Guidelines for the preformance of fusion procedures for degenerative disease of the lumbar spine

Introduction to the Lumbar Fusion Guidelines

As scientific understanding of the pathophysiology ofdegenerative disease of the lumbar spine has increased, thepossibilities for correcting the underlying problem and theresulting improvement in clinical function have expandedexponentially. Fueled by advances in material technologyand surgical technique, treatment of greater numbers ofindividuals suffering from lumbar spinal disease has prolif-erated. Using data from the National Hospital DischargeSurvey, Deyo and colleagues4 described a 200% increase inthe frequency of lumbar fusion procedures in the 1980s.Davis3 observed that the age-adjusted rate of hospitaliza-tion for lumbar surgery and lumbar fusion increased greaterthan 33% and greater than 60%, respectively, from 1979 to 1990. Lumbar fusion has been described as a treatmentof symptomatic degenerative disc disease, spinal stenosis,spondylolisthesis, and degenerative scoliosis. Lumbar fu-sion has been performed to treat acute and chronic low-back pain, radiculopathy, and spinal instability.

As practitioners have become caught up in the excite-ment of what can be accomplished, there are increasingquestions regarding what should be done and how. Thesequestions are being addressed in this current document,Guidelines for the Performance of Fusion Procedures forDegenerative Disease of the Lumbar Spine.

In January 2003, a group was formed at the request ofthe leadership of the CNS by the executive committee of

the American Association of Neurological Surgeons/CNSJoint Section on Disorders of the Spine and PeripheralNerves to perform an evidence-based review of the litera-ture on lumbar fusion procedures for degenerative diseaseof the lumbar spine and to formulate treatment recom-mendations based on this review. In March 2003, thisgroup was convened. Invitations were extended to approx-imately 12 orthopedic and neurosurgical spine surgeonsactive in the Joint Section or in the North American SpineSociety to ensure participation of nonneurosurgical spinesurgeons. The 50 recommendations that follow this intro-duction represent the product of the work of the group,with input from the Guidelines Committee of the Ameri-can Association of Neurological Surgeons/CNS and theClinical Guidelines Committee of North American SpineSociety.

The first few papers in this series deal with the meth-odology of guideline formation and the assessment of outcomes following lumbar fusion. The next series of rec-ommendations involve the diagnostic modalities helpfulfor the pre- and postoperative evaluation of patients con-sidered candidates for or treated with lumbar fusion, fol-lowed by recommendations dealing with specific patientpopulations. Finally, several surgical adjuncts, includingpedicle screws, intraoperative monitoring, and bone graftsubstitutes are discussed, and recommendations are madefor their use.

Methodology

The development of practice parameters, guidelines, orrecommendations is an onerous and time-consuming pro-cess. It consists of literature gathering (primarily through

J. Neurosurg: Spine / Volume 2 / June, 2005

J Neurosurg: Spine 2:637–638, 2005

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 1: introduction and methodology

DANIEL K. RESNICK, M.D., TANVIR F. CHOUDHRI, M.D., ANDREW T. DAILEY, M.D.,MICHAEL W. GROFF, M.D., LARRY KHOO, M.D., PAUL G. MATZ, M.D., PRAVEEN MUMMANENI, M.D., WILLIAM C. WATTERS III, M.D., JEFFREY WANG, M.D.,BEVERLY C. WALTERS, M.D., M.P.H., AND MARK N. HADLEY, M.D.

Department of Neurosurgery, University of Wisconsin, Madison, Wisconsin; Department ofNeurosurgery, Mount Sinai Medical School, New York, New York; Department of Neurosurgery,University of Washington, Seattle, Washington; Department of Neurosurgery, Indiana University,Indianapolis, Indiana; Departments of Orthopedic Surgery and Neurosurgery, University ofCalifornia at Los Angeles, California; Department of Neurosurgery, University of Alabama atBirmingham, Alabama; Department of Neurosurgery, Emory University, Atlanta, Georgia; Bone and Joint Clinic of Houston, Texas; and Department of Neurosurgery, Brown University,Providence, Rhode Island

KEY WORDS • fusion • lumbar spine • practice guidelines • treatment outcome

637

Abbreviation used in this paper: CNS = Congress of Neurologi-cal Surgeons.

SpineJune2005 5/24/05 10:42 AM Page 637

computerized literature searches), evaluation and classifi-cation of the quality of evidence provided by the literature,interpretation of this evidence to draw meaningful conclu-sions, and formulation of recommendations based on thisprocess. The process is meant to be clear, and the reader isencouraged to read the entire document as opposed to therecommendations alone.

Guideline development within the specialty of neuro-surgery has followed a rigorous process delineated early onin the advent of specialty-specific guidelines.5 Followingrecommendations proposed by other specialty societies, the process used in neurosurgical guideline developmentdivides the types of literature into classes depending on thescientific strength of the study design.6 Because the publi-cation of the ground-breaking and exemplary Guidelinesfor the Management of Severe Head Injury,1,2 an effort hasbeen made to adhere to these strict criteria for practice rec-ommendations. The definitions of classes of evidence fortherapeutic effectiveness are as follows: Class I, evidencefrom one or more well-designed, randomized controlledclinical trials, including overviews of such trials; Class II,evidence from one or more well-designed comparative cli-nical studies, such as nonrandomized cohort studies, case-control studies, and other comparable studies, includingless well-designed randomized controlled trials; and ClassIII, evidence from case series, comparative studies withhistorical controls, case reports, and expert opinion as wellas significantly flawed randomized controlled trials. Fordiagnostic tests, and clinical assessment, other study de-signs are used, and therefore the classification systems areslightly different, but still result in Classes I, II, and III evi-dence. This is reviewed in detail elsewhere.6

Class I evidence is used to support treatment recom-mendations of the strongest type, called practice standards,reflecting a high degree of clinical certainty. Class II evi-dence is used to support recommendations called guidelines,reflecting a moderate degree of clinical certainty. Othersources of information, including observational studies suchas case series and expert opinion, as well as fatally flawedrandomized controlled trials (Class III evidence), supportpractice options reflecting unclear clinical certainty.

On the surface, this appears to be a fairly straightfor-ward task, but within the process the most difficult aspectis evaluating the quality of the evidence in each type.Disappointingly, studies in which evidence should be con-sidered Class I or II because of study type have to bedowngraded to a lower class of evidence due to method-ological flaws that could cause false conclusions to bedrawn from the evidence. This is discussed extensivelywithin each topic, and all cited evidence is listed in outlineform in the evidentiary tables, so as to ensure transparen-cy of the development process.

The group culled through literally thousands of refer-ences to identify the most scientifically robust citationsavailable concerning each individual topic. Not every ref-erence identified is cited. In general, if high-quality (ClassI or II) medical evidence was available on a particulartopic, poorer-quality evidence was only briefly summa-rized and rarely included in the evidentiary tables. If nohigh-quality evidence existed, or if there was significantdisagreement between similarly classified evidence sourc-es, then the Class III and supporting medical evidencewere discussed in greater detail. If multiple reports were

available that provided similar information, a few werechosen as illustrative examples.

A consistent finding during the exploration of many ofthese topics was that many investigators reported studies inwhich the designs were unsophisticated. The use of invalidoutcome measures, the lack of an appropriate power analy-sis, and the failure to identify distinct patient populations forstudy inhibited our ability to draw meaningful conclusionsfrom many reports. Specific examples are provided in thetext of each topic. Suggestions for future research are madeat the conclusion of each paper. We, as spine surgeons, mustimprove the quality of our research practices to provide con-vincing evidence that the therapies we strongly believe inare safe, effective, and make economic sense.

During the development of these guidelines, the authorsoften found that their preconceived ideas regarding theproper treatment of patients with chronic low-back painwere founded on poor-quality or controversial medical evi-dence. Some recommendations have resulted in changes inthe authors’ practice patterns after every effort was made toclassify the evidence and to interpret the results of the vari-ous studies in a scientifically rigorous fashion. Many rec-ommendations are made at the lowest level, meaning thatdefinitive evidence is lacking to support the recommenda-tion but that evidence exists at some level. Some readerswill undoubtedly disagree with one or more of our recom-mendations or with the level of a given recommendation.The justification for all of the recommendations is includedin the scientific foundation portion and the summary sectionof each guideline. If the job has been done correctly, the rea-soning behind the recommendation should be clear.

It is our hope, as well as that of the participating orga-nizations, that these guidelines will help to elucidate thecurrent knowledge on the topic of lumbar fusion and willstimulate the development of more rigorous scientific evi-dence justifying or refining—or, if appropriate, eliminat-ing—aspects of this form of treatment.

References

1. Bullock R, Chesnut RM, Clifton G, et al: Guidelines for themanagement of severe head injury. Brain Trauma Foundation.J Neurotrauma 13:639–734, 1996

2. Bullock R, Chesnut, RM, Clifton G, et al: Guidelines for themanagement of severe traumatic brain injury. J Neurotrauma17:451–627, 2000

3. Davis H: Increasing rates of cervical and lumbar spine surgeryin the United States, 1979–1990. Spine 19:1117–1124, 1994

4. Deyo RA, Cherkin D, Conrad D, et al: Cost, controversy, crisis:low back pain and the health of the public. Annu Rev PublicHealth 12:141–156, 1991

5. Rosenberg J, Greenberg MK: Practice parameters: strategies forsurvival into the nineties. Neurology 42:1110–1115, 1992

6. Walters BC: Clinical practice parameter development, in BeanJR (ed): Neurosurgery in Transition. Baltimore: Williams &Wilkins, 1998, pp 99–111

Manuscript received December 7, 2004.Accepted in final form February 18, 2005.Address reprint requests to: Daniel K. Resnick, M.D., Depart-

ment of Neurological Surgery, University of Wisconsin MedicalSchool, K4/834 Clinical Science Center, 600 Highland Avenue,Madison, Wisconsin 53792. email: [email protected].

D. K. Resnick, et al.

638 J. Neurosurg: Spine / Volume 2 / June, 2005


Recommendations

Standards. It is recommended that functional outcomebe measured in patients treated for low-back pain due todegenerative disease of the lumbar spine by using reliable,valid, and responsive scales. Examples of these scales inthe low-back pain population include the following: TheSpinal Stenosis Survey of Stucki, Waddell–Main Ques-tionnaire, RMDQ, DPQ, QPDS, SIP, Million Scale, LBPRScale, ODI, the Short Form–12, the JOA system, theCBSQ, and the North American Spine Society LumbarSpine Outcome Assessment Instrument.

Guidelines. There is insufficient evidence to recom-mend a guideline for assessment of functional outcomefollowing fusion for lumbar degenerative disease.

Options. Patient satisfaction scales are recommendedfor use as outcome measures in retrospective case series,where better alternatives are not available. Patient satis-faction scales are not reliable for the assessment of out-come following intervention for low-back pain.

Rationale

Lumbar spinal fusion is an increasingly common pro-cedure performed as an adjunct in the surgical manage-ment of patients with degenerative lumbar disease andinstability. As the frequency and complexity of lumbarfusion surgery increases, there is a tendency for costs andcomplication rates to increase as well.20 With fewer hospi-tal resources available, the ability to assess objectively thefunctional outcome following lumbar fusion and to corre-late patient outcome with the economic consequences oftreatment is important.

Various assessment tools are available for measuringfunctional outcomes in patients who have undergone lum-bar fusion. These outcomes may vary widely in the samepopulation depending on whether subjective or objectivemeasures have been used.17 Examples of objective outcomemeasures include physiological, anatomical, economic,health-related QOL, and mortality measurements.10 Objec-tive outcome measures may be classified into functionalquestionnaires, global ratings (satisfaction), economic fac-tors (employment, disability, and cost), and physical factors(activities).21 The purpose of this review was to identifyvalid, reliable, and responsive measures of functional out-comes after lumbar fusion for degenerative disease.

Search Criteria

A computerized search of the National Library of Me-dicine database of the literature published between 1966



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 2: assessment of functional outcome



KEY WORDS • fusion • lumbar spine • practice guidelines • treatment outcome

639

Abbreviations used in this paper: CBSQ = Curtain BackScreening Questionnaire; DPQ = Dallas Pain Questionnaire; DRI =Disability Rating Index; FSQ = Functional Status Questionnaire;JOA = Japanese Orthopaedic Association; LBPR = Low Back PainRating; ODI = Oswestry Disability Index; QOL = quality of life;QPDS = Quebec Pain Disability Scale; RMDQ = Roland–MorrisDisability Questionnaire; SF-36 = Short-Form–36; SIP = SicknessImpact Profile; VAS = visual analog scale.


and 2003 was performed. A search using the subject head-ing “lumbar fusion” yielded 3708 citations. The followingsubject headings were combined: “lumbar fusion and out-comes.” Approximately 204 citations were acquired. Onlycitations in English were selected. A search of this set ofpublications with the key words “functional outcome” and“satisfaction” resulted in 107 matches. Alternative search-es included each disability index by name. Titles and ab-stracts of the articles were reviewed and clinical seriesdealing with adult patients treated with lumbar fusion fordegenerative lumbar disease were selected for detailed an-alysis. Additional references were culled from the ref-erence lists of remaining articles. Among the articles reviewed, 30 studies were included that dealt with lum-bar fusion, functional outcomes, and satisfaction surveys.Nineteen of these articles were studies in which the au-thors examined the reliability of functional outcome mea-sures. In another seven articles investigators examined theutility of these functional outcome measures in the settingof lumbar fusion. Two articles were overviews on func-tional outcome and lumbar degenerative disease. All pa-pers providing Class I medical evidence are summarizedin the evidentiary table (Table 1).

Scientific Foundation

Assessment of Functional Outcome

To assess outcome following treatment properly, afunctional instrument must fulfill three criteria.11,21 First, itmust be reliable.10,11 Repetition of the functional assess-ment should be consistent within (internal reliability) andbetween (external reliability) observers. If a functional in-strument contains multiple domains, each should correlatewith the final outcome (internal consistency). Second, afunctional instrument must be valid.21 It should measure theproperty intended. For example, an instrument assessingdysfunction due to leg pain would be expected to correlatewith a reduction in the ability to walk a given distance.Finally, the instrument should be responsive.21 The instru-ment should be able to detect differences in severity amongpopulations. If an instrument measures low-back pain andthis pain improves with physical therapy, the instrumentshould reflect that improvement quantitatively. When eval-uating the utility of a functional tool, the initial assessmentshould emphasize reliability. If a functional instrumentdoes not produce reliable results, its validity and respon-siveness are irrelevant.

In terms of grading the quality of outcomes instruments,k and a values are used. The k value refers to the degreeof correlation of interrater observations (reliability). Inpatient-based assessments, it indicates consistency in res-ponse at a given time point. The a value, often calculatedusing the Cronbach a test, reflects the degree to whicheach domain of a multidomain outcome measure corre-lates with the final result.7 For example, an assessmenttool for pain may contain physical, psychological, and so-cial domains. Each domain score should correlate with thefinal score. For a study to provide Class I medical evi-dence regarding functional outcomes, the outcomes toolused must have a k value greater than 0.8. Class II med-ical evidence requires an outcomes tool to have a k greaterthan 0.6. Any outcome scale with a k value less than 0.6

is considered to provide Class III medical evidence for theassessment of outcomes following an intervention.18

Roland and Morris30,31 followed 230 patients of whom193 were studied up to 4 weeks after their initial presenta-tion. Functional disability was assessed using a 24-itemdisability questionnaire (the RMDQ) with statements de-rived from the SIP and relating to the lower back. Re-liability was ascertained in 20 patients with an externalreliability greater than 0.91. Internal consistency appearedto be greater than 0.8. Validity was confirmed after com-parisons to a six-point pain rating scale and physical signsascertained by an examining physician.31 In this group,60% of patients appeared to improve over the 4-weekperiod, whereas 20% worsened. Absence from work ap-peared to correlate less well with disability, as only 8% of the employed were unable to work.30 Using the ODI,Fairbank and colleagues12 followed 25 patients with acutelow-back pain in whom a reasonable prognosis was ex-pected. The questionnaire has 10 categories with six gra-dations each, for a total score of 50. It was completed atweekly intervals over a period of 3 weeks. Reliability (k . 0.95) was confirmed in 22 patients who repeated thequestionnaire over 2 days. Validity was demonstrated aspatients improved over 3 weeks. Paired t-tests revealed asignificant improvement in ODI scores during this timeperiod (p , 0.005).

Leclaire and colleagues24 observed patients who pre-sented with acute low-back pain alone (100 cases) or ac-companied by radiculopathy (100 cases). The cohort wasfollowed using the RMDQ and ODI questionnaires. In theradiculopathy group, ODI and RMDQ scores were signif-icantly more severe (higher) than in the low-back pain–alone group (p , 0.0001). The two scales had a moderatecorrelation to each other in each subgroup (r = 0.72 [ra-diculopathy]; r = 0.66 [lumbago]; p , 0.0001). In a cohortof patients with low-back pain, the JOA score was used asa psychometric measure. External reliability was strong (k . 0.90) when 15 patients reassessed their status withno change in their symptomatology. Interobserver externalreliability among physicians was also sound (k . 0.90) in30 patients reassessed using the JOA. Validity was estab-lished by a strong correlation to the RMDQ, ODI, and theSF-36.15 In several different groups with lumbar degener-ative disease, the North American Spine Society LumbarSpine Outcome Assessment tool was used to assesspatients who had undergone conservative or decompres-sive therapy.8 In this study, 136 of 206 questionnaires weresuccessfully completed. External reliability was assessedin 64 patients. Both internal and external reliability wasstrong (k . 0.90). The test was determined to be a validmeasure compared with existing instruments.

The SIP is a traditional general functional outcomemeasure, with 136 items in 12 categories, that has beenevaluated in the general populace for a variety of condi-tions. It has been applied to patients with low-back painand degenerative lumbar disease. Bergner, et al.,1 exam-ined the use of this general health instrument in 1108 pa-tients with multiple medical problems including rheuma-toid arthritis and hip osteoarthritis.1 Simultaneous withthis questionnaire were a clinician’s assessment of physi-cal function and patients’ self-assessment of the severityof sickness and dysfunction. In this setting, the test–retest(external) reliability of SIP was greater than 0.90, and its





Functional Outcome

641

TAB

LE

1E

vide

ntia

ry t

able

sum

mar

izin

g pu

blis

hed

stud

ies

invo

lvin

g C

lass

I m

edic

al d

ata*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nR

esul

tsC

oncl

usio

ns

Fair

bank

, et a

l., 1

980

I25

pat

ient

s w

/ acu

te L

BP

& r

easo

nabl

e pr

ogno

sis

wer

eTe

st–r

etes

t rel

iabi

lity

was

k.

0.95

(p

,0.

001)

in

The

OD

I is

a r

elia

ble

& v

alid

mea

sure

in d

etec

ting

stud

ied

at w

kly

inte

rval

s fo

r 3

wks

w/ a

fun

ctio

nal

22 p

atie

nts.

Ove

r th

e 3-

wk

inte

rval

, sig

nifi

cant

im-

chan

ges

in th

e L

BP

& it

s fu

nctio

nal s

ever

ity.

disa

bilit

y su

rvey

. The

OD

I ha

s 10

cat

egor

ies

each

pr

ovem

ent w

as n

oted

clin

ical

ly &

was

det

ecte

d w

/ 6 r

espo

nses

gra

ded

0–5.

Ato

tal o

f 50

poi

nts

are

usin

g th

e O

DI.

Apa

ired

t-te

st r

evea

led

a si

gnif

ican

t po

ssib

le.

impr

ovem

ent o

n th

e O

DI

over

3 w

ks (

p ,

0.05

).B

ergn

er, e

t al.,

198

1I

1108

pat

ient

s in

a g

ener

al p

opul

ace

w/ m

ultip

le p

rob-

Ext

erna

l rel

iabi

lity

w/in

& b

twn

obse

rver

s w

as

SIP

mea

sure

s in

depe

nden

t fun

ctio

n, p

hysi

cal w

elln

ess,

le

ms

incl

udin

g R

A&

hip

ost

eoar

thri

tis. P

atie

nts

k.

0.90

. Int

erna

l con

sist

ency

was

a.

0.90

. &

psy

chos

ocia

l wel

lnes

s. I

t is

relia

ble

& v

alid

. Rea

-w

ere

eval

uate

d us

ing

the

SIP.

Ass

essm

ent w

as d

one

Self

-ass

essm

ent o

f si

ckne

ss &

dys

func

tion

had

sona

ble

mea

sure

s to

use

for

out

com

e ar

e SI

P&

sel

f-by

a c

linic

ian

for

phys

ical

mea

sure

s. S

elf-

asse

ss-

a re

liabi

lity

of k

.0.

60. T

he S

IPap

pear

ed to

cor

-as

sess

men

t of

sick

ness

& d

ysfu

nctio

n.m

ent w

as c

ompl

eted

for

sev

erity

of

sick

ness

&

rela

te w

/ the

sel

f-as

sess

men

t of

sick

ness

& d

ys-

dysf

unct

ion.

func

tion

(cor

rela

tion

.0.

50).

Mill

ion,

et a

l., 1

982

I19

pat

ient

s w

/ chr

onic

LB

P. T

heir

fun

ctio

nal d

isab

ility

Ext

erna

l rel

iabi

lity

was

str

ong

btw

n &

w/in

obs

erve

rs

The

Mill

ion

Scal

e is

a r

elia

ble

indi

cato

r of

the

seve

rity

w

as s

tudi

ed u

sing

the

Mill

ion

Scal

e w

hich

was

a

k.

0.90

. As

a va

lidity

mea

sure

, the

Mill

ion

of lu

mba

go &

is r

espo

nsiv

e in

the

earl

y ph

ase

of

VA

S ex

amin

ing

15 s

ubje

ctiv

e va

riab

les

refl

ectin

g Sc

ale

appe

ared

to r

efle

ct c

hang

es in

phy

sica

l mea

-tr

eatm

ent.

Its

resp

onsi

vene

ss a

ppea

rs b

ette

r th

an th

at

the

seve

rity

of

lum

bago

. Aso

ft c

orse

t w/ &

w/o

su

rem

ents

. At 4

& 8

wks

aft

er r

igid

bra

cing

, pa-

of o

bjec

tive

mea

sure

men

ts in

clud

ing

lum

bar

mot

ion

supp

ort w

as u

sed

to te

st th

e re

spon

sive

ness

of

the

tient

s im

prov

ed c

linic

ally

, & th

is r

espo

nsiv

enes

s &

str

aigh

t leg

rai

sing

.M

illio

n Sc

ale.

was

det

ecte

d by

the

Mill

ion

Scal

e (p

,0.

05 a

t 4

wks

& p

,0.

01 a

t 8 w

ks).

Rol

and

& M

orri

s, 1

983

I23

0 pa

tient

s w

/ acu

te lu

mba

go; 1

93 w

ere

stud

ied

at 0

,E

xter

nal r

elia

bilit

y w

as k

.0.

90 &

inte

rnal

con

sist

-T

he R

MD

Q is

rel

iabl

e fo

r as

sess

men

t of

acut

e L

BP.

1, &

4 w

ks a

fter

the

epis

ode.

Tes

t–te

st r

elia

bilit

y en

cy a

.0.

80. C

onst

ruct

val

idity

dem

onst

rate

d w

as d

one

on 2

0/23

0 pa

tient

s. T

he c

onst

ruct

val

idity

that

the

Rol

and–

Mor

ris

ques

tionn

aire

was

abl

e to

w

as q

ualit

ativ

ely

asse

ssed

by

com

pari

ng th

is f

unc-

dete

ct q

ualit

ativ

ely

patie

nts

w/ p

oore

r ou

tcom

es

tiona

l que

stio

nnai

re to

the

pain

rat

ing

scal

e.fr

om a

cute

lum

bago

; how

ever

, no

spec

ific

ana

ly-

sis

was

don

e.R

olan

d &

Mor

ris,

198

3I

230

patie

nts

w/ a

cute

lum

bago

who

wer

e st

udie

d at

0,

. 6

0% o

f pa

tient

s ha

d im

prov

emen

t ove

r th

e 4-

wk

No

spec

ific

sta

tistic

s te

sted

the

corr

elat

ion

in th

is s

tudy

. 1,

& 4

wks

. The

dis

abili

ty q

uest

ionn

aire

was

ad-

peri

od, w

here

as 2

0% h

ad a

n in

crea

se in

dis

abili

ty.

The

RM

DQ

is r

elia

ble

but t

his

man

uscr

ipt d

id n

ot

min

iste

red

& c

ompl

eted

at a

ll tim

e in

terv

als

in 1

93

The

se c

hang

es a

ppea

red

to b

e re

flec

ted

in th

e di

s-as

sess

its

resp

onsi

vene

ss to

a s

tand

ard

mea

sure

in s

ta-

patie

nts.

Cor

rela

tion

was

qua

litat

ivel

y do

ne w

/ ab

ility

que

stio

nnai

re. A

bsen

ce f

rom

wor

k ap

pear

-tis

tical

fas

hion

.ba

ck-t

o-w

ork

stat

us.

ed to

cor

rela

te le

ss w

ell a

s on

ly 8

% o

f em

ploy

ed

wer

e un

able

to w

ork

4 w

ks a

fter

acu

te lu

mba

go.

Wad

dell

& M

ain,

198

4I

160

patie

nts

w/ 1

2 w

ks o

f lu

mba

go (

chro

nic)

w/ s

e-D

isab

ility

as

dete

rmin

ed b

y fu

nctio

nal o

utco

me

on

Wad

dell

Scal

e de

scri

bes

func

tiona

l dis

abili

ty w

/ chr

on-

veri

ty s

tudi

ed b

y a

9-ca

tego

ry d

isab

ility

inde

x &

qu

estio

nnai

re h

ad a

rel

iabi

lity

. 0

.80

& c

orre

late

dic

LB

P. A

ll 9

scal

es c

orre

late

w/ f

inal

sco

re (

cont

ent

phys

ical

cha

ract

eris

tics.

Rel

iabi

lity

dete

rmin

ed u

s-w

/ the

OD

I (r

= 0

.70)

. For

phy

sica

l cha

ract

eris

tics

valid

ity)

& th

e sc

ale

is r

elia

ble.

It a

lso

has

cons

truc

t in

g a

subg

roup

of

30 p

atie

nts.

(lum

bar

flex

ion,

str

aigh

t leg

rai

sing

, roo

t com

pres

-va

lidity

as

it co

rrel

ates

w/ O

DI.

sion

sig

ns)

relia

bilit

y w

as .

0.9

0.D

eyo,

198

6I

136

patie

nts

who

wer

e ex

amin

ed in

a c

linic

for

a c

hief

Rel

iabi

lity

for

both

sca

les

was

k.

0.80

in p

atie

nts

The

SIP

& th

e m

odif

ied

RM

DQ

(sh

orte

r) a

re r

elia

ble

com

plai

nt o

f lu

mba

go. E

valu

atio

n w

as d

one

usin

g (1

0) w

ho h

ad n

o ch

ange

in p

ain.

For

pat

ient

s w

ho

scal

es f

or th

e as

sess

men

t of

lum

bago

, whi

ch s

eem

to

SIP

& th

e m

odif

ied

RM

DQ

Sca

le (

shor

tene

d ve

r-di

d no

t res

ume

full

activ

ity (

47),

the

relia

bilit

y fo

llow

the

phys

ical

dim

ensi

on o

f fu

nctio

nal d

isab

ility

. si

on o

f SI

P) in

itial

ly &

3 w

ks la

ter.

was

a.

0.60

. Ast

rong

cor

rela

tion

exis

ted

btw

n T

he m

odif

ied

RM

DQ

is le

ss w

ell s

uite

d to

fol

low

the

the

scal

es (

r =

0.8

5) &

bet

wee

n th

e ph

ysic

al d

i-ps

ycho

soci

al d

imen

sion

of

func

tiona

l dis

abili

ty.

men

sion

of

the

SIP

& th

e m

odif

ied

RM

DQ

(r

=

0.89

). T

he m

odif

ied

RM

DQ

cor

rela

ted

less

wel

l w

/ the

psy

chos

ocia

l dim

ensi

on o

f th

e SI

P(r

= 0

.56)

.L

awlis

, et a

l., 1

989

I14

3 pa

tient

s ov

eral

l (24

nor

mal

, 15

chro

nic

lum

bago

Ext

erna

l rel

iabi

lity

was

k.

0.90

. Con

stru

ct v

alid

ity

The

DPQ

is a

rel

iabl

e te

st in

ass

essi

ng c

hron

ic L

BP

&

but w

orki

ng, 1

04 c

hron

ic lu

mba

go u

nder

goin

g in

- w

as s

how

n by

cor

rela

tion

of th

e 1s

t 2 c

ateg

orie

s ap

pear

s re

spon

sive

in d

efin

ing

diff

eren

ces

btw

n pa

tient

ther

apy)

. Fun

ctio

nal a

sses

smen

t per

form

ed

of D

PQ w

/ fun

ctio

nal c

apac

ity s

core

s re

latin

g to

pa

tient

s w

/ chr

onic

lum

bago

& th

ose

w/o

.us

ing

the

DPQ

whi

ch a

sses

ses

daily

act

iviti

es, w

ork

the

phys

ical

dem

ands

of

wor

k. R

espo

nsiv

enes

s w

as&

leis

ure

activ

ities

, anx

iety

/dep

ress

ion,

& s

ocia

l as

sess

ed b

y co

mpa

ring

DPQ

sco

res

in th

e 10

4 in

tere

st. R

elia

bilit

y te

sted

on

15 c

hron

ic p

ain

patie

nts

chro

nic

lum

bago

pat

ient

s to

the

24 n

orm

al p

atie

nts.

& 1

3 no

rmal

pat

ient

s.D

PQ s

core

s w

ere

sign

ific

antly

hig

her

in th

e fo

rmer

.

cont

inue

d




TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nR

esul

tsC

oncl

usio

ns

Man

nich

e, e

t al.,

199

4I

58 p

atie

nts

who

und

erw

ent l

umba

r di

sc o

p w

ere

sur-

The

LB

PR s

cale

com

pris

ed 6

0 po

ints

for

pai

n, 3

0 fo

rT

he L

BPR

Sca

le c

ombi

nes

elem

ents

of

phys

ical

fun

c-ve

yed

14–6

0 m

os p

osto

p. T

he a

sses

smen

t was

an

leve

l of

func

tion,

& 4

0 fo

r ph

ysic

al im

pair

men

t. tio

n, p

ain

inte

nsity

, & o

vera

ll di

sabi

lity.

It i

s a

relia

ble

LB

PR s

cale

that

exa

min

ed p

hysi

cal i

mpa

irm

ent,

dis-

Inte

rrat

er r

elia

bilit

y w

as k

.0.

95. U

sing

con

tin-

indi

cato

r of

dys

func

tion

& a

ppea

rs v

alid

com

pare

d w

/ ab

ility

, & p

ain

inte

nsity

. Com

pari

son

was

don

e ge

ncy

tabl

es, t

he s

cale

cor

rela

ted

with

the

doct

or’s

obje

ctiv

e m

easu

res

(doc

tor’

s as

sess

men

t) &

sub

jec-

agai

nst a

doc

tor’

s gl

obal

ass

essm

ent &

a p

atie

nt’s

as

sess

men

t and

pat

ient

’s a

sses

smen

t (p

,0.

0000

5).

tive/

satis

fact

ion

mea

sure

s (p

atie

nt’s

ass

essm

ent)

.gl

obal

ass

essm

ent.

Rut

a, e

t al.,

199

4I

354

patie

nts

w/ l

umba

go in

itial

ly e

xam

ined

in c

linic

&18

3 pa

tient

s ha

d no

clin

ical

cha

nges

& u

nder

wen

tT

his

LB

Psc

ale

is a

rel

iabl

e &

val

id in

dica

tor

of th

e su

rvey

ed s

hort

ly th

erea

fter

to a

sses

s fu

nctio

nal d

is-

exte

rnal

rel

iabi

lity

test

ing

(k.

0.90

). T

he q

uest

ion-

func

tiona

l dis

abili

ty r

elat

ing

to lu

mba

go. N

o us

age

abili

ty. 2

73 p

atie

nts

wer

e re

test

ed f

or r

elia

bilit

y of

na

ire

corr

elat

ed w

ell w

/ all

8 do

mai

ns o

f th

e SF

-36

desc

ribe

d in

the

setti

ng o

f lu

mba

r fu

sion

. No

acui

ty

who

m 1

83 r

epor

ted

no c

hang

e in

clin

ical

sev

erity

. us

ing

linea

r re

gres

sion

(p

,0.

001)

& w

/ per

cep-

give

n fo

r th

e lu

mba

go.

Cor

rela

tion

to th

e SF

-36

gene

ral h

ealth

pro

file

was

tio

ns o

f di

seas

e se

veri

ty.

done

for

con

stru

ct v

alid

ity.

Sale

n, e

t al.,

199

4I

1445

pat

ient

s w

ere

divi

ded

into

3 g

roup

s: 1

092

vol-

Ext

erna

l rel

iabi

lity

for

the

DR

I w

as k

.0.

80. T

here

T

he D

RI

is a

rel

iabl

e, v

alid

, & r

espo

nsiv

e m

easu

re in

un

teer

con

trol

s, 3

06 w

/ axi

al s

kele

tal p

ain,

& 4

7 w

/w

as a

cor

rela

tion

to th

e FS

Q. T

he D

RI

was

res

- pa

tient

s w

/ axi

al s

kele

tal p

ain.

join

t pai

n. P

atie

nts

wer

e ev

alua

ted

usin

g th

e D

RI

pons

ive

in d

etec

ting

impr

ovem

ent a

fter

join

t re-

& a

n FS

Q.

plac

emen

t.H

arpe

r, et

al.,

199

5I

150

patie

nts

wer

e di

vide

d in

to 3

gro

ups

(Gro

up I

: E

xter

nal r

elia

bilit

y in

all

3 gr

oups

was

k.

0.90

. In-

The

CB

SQ is

a r

elia

ble

& v

alid

mea

sure

for

det

erm

inin

g ch

roni

c lu

mba

go .

4 w

ks/d

isab

led;

Gro

up I

I: a

cute

te

rnal

rel

iabi

lity

was

a.

0.80

. The

re w

as a

str

ong

the

func

tiona

l dis

abili

ty a

ssoc

iate

d w

ith L

BP.

No

lum

bago

/wor

king

; Gro

up I

II: n

orm

al).

Eva

luat

ion

ofco

rrel

atio

n bt

wn

each

cat

egor

y in

CB

SQ &

its

sim

-te

stin

g of

res

pons

iven

ess

was

und

erta

ken.

func

tiona

l dis

abili

ty w

as d

one

usin

g th

e C

BSQ

&

ilar

cate

gory

in th

e SI

P(r

= 0

.56–

0.72

). F

inal

ly,

the

SIP.

The

CB

SQ te

sts

11 c

ateg

orie

s of

fun

ctio

nal

CB

SQ s

core

s ap

pear

ed r

espo

nsiv

e w

/ hig

her

scor

esdi

sabi

lity.

Test

–ret

est c

orre

latio

n &

cor

rela

tion

btw

nin

the

mor

e se

vere

ly a

ffec

ted

grou

ps.

tCB

SQ a

nd S

IPw

as d

one

usin

g th

e Pe

arso

n co

r-re

latio

n te

st.

Kop

ec, e

t al.,

199

5I

242

patie

nts

with

a h

isto

ry o

f lu

mba

go in

Que

bec.

80%

Ext

erna

l rel

iabi

lity

was

k.

0.90

w/ i

nter

nal c

onsi

s-T

he Q

PDS

is s

uita

ble

for

the

relia

ble

func

tiona

l mea

-ha

d pr

ior

lum

bago

w/ 2

9% r

ecei

ving

com

pens

atio

n.

tenc

y of

a.

0.90

. Con

stru

ct v

alid

ity w

as s

how

n su

rem

ent o

f L

BP.

Patie

nts

wer

e as

sess

ed f

or f

unct

iona

l dis

abili

ty u

s-by

a s

tron

g co

rrel

atio

n in

this

fun

ctio

nal i

ndex

w/

ing

the

QPD

S. R

elia

bilit

y w

as e

xam

ined

in a

98-

the

OD

I (r

= 0

.80)

, RM

DQ

(r

= 0

.77)

, & S

F-36

pa

tient

sam

ple

w/in

1–1

4 da

ys a

fter

initi

al s

urve

y.

(r =

0.7

2).

Con

stru

ct v

alid

ity w

as d

one

by c

ompa

ring

res

ults

to

fun

ctio

nal s

cale

s of

OD

I, R

MD

Q, &

SF-

36.

Dal

troy

, et a

l., 1

996

I20

6 pa

tient

s in

6 o

rtho

pedi

c pr

actic

es w

ere

eval

uate

d.

Ext

erna

l & in

tern

al r

elia

bilit

y w

ere

stro

ng (

k.

0.90

)T

he N

ASS

LSO

Ais

a v

alid

& r

elia

ble

outc

ome

mea

sure

Pa

tient

s w

ere

in s

ever

al c

ateg

orie

s in

clud

ing

thos

e w

hen

asse

ssed

in 6

4 pa

tient

s. T

he m

easu

re w

as

for

func

tiona

l eva

luat

ion

of th

e lu

mba

r sp

ine.

w/ L

BP

& s

ciat

ica.

Als

o in

clud

ed w

ere

patie

nts

who

valid

com

pare

d w

/ kno

wn

inst

rum

ents

.un

derw

ent l

umba

r de

com

pres

sion

but

not

fus

ion.

Stuc

ki, e

t al.,

199

6I

193

patie

nts

w/ l

umba

r de

gene

rativ

e st

enos

is u

nder

go-

23/1

93 s

tudi

ed f

or r

elia

bilit

y w

/ k.

0.80

. Int

erna

l T

his

outc

ome

ques

tionn

aire

was

rel

iabl

e in

lum

bar

sten

-in

g de

com

pres

sion

. Pro

spec

tive

mul

ticen

ter

stud

y of

cons

iste

ncy

a.

0.80

; 130

/193

stu

died

for

res

pon-

osis

pat

ient

s w

ho u

nder

wen

t op

& h

ad c

onst

ruct

val

id-

self

-adm

inis

tere

d ou

tcom

e m

easu

re a

sses

sed

w/in

si

vene

ss. R

espo

nsiv

e &

val

id o

ver

6 m

os to

det

ect

ity c

ompa

red

w/ e

stab

lishe

d sc

ale

& w

as r

espo

nsiv

e 6

mos

. L

iker

t res

pons

e sc

ales

use

d in

dom

ains

of

impr

ovem

ent p

osto

p.in

det

ectio

n of

dif

fere

nces

w/in

6 m

os f

or f

unct

iona

l ph

ysic

al d

ysfu

nctio

n, s

ympt

om s

ever

ity, &

sat

isfa

c-im

prov

emen

t.tio

n. R

esul

ts c

ompa

red

w/ S

IP&

VA

S.Fu

jiwar

a, e

t al.,

200

3I

97 p

atie

nts

obse

rved

clin

ical

ly w

/ LB

P&

fol

low

ed

Test

–ret

est r

elia

bilit

y w

as k

.0.

90 w

hen

patie

nts

The

JO

Ais

a r

elia

ble

& v

alid

indi

cato

r of

LB

P.us

ing

JOA

, OD

I, a

nd R

MD

Q. C

orre

latio

n w

as c

al-

(15)

or

phys

icia

ns (

30)

did

repe

at m

easu

rem

ents

. cu

late

d bt

wn

thes

e m

easu

res

& e

xter

nal r

elia

bilit

ySt

rong

cor

rela

tion

was

obs

erve

d bt

wn

JO

A&

w

as a

sses

sed

by r

epea

ted

phys

icia

n &

pat

ient

ob-

OD

I &

RM

DQ

.se

rvat

ion.

Luo

, et a

l., 2

003

I25

20 p

atie

nts

w/ L

BP;

506

pat

ient

s as

sess

ed o

ver

3–6

Ext

erna

l rel

iabi

lity

of th

e SF

-12

was

per

form

ed b

y T

he S

F-12

is c

apab

le o

f as

sess

ing

& f

ollo

win

g L

BP

mos

. SF-

12 s

urve

y w

as u

sed

& c

ompa

red

w/ s

ubje

c-W

are,

et a

l., in

a d

iffe

rent

pat

ient

gro

up; h

owev

er,

relia

bly.

tive

quan

tific

atio

n of

LB

Pin

tens

ity.

inte

rnal

rel

iabi

lity

& r

espo

nsiv

enes

s w

as f

ound

in

this

stu

dy.

*L

BP

= lo

w-b

ack

pain

; NA

SS L

SOA

= N

orth

Am

eric

an S

pine

Soc

iety

Lum

bar

Spin

e O

utco

me

Ass

essm

ent;

RA

= r

heum

atoi

d ar

thri

tis.



Functional Outcome

643

internal consistency was greater than 0.90. Self-assess-ment of sickness and dysfunction had a reliability greaterthan 0.60. The SIP appeared to correlate (. 0.50) with theself-assessment of sickness and dysfunction. Deyo9 usedthe SIP and a modified RMDQ when evaluating 136 pa-tients with a chief complaint of low-back pain at an initialindex visit and 3 weeks later. Reliability was examined in10 patients who claimed no interval improvement in painand in 47 patients who did not resume full activity. Forpatients with no change in pain, the correlation was great-er than 0.80. In those patients who may have improved butdid not resume normal activity, reliability was greater than0.60. A strong correlation was observed between the SIPand the modified RMDQ (r = 0.85). The physical dimen-sion of the SIP (r = 0.89) correlated more strongly with theRMDQ than the psychosocial dimension (r = 0.56). TheSIP appears to be a reliable and valid measure of the se-verity of low-back pain in the acute phase.

Million and colleagues27 assessed 19 patients with chron-ic low-back pain by using a VAS examining 15 subjectivevariables reflecting its severity. External reliability amongand within observers was greater than 0.90. To determinevalidity, they compared their results with physical mea-surements of spinal movements and straight leg raising.These objective assessments had a reliability greater than0.90 and correlated with the Million Scale. After bracingwith a rigid support, low-back pain improved clinically and this responsiveness was detected by the Million Scale.The Waddell–Main Disability Index was used to evaluatechronic low-back pain (duration . 12 weeks) in a 160-pa-tient cohort.37 Reliability in this study was evaluated in arandom subgroup of 30 patients. Measures were also ob-tained of objective physical characteristics including lum-bar flexion, straight leg raising, and root compression signs.The external reliability on the Waddell–Main Disability In-dex was greater than 0.80, and its validity was establishedby a strong correlation with the ODI (r = 0.70). The physi-cal characteristics, when evaluated for objective reliability,had a correlation greater than 0.80.

Using the DPQ, Lawlis and colleagues23 studied 143 pa-tients of whom 119 had chronic low-back pain. Fifteenpatients in this group were working, whereas the remain-ing 104 were undergoing inpatient therapy. Twenty-fourhealthy volunteers served as controls. The DPQ was usedto assess daily activities, work/leisure activities, anxiety/depression, and social interest. Reliability was tested in 15patients with chronic back pain and 13 controls. Externalreliability was greater than 0.90. Construct validity wasshown through a positive correlation to other assessmentsof functional capacity relating to the physical demands ofwork. The DPQ was responsive to differences betweenpatients with chronic low-back pain and controls.

Ruta, et al.,32 devised an outcome measure based onquestions commonly used in the clinical assessment ofpatients with low-back pain. A total of 354 patients withlow-back pain seen by primary and specialty practitionerswere studied. Within this group, 273 patients were testedfor reliability. One hundred eighty-three reported no clin-ical changes over a 2-week interval. External reliabilitywas tested in these 183 patients with correlations greaterthan 0.90. Validity was demonstrated by a strong correla-tion (p , 0.001 on regression) with the SF-36 generalhealth assessment. Harper and colleagues19 examined 150

patients in three subgroups (chronic low-back pain [50cases], acute lumbago [49 cases], and control [51 cases]).They employed the CBSQ, which evaluated 11 categoriesof functional disability and compared results with those ofthe SIP. External reliability for the CBSQ was greater than0.90, with internal reliability greater than 0.80. A strongcorrelation was observed between each category in theCBSQ and its similar category in the SIP (r = 0.56–72), andthe CBSQ appeared responsive in distinguishing the sever-ity of dysfunction among the three groups of patients.

Several other groups undertook studies on the functionalassessment of chronic low-back pain. Using the QPDS,Kopec, et al.,22 analyzed 242 patients with a history ofchronic low-back pain. Twenty-nine percent of this groupwere disabled and receiving compensation. This scale con-tains 48 items assessing the difficulty in simple daily activ-ities pertaining to domains relevant to low-back pain. Re-liability was gauged using a random sample (98 cases) whowere retested after 14 days. External reliability was greaterthan 0.90, with an internal consistency coefficient greaterthan 0.90. Construct validity was determined by a strongcorrelation with the ODI (r = 0.80), RMDQ (r = 0.77), andSF-36 (r = 0.72) Scales. Using the LBPR scale, Mannicheand colleagues26 surveyed 58 patients 14 to 60 monthsafter they underwent lumbar disc surgery. This scale com-prises 60 points for back and leg pain, 30 points for levelof function, and 40 points for physical impairment. Exter-nal reliability had a coefficient greater than 0.95. Validitywas determined by dichotomizing the scale into good andbad outcomes. The mean score of the study population was39, and therefore a value greater than 39 implied greaterdysfunction than the mean. The results on the LBPR Scalecorrelated (p , 0.00005) with a Global Assessment Scale(a graded evaluation tool) performed by both patient andphysician.

Stucki, et al.,35 evaluated 193 patients with degenerativelumbar stenosis from multiple centers who were to under-go lumbar decompression. A functional survey was under-taken preoperatively and 6 months after surgery. Interob-server reliability was studied in a random sample of 23patients. Correlation (k) was greater than 0.80 in this group.Internal consistency was greater than 0.80. This lumbaroutcome scale was responsive to functional improvementin this cohort of patients when reassessed 6 months follow-ing surgery. Comparison to the SIP and the VAS for painconfirmed the validity of this instrument in detecting over-all dysfunction associated with lumbar stenosis.

Bernstein and colleagues2 followed 291 patients withchronic low-back pain by using the 90-item SymptomChecklist, which measures psychological dysfunction. Ithas nine major scales with one common factor—generalpsychological discomfort. The somatization scale coversgeneral physical discomfort. The reliability of this check-list was not reported in this study, but validity was ascer-tained by comparison with the Minnesota Multiphasic In-ventory and the McGill Pain Inventory. In this group ofpatients, the scale had a high correlation with the Minne-sota Multiphasic Inventory and McGill Inventory scalesfor detecting general discomfort; however, external relia-bility was not reported. In a 5-year period, Greenough andFraser16 studied 300 patients with low-back pain by usinga Low-Back Outcome Score that examined 13 functionalfactors related to pain. Comparison was made to the ODI


and Waddell–Main scales. Despite a statement that exter-nal reliability was studied, no mention was made of thestatistical analysis in their study. This scale had a high cor-relation with the ODI (20.87; p , 0.001) and Waddell–Main scale (20.74; p , 0.001). Moffroid and colleagues28

assessed 115 patients undergoing physical therapy refer-red for low-back pain, 112 asymptomatic volunteers wereused as a control group. The physical capabilities of bothgroups were quantified using the National Institute for Oc-cupational Safety and Health Low Back Atlas score. Al-though external reliability was described, it was not spe-cifically reported in this study. The authors did find clustersof patients with imbalances in muscle strength and symme-try. Those patients were more apt to suffer from low-backpain.

General health may be measured in addition to low-back pain. In addition to the use of the SIP as a generalhealth measure, Brazier and colleagues3 studied the SF-36Scale in 1582 patients in a general medical practice. TheSF-36 Scale focuses on functional status, general well-ness, and an overall assessment of health in eight domainsby asking 36 questions. Results were compared for valid-ity with the Nottingham Scale. In the general population,the external reliability coefficient was greater than 0.60.Construct validity was determined through a correlationwith the Nottingham Scale (r . 0.50). Ware and col-leagues38 used regression methods to shorten the SF-36 toa 12-item format (SF-12) focusing on physical and mentalaspects. Reliability in an initial evaluation of two differentsets of patients was strong (k . 0.80). Luo and col-leagues25 used the SF-12 in 2520 patients with low-backpain. Although no external reliability was performed inthis setting, internal consistency was sound, and the SF-12appeared valid and responsive to changes in patients withlow-back pain.

Salen, et al.,33 assessed 1092 healthy volunteers andcompared observation with 306 patients with axial skele-tal pain or 47 with joint pain by using a DRI. External reli-ability for this group was greater than 0.80. The DRI wasvalid with correlation to the FSQ. The DRI was respon-sive in detecting improvement after joint replacement.

Examples of the Application of Functional Assessments toLumbar Fusion

The appropriateness of an outcome instrument designedto assess low-back pain does not necessarily generalize to the assessment of patients treated with lumbar spinalfusion procedures. Despite this fact, these same outcomemeasures have been used to assess outcome followinglumbar fusion procedures. In an attempt to correct thisapparent deficiency, many investigators have used multi-ple outcome instruments for correlation.

Several groups have used more formalized methods ofassessing patient outcome. Moller and Hedlund29 studied111 patients with isthmic spondylolisthesis and a 1-yearhistory of back or leg pain. Patients were randomized tosurgery (80 cases) or exercise (34 cases). Evaluation wascompleted at 1 and 2 years by using the DRI and a patientassessment survey involving broad categories (much bet-ter, better, unchanged, or worse). In this patient popula-tion, the DRI appeared responsive with improvement inthe surgical group at 12 and 24 months (p , 0.0001,

Mann–Whitney U-test). Similarly, the broad patient assess-ment survey revealed that a higher proportion of “good”responses occurred in the surgery group (p , 0.01). In asimilar cohort study, Christensen and colleagues6 followed129 patients with chronic low-back pain and either isthmicspondylolisthesis, primary lumbar degeneration, or second-ary lumbar degeneration. Comparison was made betweenposterior fusion with and without instrumentation by usingthe DPQ and LBPR Scale in a 5-year period. Patients inboth groups improved significantly from their preopera-tive status on the DPQ during this period. With the excep-tion of patients with isthmic spondylolisthesis, no differ-ences were observed between groups when using the DPQor LBPR Scale. For patients with isthmic spondylolisthe-sis, fusion without instrumentation resulted in significant-ly better results as measured by the DPQ.

In a different cohort study, Fritzell and colleagues14

studied 294 patients with L4–S1 disc degeneration andlow-back pain who underwent surgical (222) or expectant(72) management during a 6-year period. Evaluation wascompleted at 6, 12, and 24 months by using the ODI,Million, and General Function Score Scales. Disabilitysignificantly decreased in the surgical group over a 2-yearperiod when assessed using all of these scales (p , 0.02).Using a general, subjective assessment, 63% in the surgi-cal group indicated they were better or much better com-pared with 29% in the nonsurgical group (p , 0.0001).Burkus, et al.,4 reported on 46 patients randomized toanterior interbody fusion with or without bone morpho-genetic protein–2. Outcome was recorded over a 24-monthperiod by using the ODI, SF-36, and satisfaction scales.Neurological function, satisfaction, and general healthmeasures were no different between groups. The ODIscore indicated an improvement in the bone morphogen-etic protein–2 group as early as 3 months after surgery.These outcome measures were responsive to low-backpain after lumbar fusion, and the use of multiple outcomemeasures conferred apparent validity.

Other Outcome Measures

Turner and colleagues36 undertook a metaanalysis of alllumbar fusion Medline literature published between 1966and 1991. Studies were required to have a minimum 1-year follow-up period and classification of clinical out-come as satisfactory or unsatisfactory in at least 30 patients.Forty-seven articles met their inclusion criteria. No ran-domized trials were identified at that time. A mean of 68%of the patients had a satisfactory outcome (range 16–95%).Substratification revealed outcomes of excellent/good in66% (range 16–93%), fair in 22% (range 5–68%), and poorin 13% (range 2–54%). No defined criteria were reportedfor external reliability. Their analysis demonstrates thatoutcomes may be dichotomized into broad categories toassess overall outcome following lumbar fusion.

Patient satisfaction has been used as an outcome mea-sure for patients undergoing lumbar fusion. Patient satis-faction surveys are frequently used in the setting of retro-spective series because preintervention data may not beavailable. Patient satisfaction is easily surveyed but is dependent on multiple external factors independent of thesurgical procedure. Furthermore, satisfaction outcomemeasures are hampered by the inherent inability to measure




responsiveness. The validity of satisfaction measures hasbeen examined but their external reliability has not.

Slosar and colleagues34 followed 141 patients who un-derwent circumferential lumbar fusion. A satisfaction sur-vey was used as a follow-up instrument, as was return toemployment. Patients were asked if: 1) surgery met theirexpectations; 2) surgery improved their condition; 3) sur-gery improved their condition but they would not repeat it;and 4) surgery worsened their condition. One hundred thir-ty-three patients were followed for more than 37 months.The outcomes were classified as follows: 10.5% in Cate-gory 1, 51.1% in Category 2, 19.5% in Category 3, and18.8% in Category 4. Christensen and colleagues5 followed148 patients who underwent posterior lumbar fusion withor without supplemental anterior interbody fusion. Satis-faction surveys and the DPQ and LBPR Scale were used.In addition to improvements on the LBPR Scale and DPQ,satisfaction was high in both groups, with 77% of patientsin the posterior fusion group and 79% of patients in the cir-cumferential fusion group stating they would undergo sur-gery again if indicated.

In a study of 388 Workers’ Compensation patients inWashington state, Franklin and colleagues13 undertook anassessment of broad satisfaction surveys. Simple surveysexamined back and leg pain, QOL, and the decision to un-dergo surgery at 2 years following lumbar fusion. Patientswere dichotomized into two outcome groups: poor (re-ceiving Workers’ Compensation) and good (not receivingWorkers’ Compensation) at 2 years. There was a higher in-cidence of poor outcomes among those who stated thatback or leg pain was worse than expected (76% comparedwith 54%; p , 0.0003) and in those whose QOL was nobetter or worse than expected (69% compared with 34%; p , 0.0001). There was a lower incidence of poor out-comes in patients who would undergo surgery again for thesame indications (52% compared with 80%; p , 0.0001).

Although patient satisfaction surveys are easy and areintuitively valuable, they have never been validated andthe responsiveness of such measures cannot be measured.Furthermore, wide discrepancies exist when results ofpatient satisfaction surveys are compared with validatedoutcome measures. These inadequacies limit their abilityto provide high-quality medical evidence for or againstany treatment modality.

Summary

Functional disability secondary to acute low-back pain,chronic low-back pain, lumbar stenosis, and lumbar discdisease may be reliably and validly assessed using func-tional outcome surveys that are valid, reliable, and respon-sive. Outcome instruments supported by Class I and ClassII medical evidence for the evaluation of low-back paininclude the Spinal Stenosis Survey of Stucki, Waddell–Main, RMDQ, DPQ, QPDS, SIP, Million Scale, LBPRScale, ODI, and CBSQ. Many of these outcome measureshave been applied to patients who have been treated withlumbar fusion for degenerative lumbar disease and haveproven to be valid and responsive; however, the reliabilityof these instruments has never been specifically assessed inthe lumbar fusion patient population. Patient satisfactionsurveys have been used to measure outcome followinglumbar fusion. Their usefulness resides in their insight in-

to patient attitudes toward the treatment experience but islimited because of their inability to measure responsive-ness and the lack of information on their reliability.

Key Issues for Future Investigation

Although the functional outcome instruments discussedin this review appear valid and responsive in the low-backpain patient population, their external reliability has notbeen confirmed in the clinical setting of lumbar fusion.This may be important for the comparison of differentlumbar fusion techniques. Another key issue appears to bethe timing of administration of the outcomes instruments.The aforementioned functional outcome measures appearto be responsive both initially and over a few years. Whet-her the benefits associated with any sort of interventionfor low-back pain are durable beyond this period has notbeen established.

References

1. Bergner M, Bobbitt RA, Carter WB, et al: The Sickness ImpactProfile: development and final revision of a health status mea-sure. Med Care 19:787–805, 1981

2. Bernstein IH, Jaremko ME, Hinkley BS: On the utility of theSCL-90-R with low-back pain patients. Spine 19:42–48, 1994

3. Brazier JE, Harper R, Jones NM, et al: Validating the SF-36health survey questionnaire: new outcome measure for primarycare. BMJ 305:160–164, 1992

4. Burkus JK, Transfeldt EE, Kitchel SH, et al: Clinical and radio-graphic outcomes of anterior lumbar interbody fusion using re-combinant human bone morphogenetic protein-2. Spine 27:2396–2408, 2002

5. Christensen FB, Hansen E, Eiskjaer SP, et al: Circumferentiallumbar spinal fusion with Brantigan cage versus posterolateralfusion with titanium Cotrel-Dubousset instrumentation: a pros-pective, randomized clinical study of 146 patients. Spine 27:2674–2683, 2002

6. Christensen FB, Hansen ES, Laursen M, et al: Long-term func-tional outcome of pedicle screw instrumentation as a supportfor posterolateral spinal fusion. Spine 27:1269–1277, 2002

7. Cronbach LJ: Coefficient alpha and the internal structure oftests. Psychometrika 16:297–334, 1951

8. Daltroy LH, Cats-Baril W, Katz JN, et al: The North AmericanSpine Society Lumbar Spine Outcome Assessment Instrument:Reliability and Validity Tests. Spine 21:741–748, 1996

9. Deyo RA: Comparative validity of the sickness impact profileand shorter scales for functional assessment in low-back pain.Spine 11:951–954, 1986

10. Deyo RA, Andersson G, Bombardier C, et al: Outcome mea-sures for studying patients with low back pain. Spine 19 (18Suppl):S2032–S2036, 1994

11. Deyo RA, Diehr P, Patrick DL: Reproducibility and responsive-ness of health status measures. Statistics and strategies for evalu-ation. Control Clin Trials 12 (4 Suppl):S142–S158, 1991

12. Fairbank JC, Couper J, Davies JB, et al: The Oswestry low backpain disability questionnaire. Physiotherapy 66:271–273, 1980

13. Franklin GM, Haug J, Heyer NJ, et al: Outcome of lumbar fu-sion in Washington State workers’ compensation. Spine 19:1897–1904, 1994

14. Fritzell P, Hagg O, Wessberg P, et al: 2001 Volvo Award Win-ner in Clinical Studies: Lumbar fusion versus nonsurgical treat-ment for chronic low back pain: a multicenter randomized controlled trial from the Swedish Lumbar Spine Study Group.Spine 26:2521–2534, 2001

15. Fujiwara A, Kobayashi N, Saiki K, et al: Association of the Jap-anese Orthopaedic Association score with the Oswestry Dis-


Functional Outcome

645


ability Index, Roland-Morris Disability Questionnaire, andShort-Form 36. Spine 28:1601–1607, 2003

16. Greenough CG, Fraser RD: Assessment of outcome in patientswith low-back pain. Spine 17:36–41, 1992

17. Greenough CG, Peterson MD, Hadlow S, et al: Instrumentedposterolateral lumbar fusion. Results and comparison with ante-rior interbody fusion. Spine 23:479–486, 1998

18. Hadley MN, Walters BC, Grabb PA: Guidelines for the man-agement of acute cervical spine and spinal cord injuries. Neu-rosurgery 50 (Suppl):S2–S6, 2002

19. Harper AC, Harper DA, Lambert LJ, et al: Development andvalidation of the Curtin Back Screening Questionnaire (CBSQ):a discriminative disability measure. Pain 60:73–81, 1995

20. Katz JN: Lumbar spinal fusion. Surgical rates, costs, and com-plications. Spine 20 (24 Suppl):S78S–S83, 1995

21. Kopec JA, Esdaile J: Functional disability scales for back pain.Spine 20:1943–1949, 1995

22. Kopec JA, Esdaile J, Abrahamowicz M, et al: The Quebec BackPain Disability Scale. Measurement properties. Spine 20:341–352,1995

23. Lawlis GF, Cuencas R, Selby D, et al: The development of theDallas Pain Questionnaire. An assessment of the impact ofspinal pain on behavior. Spine 14:511–516, 1989

24. Leclaire R, Blier F, Fortin L, et al: A cross-sectional study com-paring the Oswestry and Roland-Morris Functional Disabilityscales in two populations of patients with low back pain of dif-ferent levels of severity. Spine 22:68–71, 1997

25. Luo X, Lynn George M, Kakouras I, et al: Reliability, validity,and responsiveness of the short form 12-item survey (SF-12) inpatients with back pain. Spine 28:1739–1745, 2003

26. Manniche C, Asmussen K, Lauritsen B, et al: Low Back PainRating scale: validation of a tool for assessment of low backpain. Pain 57:317–326, 1994

27. Million R, Hall W, Nilsen KH, et al: Assessment of the progressof the back-pain patient. 1981 Volvo Award in Clinical Scien-ces. Spine 7:204–208, 1982

28. Moffroid MT, Haugh LD, Henry SM, et al: Distinguishablegroups of musculoskeletal low back pain patients and asymptom-atic control subjects based on physical measures of the NIOSH

Low Back Atlas. Spine 19:1350–1358, 1994 (Erratum in Spine19:2137, 1994)

29. Moller H, Hedlund R: Surgery versus conservative managementin adult isthmic spondylolisthesis—a prospective, randomizedstudy: part 1. Spine 25:1711–1715, 2000

30. Roland M, Morris R: A study of the natural history of back pain.Part I: development of a reliable and sensitive measure of dis-ability in low-back pain. Spine 8:141–144, 1983

31. Roland M, Morris R: A study of the natural history of low-backpain. Part II: development of guidelines for trials of treatmentin primary care. Spine 8:145–150, 1983

32. Ruta DA, Garratt AM, Wardlaw D, Russell IT: Developing avalid and reliable measure of health outcome for patients withlow back pain. Spine 19:1887–1896, 1994

33. Salen BA, Spangfort EV, Nygren AL, et al: The DisabilityRating Index: an instrument for the assessment of disability inclinical settings. J Clin Epidemiol 47:1423–1435, 1994

34. Slosar PJ, Reynolds JB, Schofferman J, et al: Patient satisfactionafter circumferential lumbar fusion. Spine 25:722–726, 2000

35. Stucki G, Daltroy L, Liang MH, et al: Measurement propertiesof a self-administered outcome measure in lumbar spinal steno-sis. Spine 21:796–803, 1996

36. Turner JA, Ersek M, Herron L, et al: Patient outcomes afterlumbar spinal fusions. JAMA 268:907–911, 1992

37. Waddell G, Main CJ: Assessment of severity in low-back dis-orders. Spine 9:204–208, 1984

38. Ware JE, Kosinski M, Keller SD: A 12-Item Short-Form HealthSurvey: construction of scales and preliminary tests of reliabil-ity and validity. Med Care 34:220–233, 1996

Manuscript received December 7, 2004.Accepted in final form March 22, 2005.Address reprint requests to: Daniel K. Resnick, M.D., Depart-





Recommendations

Standards. There is insufficient evidence to recommenda standard for assessment of economic outcome followinglumbar fusion for degenerative disease.

Guidelines. There is insufficient evidence to recom-mend a guideline for assessment of economic outcomefollowing lumbar fusion for degenerative disease.

Options. It is recommended that valid and responsiveeconomic outcome measures be included in the assess-ment of outcomes following lumbar fusion surgery fordegenerative disease. Return-to-work rates and termina-tion of disability compensation are two such measures. Itis recommended that cost analyses related to lumbarspinal fusion include perioperative expenses as well as ex-penses associated with long-term care, including thoseincurred in both the operative and nonoperative settings.

Rationale

Lumbar fusion is commonly performed as an adjunct tothe surgical treatment of patients with low-back pain dueto degenerative lumbar disease. Using data from the Na-tional Hospital Discharge Survey, both Deyo, et al.,4 andDavis3 observed a dramatic increase in the frequency of

lumbar fusion procedures in the 1980s. Lumbar fusion hasbeen undertaken in the setting of degenerative disc disease,spinal stenosis, spondylolisthesis, and degenerative scolio-sis and is commonly supplemented with internal fixationinvolving a variety of devices. As the frequency and com-plexity of lumbar fusion surgery increases, there is a ten-dency for costs and complication rates to follow.9 In a timeof contracting hospital resources, it is important to under-stand the economic impact of lumbar fusion. The purposeof this review is to examine the economic impact of lum-bar fusion for degenerative lumbar spine disease asassessed by cost, complication rates, and rates of reopera-tion. These expenses of lumbar fusion must be contrastedwith the return-to-work rate and the potential for im-proved productivity following treatment. These end pointswill be examined as economic outcome measures follow-ing lumbar fusion.

Search Criteria

A computerized search of the National Library of Me-dicine database of the literature published between 1966and 2001 was performed. A search using the subject head-ing “lumbar fusion” yielded 3708 citations. The followingsubject headings were combined: “lumbar fusion and out-comes.” Approximately 204 citations were acquired. Onlycitations in English were selected. A search of this set of publications with the key words “employment status,”“mortality,” “medical care costs,” “cost containment/com-



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 3: assessment of economic outcome



KEY WORDS • fusion • lumbar spine • practice guidelines • treatment outcome •economic outcome

647

Abbreviations used in this paper: CI = confidence interval; LOS = length of stay; QOL = quality of life; RR = relative risk.


parison,” or “cost effectiveness” resulted in 58 matches.Titles and abstracts of the articles were reviewed. Clinicalseries dealing with adult patients who had lumbar fusionfor degenerative disease were selected. Additional refer-ences were culled from the reference lists of remainingarticles.

Among the articles reviewed, 13 studies were includedthat dealt with lumbar fusion, complication rates, reopera-tion rates, and costs. Six of these articles were cohort stud-ies that examined the economic impact of lumbar fusioncompared with surgery for degenerative lumbar diseasethat did not involve fusion. One article was a cohort studyinvestigating fusion with and without fixation. Two stud-ies examined the cost benefit or cost effectiveness of lum-bar fusion compared with decompression alone. The re-maining study examined the responsiveness of returningto work as an economic indicator in a large series. Thesearticles are summarized in Table 1.


One of the more difficult results to ascertain followinga medical or surgical treatment is economic outcome. Typ-ical medical economic analyses seek to ascertain whethera given treatment-related benefit accrues in light of theexpenditures required to provide that treatment. With re-gard to lumbar fusion procedures, benefits from treatmentmay include an overall improvement in low-back pain andfunction, an increased return-to-work rate, and/or improv-ed patient satisfaction. The expenses of the procedures arethe measured costs of the surgery, the devices implanted,and operative time. Other measurable outlays include thecost of complications and time and expenses associatedwith reoperation. Deyo and colleagues4 examined dataconcerning lumbar spinal disease and lumbar spinal fu-sion from the National Hospital Discharge Survey be-tween 1979 and 1987. In addition to a 200% increase inspinal fusion procedures performed during this period, the authors reported significant regional variations in theperformance of lumbar fusion procedures as reflected bya ninefold regional variation in frequency between thenortheastern US (four/100,000) and the western US (35/100,000). Because of the increasing incidence of lumbarfusion procedures in the treatment of degenerative spinedisease, it is important to examine the economic impact oflumbar fusion as a specific outcome measure.

Costs, Complications, Hospitalizations, and Reoperations

Malter and colleagues12 performed a population-basedstudy of patients who underwent lumbar surgery for de-generative disease in Washington state in 1988. The studywas not prospective, nor was it clear that all patients wereeligible for all therapies. Using diagnosis and procedurecodes from the Washington State Department of Health’scomputerized system, the authors obtained data on 6376patients of whom 1041 underwent lumbar fusion. Rates ofreoperation, complications, and associated costs (in 1988US dollars) were examined through the next 5 years. Thecomplication rates associated with lumbar arthrodesis pro-cedures were 18% compared with a 7% complication ratefollowing lumbar surgery without arthrodesis (chi-squaretest, p , 0.001). The LOS was significantly longer for

fusion-treated patients (7 days compared with 5.1 days; p , 0.001). In 1988 dollars, hospital costs averaged $7101per patient treated with fusion and $4161 per patient treat-ed without fusion (p , 0.001). These authors examinedreoperation rates to determine if fusion reduced the needfor repeated lumbar surgery within 5 years. Reoperationrates were similar between those treated with fusion (RR1.1; 95% CI 0.9–1.3) and those not. Because the indica-tions for surgery were not examined, the only conclusionsthat could be drawn from this study were that lumbarfusion procedures are associated with increased costs andcomplications.

Using similar methods and a hospital discharge registryin Washington state, Deyo and colleagues5 examined18,122 hospitalizations for lumbar surgery between 1986and 1988. The majority (84%) of cases requiring surgeryinvolved spinal stenosis or disc displacement. Excludedwere cases involving malignant lesions, infection, or frac-tures. Approximately 15% of patients in this cohort under-went arthrodesis in addition to decompression. The report-ed mortality rate was less than 1%. The complication ratewas 17.4% among patients treated with fusion comparedwith a 7.6% rate for those with lumbar disease treated sur-gically without fusion (chi square test, p , 0.0005). TheLOS among patients who were treated with fusion wasapproximately 7.6 days compared with 5.4 days for thosewho did not undergo fusion (p , 0.0005). In 1986 to 1988dollars, the cost of hospitalization was $6491 for fusion-treated patients compared with $3793 for patients treatedsurgically without fusion (p = 0.0005). Logistic-regres-sion models were used to examine the risk of complica-tions or prolonged hospitalization and indicated that theRR for a complication or prolonged hospitalization withany type of lumbar fusion procedure was 2.7 (95% CI1.5–4.9). The lack of information regarding the indica-tions for surgery and the clinical outcome followingsurgery limit the usefulness of this information.

Deyo, et al.,6 examined lumbar surgery data for 1985obtained from the Health Care Financing Administrationfor all Medicare recipients, excluding those on Medicarefor chronic renal failure or Social Security Disability.Using ICD-9-CM diagnosis and procedure codes, datawere accrued on the frequency of lumbar surgery per-formed with or without fusion and the incidence of asso-ciated complications. The study was not undertaken pros-pectively nor was it certain that all patients were eligiblefor all therapies. Specific data were obtained for 6-weekmortality rates, requirements for assisted living, and theneed for blood transfusion. An economic analysis wascompleted for LOS and cost. These data were comparedwith similar data from 1 year prior to 4 years after thestudy date. A study population of 27,111 patients wasobtained of whom 1524 (5.6%) underwent lumbar fusion.For patients treated surgically with fusion, the mean hos-pital costs (1985 US dollars) were $10,091 compared with$6754 for patients treated without fusion (chi-square test,p , 0.0005). A logistic regression was completed to deter-mine RR and (95%) CIs for several variables. In the fu-sion group, the RR was 1.9 (95% CI 1.6–2.2) for the pres-ence of complications, 5.8 for blood transfusion (5.2–6.6),2.0 for 6-week mortality (1.2–3.4), and 2.2 for dischargeto a nursing home (1.7–3.0). This cohort study revealedthat lumbar surgery with fusion was more expensive and





Economic Outcome

649

TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

ass

essm

ent

of e

cono

mic

out

com

e af

ter

lum

bar

spin

al s

urge

ry*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nR

esul

tsC

oncl

usio

ns

Tun

turi

, et a

l., 1

979

III

133

patie

nts

unde

rwen

t lum

bosa

cral

fus

ion

w/ 1

18 F

Us

The

cos

t/ben

efit

ratio

for

lum

bosa

cral

fus

ion

was

1:2

.9

Lum

bosa

cral

fus

ion

in a

sel

ecte

d po

pula

tion

has

a in

clud

ing

2 de

aths

. Cos

ts w

ere

calc

ulat

ed in

197

6 w

/ the

cos

t in

1976

US

dolla

rs a

s $5

569

& b

enef

it as

po

sitiv

e co

st/b

enef

it ra

tio.

US

dolla

rs b

ased

on

peri

op h

ospi

taliz

atio

n &

FU

s.

$16,

075.

Ben

efit

was

def

ined

as

the

time

over

the

mea

n FU

(4

.8 y

rs)

for

whi

ch th

e pa

tient

was

em

ploy

ed &

was

calc

ulat

ed b

ased

on

mea

n sa

lary

dur

ing

this

per

iod.

Dey

o, e

t al.,

199

1II

IA

ll M

edic

are

reci

pien

ts u

nder

goin

g lu

mba

r op

in 1

985.

For

the

fusi

on g

roup

, RR

w/ 9

5% C

I fo

r co

mpl

icat

ions

A

grea

ter

econ

omic

cos

t of

fusi

on in

the

Med

icar

e D

ata

prov

ided

fro

m H

CFA

on th

ese

patie

nts

from

w

as 1

.9 (

1.6–

2.2)

, blo

od tr

ansf

usio

n 5.

8 (5

.2–6

.6),

po

pula

tion.

Lum

bar

fusi

on is

ass

ocia

ted

w/ g

reat

- 1

yr p

reop

to 4

yrs

pos

top.

27,

111

patie

nts

wer

e st

ud-

6-w

k m

orta

lity

2.0

(1.2

–3.4

), a

ssis

ted

livin

g 2.

2 (1

.7–

er m

orbi

dity

, mor

talit

y, &

use

of

hosp

ital r

esou

rces

ied

of w

hom

5.6

% (

1524

) re

ceiv

ed lu

mba

r fu

sion

. 3.

0) (

p ,

0.05

). T

hese

res

ults

wer

e co

nsis

tent

btw

n

in o

lder

adu

lts. N

o cl

ear

coho

rt o

f lu

mba

r de

gen-

Hos

pita

lizat

ions

exa

min

ed f

or c

ompl

icat

ions

, mor

tal-

sp

inal

ste

nosi

s &

spo

ndyl

olis

thes

is w

/ hos

pita

l cos

ts

erat

ive

popu

latio

n de

fine

d fo

r co

st c

ompa

riso

n.ity

at 6

wks

, nee

d fo

r bl

ood

tran

sfus

ion,

&re

quir

e-of

$10

,091

(fu

sion

) vs

$67

54 (

w/o

) in

198

5 do

llars

m

ents

for

ass

iste

d liv

ing.

Eco

nom

ic a

naly

sis

was

&

a s

igni

fica

ntly

sho

rter

LO

S (p

,0.

05 in

eac

h ca

te-

com

plet

e fo

r L

OS

& c

osts

.go

ry).

Dey

o, e

t al.,

199

2II

I18

,122

hos

pita

lizat

ions

for

lum

bar

spin

e op

(84

% in

-~1

5% o

f pa

tient

s un

derw

ent a

rthr

odes

is. T

he c

ompl

ica-

Patie

nts

who

und

ergo

fus

ion

in a

bro

ad p

opul

atio

n vo

lved

spi

nal s

teno

sis

or d

isc

disp

lace

men

t) f

rom

tio

n ra

te w

as 1

7.4%

w/ f

usio

n &

7.6

% w

/o (

p ,

are

mor

e ap

t to

have

long

er L

OSs

w/ g

reat

er c

om-

1986

–198

8. 1

5,28

0 su

rger

ies

w/o

art

hrod

esis

&

0.00

05).

The

LO

S w

as 7

.6 d

ays

w/ f

usio

n &

5.4

w/o

plic

atio

n ra

tes

& u

tiliz

atio

n of

hea

lthca

re r

esou

rc-

2785

incl

uded

art

hrod

esis

. Hos

pita

lizat

ions

exa

min

-(p

,0.

0005

). T

he c

ost i

n 19

86–1

988

dolla

rs w

as $

6491

es. N

o cl

ear

coho

rt o

f lu

mba

r fu

sion

pop

ulat

ion

ed f

or c

ompl

icat

ions

. Eco

nom

ic a

naly

sis

was

com

-w

/ fus

ion

& $

3793

w/o

(p

,0.

0005

). N

o de

tails

wer

ede

fine

d fo

r co

st c

ompa

riso

n.pl

ete

for

LO

S &

cos

ts.

give

n fo

r m

orta

lity.

Fran

klin

, et a

l., 1

994

III

388

patie

nts

in W

orke

rs’C

ompe

nsat

ion

syst

em in

Em

ploy

men

t was

16,

32,

& 4

9% o

ver

1, 2

, & 3

yrs

. It

Em

ploy

men

t as

an e

cono

mic

indi

cato

r m

ay b

e us

ed

Was

hing

ton

stat

e (1

986–

1987

) w

ho u

nder

wen

t fu-

was

less

like

ly to

occ

ur in

this

coh

ort t

han

hist

oric

al

as a

n ou

tcom

e m

easu

re b

ut o

ther

con

trol

gro

ups

sion

. Pat

ient

sat

isfa

ctio

n st

udie

d al

ong

w/ e

cono

mic

cont

rols

(R

R =

0.6

6, 0

.88,

&0.

93)

at 1

, 2, &

3 y

rs;

shou

ld b

e co

nsid

ered

.re

cove

ry b

y pa

tient

. Sim

ple

satis

fact

ion

surv

ey e

xam

-23

% r

equi

red

reop

& in

stru

men

tatio

n do

uble

d th

is r

isk.

ined

bac

k/le

g pa

in, Q

OL

, dec

isio

n to

und

ergo

op,

& e

mpl

oym

ent a

t 2 y

rs.

Kat

z, e

t al.,

199

7II

I27

2 pa

tient

s w

/ deg

ener

ativ

e lu

mba

r st

enos

is. S

urge

ry:

Indi

vidu

al s

urge

on w

as p

redi

ctor

for

art

hrod

esis

. Hos

pi-

Hos

pita

l cos

ts o

f ar

thro

desi

s/fi

xatio

n ar

e hi

ghes

t w/

deco

mpr

essi

on (

194)

, dec

ompr

essi

on w

/ art

hrod

e-ta

l cos

ts w

ere

$12,

615

(no

arth

rode

sis)

, $18

,495

(ar

th-

no c

lear

def

ined

ben

efit.

Art

hrod

esis

alo

ne

sis

(37)

, &de

com

pres

sion

w/ a

rthr

odes

is/f

ixat

ion

rode

sis)

, $25

,914

(ar

thro

desi

s/fi

xatio

n) (

p ,

0.00

01).

sh

owed

impr

oved

rel

ief

of lu

mba

go a

t 6 &

24

(41)

. Out

com

es a

sses

sed

w/ r

espe

ct to

wal

king

ca-

No

relia

bilit

y gi

ven

for

wal

king

, sat

isfa

ctio

n, o

r he

alth

mos

w/o

rel

iabi

lity.

Sig

nifi

cant

var

iabi

lity

intr

o-pa

city

, bac

k/le

g pa

in, s

atis

fact

ion,

hea

lth s

tatu

s

stat

us.

duce

d by

sur

geon

cho

ice

for

arth

rode

sis.

(SF-

36),

& h

ospi

tal c

ost.

Mal

ter,

et a

l., 1

998

III

6376

pat

ient

s ha

d op

for

lum

bar

dege

nera

tive

dise

ase

Com

plic

atio

n ra

te: 1

8% (

arth

rode

sis)

to 7

% (

none

) (p

,T

he e

cono

mic

cos

ts o

f lu

mba

r ar

thro

desi

s in

the

set-

(104

1 fo

r op

incl

udin

g ar

thro

desi

s, 5

335

for

op w

/o

0.00

1). H

ospi

tal c

osts

gre

ater

w/ f

usio

n ($

7101

&

ting

of s

teno

sis,

dis

c di

spla

cem

ent,

spon

dylo

lis-

arth

rode

sis)

. Eco

nom

ic a

naly

sis

of h

ospi

taliz

atio

n.$4

161

in 1

988

dolla

rs)

(p ,

0.00

1). R

eop

rate

sim

ilar

thes

is, &

deg

ener

atio

n ar

e gr

eate

r. N

o cl

ear

coho

rt

btw

n gr

oups

, RR

1.1

(95

% C

I 0.

9–1.

3).

of lu

mba

r deg

ener

ativ

e pa

tient

s us

ed fo

r com

pari

son.

Kun

tz, e

t al.,

200

0II

IA

cost

-eff

ectiv

enes

s st

udy

of la

min

ecto

my,

lam

inec

-T

he Q

ALY

s &

cos

ts w

ere

calc

ulat

ed &

fou

nd to

be

Lum

bar

lam

inec

tom

y w

/ non

inst

rum

ente

d fu

sion

com

-to

my

w/ n

onin

stru

men

ted

fusi

on, &

lam

inec

tom

y $5

6,50

0 fo

r la

min

ecto

my

w/ n

onin

stru

men

ted

fusi

onpa

red

favo

rabl

y w

/ dec

ompr

essi

on. N

ot e

noug

h w

/ ins

trum

ente

d fu

sion

. Out

com

e w

as a

sses

sed

at

com

pare

d w

/ lam

inec

tom

y al

one.

Ins

trum

ente

d fu

sion

da

ta e

xist

ed o

n ou

tcom

e w

/ fix

atio

n to

pre

sent

it

6 m

os &

long

term

& b

ased

on

prio

r re

port

s. P

eri-

w

as s

ubst

antia

lly h

ighe

r ($

3,11

2,80

0). I

mpr

oved

out

-po

sitiv

ely

on a

n ec

onom

ic s

cale

.op

com

plic

atio

ns &

cos

ts &

reo

p ra

tes

wer

e al

l co

me

w/ i

nstr

umen

tatio

n (9

0 vs

80%

) re

duce

d th

e re

l-ba

sed

on p

rior

rep

orts

.at

ive

cost

of

fixa

tion.

Mol

ler

& H

edlu

nd, 2

000

II11

1 pa

tient

s w

/ spo

ndyl

olis

thes

is w

ho u

nder

wen

t fu-

The

fus

ion

& e

xerc

ise

grou

ps h

ad s

imila

r nu

mbe

rs o

f pa

-R

TE

app

ears

to b

e an

indi

cato

r of

impr

ovem

ent.

Asi

on (

77)

or e

xerc

ise

(34)

. Pat

ient

s w

ere

rand

om-

tient

s on

dis

abili

ty a

t 2 y

rs (

46 v

s 45

%);

how

ever

, the

sa

tisfa

ctio

n su

rvey

was

not

rel

iabl

y st

udie

d bu

t iz

ed to

thes

e gr

oups

if th

ey h

ad $

1 yr

of

pain

/sci

-ov

eral

l red

uctio

n w

as g

reat

er f

or f

usio

n (p

,0.

0001

) di

d ap

pear

to b

e a

resp

onsi

ve in

dica

tor

for

out-

atic

a. E

valu

atio

n w

as c

ompl

eted

at 1

& 2

yrs

usi

ng

com

pare

d w

/ exe

rcis

e (p

= 0

.23)

. The

sat

isfa

ctio

n su

r-co

me

& s

atsf

actio

n ha

d im

prov

ed m

ore

afte

r th

e D

RI,

a s

atis

fact

ion

surv

ey (

muc

h be

tter,

bette

r,ve

y sh

owed

goo

d re

spon

ses

to b

e si

gnif

ican

tly h

ighe

r fu

sion

.un

chan

ged,

wor

se; w

ould

you

rep

eat o

p?),

& R

TE

.in

the

op g

roup

(p

,0.

01).

Con

tinu

ed


associated with higher risk for complications than lumbarsurgery without fusion.

Katz and colleagues10 completed a prospective observa-tional study of 272 patients with radiographically and clin-ically documented lumbar stenosis. Patients were treated atfour centers by eight surgeons over a 4-year period. Sur-gical treatment included decompression (194 cases), de-compression with fusion (37 cases), or decompression withfusion and internal fixation (41 cases). Patients were fol-lowed for 24 months and assessed for walking capacity,back and leg pain, satisfaction, and health status based onthe Sickness Impact Profile. Internal reliability was calcu-lated for the walking, pain, and satisfaction scales. Hospitalcosts were analyzed for each group. The individual sur-geon, in this study, was the greatest predictor for the per-formance of a fusion, with an RR of more than 10 based onlogistic regression. At 6 and 24 months, decompressionand fusion without internal fixation resulted in better reliefof back pain (p , 0.004 at 6 months; p , 0.01 at 24months) compared with the other treatment groups. Withmultivariate analysis, a trend was evident but did not reachstatistical significance. The reoperation rates were similarin all three groups (p = 0.15). Mean hospital costs were$12,615 per patient for decompression without fusion,$18,495 per patient for decompression with fusion, and$25,914 per patient for decompression with fusion and in-ternal fixation (p , 0.0001). This study indicated that in-creased costs for lumbar fusion may be offset by significantfunctional gains in the patients who undergo fusion withoutinstrumentation. The medical evidence cited in this report isconsidered Class III because of the retrospective nature ofthe study and selection bias by the operating surgeons as towhich patients were treated with internal fixation.

Return to Employment

Two studies described resumption of employmentamong patients with low-back pain and compared thosetreated with lumbar fusion with those treated nonopera-tively. Moller and Hedlund13 examined 111 patients overa 5-year period who had chronic low-back pain or sciati-ca for a minimum of 1 year as a result of isthmic spondy-lolisthesis. Treatments included arthrodesis with internalfixation (37 cases), arthrodesis without internal fixation(40 cases), or exercise (34 cases). Evaluation was per-formed at 1 and 2 years by using a Disability Rating In-dex, a satisfaction survey, and return-to-work rates. In thefusion and exercise groups there were similar numbers ofpatients receiving disability payments at 2 years (46 and45%, respectively); however, the surgical group had agreater degree of improvement (75 and 46%, respective-ly; p , 0.0001) compared with the exercise group (61 and45%, respectively; p = 0.23). Fritzell, et al.,8 reported astudy of 294 patients with lumbar degenerative diseaseinvolving chronic lumbago of at least 2 years’ durationdue to L4–5 and/or L5–S18 disc degeneration. Patientswere randomized to surgical or nonsurgical groups. Pa-tients in the surgical group underwent posterolateral fu-sion (73 cases), posterolateral fusion with internal fixation(74 cases), or interbody fusion with internal fixation (75cases). Seventy-two patients received medical manage-ment including physical therapy. Evaluation was accom-plished at 6, 12, and 24 months by using functional out-



TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nR

esul

tsC

oncl

usio

ns

Slos

ar, e

t al.,

200

0II

I14

1 pa

tient

s un

derw

ent c

ircu

mfe

rent

ial l

umba

r in

stru

-13

3 FU

pat

ient

s (1

0.5%

= 1

, 51.

1% =

2, 1

9.5%

= 3

, &Sa

tisfa

ctio

n ap

pear

s to

be

a re

spon

sive

out

com

e m

ea-

men

ted

fusi

on, e

ither

pri

mar

y (3

1%)

or s

econ

dary

18

.8%

= 4

). R

TE

occ

urre

d in

38%

of

patie

nts;

it w

as

sure

at 3

7 m

os; h

owev

er, i

ts r

epro

duci

bilit

y w

as

(69%

). F

U a

vera

ged

37 m

os &

was

don

e by

a b

asic

mor

e lik

ely

in th

ose

not i

nvol

ved

w/ W

orke

rs’C

om-

not t

este

d. R

TE

is a

res

pons

ive

mea

sure

& im

-sa

tisfa

ctio

n su

rvey

: 1)

op m

et e

xpec

tatio

ns; 2

) op

pe

nsat

ion

(57

vs 2

2%; p

,0.

001)

. The

re w

as a

20%

pr

oves

in th

e no

ncom

pens

atio

n pa

tient

s. A

com

-im

prov

ed m

y co

nditi

on; 3

) op

impr

oved

but

wou

ld

com

plic

atio

n ra

te in

clud

ed tr

ansi

ent w

eakn

ess,

infe

c-pl

icat

ion

rate

of

20%

was

see

n, s

ugge

stin

g a

nega

-no

t red

o; 4

) op

wor

sene

d co

nditi

on.

tion,

&gr

aft e

xtru

sion

.tiv

e ec

onom

ic im

pact

.

Fritz

ell,

et a

l., 2

001

II29

4 pa

tient

s w

/ L4–

S1 d

isc

dege

nera

tion

& L

BP

who

w

/ ove

rall

asse

ssm

ent,

63%

in th

e op

gro

up in

dica

ted

Satis

fact

ion

surv

eys

seem

to b

e re

spon

sive

ove

r th

e un

derw

ent o

p (2

22)

or e

xpec

tant

(72

) m

anag

emen

t th

ey w

ere

bette

r or

muc

h be

tter

com

pare

d w

/ 29%

sa

me

time

inte

rval

, & R

TE

indi

cate

s re

spon

sive

-ov

er a

6-y

r pe

riod

. Eva

luat

ion

was

com

plet

ed a

t 6,

in th

e no

nop

grou

p (p

,0.

0001

). T

he n

et R

TE

rat

e ne

ss a

t 24

mos

.12

, &24

mos

usi

ng th

e O

DI,

Mill

ion,

& g

ener

al

was

hig

her

in th

e op

gro

up (

36%

) th

an th

e no

nop

func

tion

scor

e al

ong

w/ p

atie

nt a

sses

smen

t. R

TE

grou

p (1

3%, p

,0.

002)

.w

as a

lso

mon

itore

d.C

hris

tens

en, e

t al.,

200

2II

I14

8 pa

tient

s un

derw

ent l

umba

r fu

sion

w/ 7

3 in

PL

F R

TE

was

sim

ilar

in b

oth

grou

ps b

ut im

prov

ed f

rom

24%

RT

E d

id n

ot a

ppea

r to

wor

k as

a r

espo

nsiv

e in

dica

-gr

oup

and

75 in

AL

IF/P

LF

grou

p. T

hree

pat

ient

s to

36%

(no

sta

tistic

s us

ed to

ana

lyze

) w

/ no

diff

eren

ce

tor

of im

prov

emen

t, &

RT

E s

eem

ed to

cor

rela

te

wer

e lo

st to

FU

. RT

E w

as a

lso

follo

wed

at 2

yea

rs.

btw

n su

bgro

ups.

w/ a

n im

prov

emen

t in

DPQ

sco

re. N

o co

ntro

ls

FU p

erfo

rmed

at 0

, 1, &

2 y

rs.

wer

e us

ed.

Chr

iste

nsen

, et a

l., 2

002

II12

9 pa

tient

s w

/ chr

onic

LB

P&

isth

mic

spo

ndyl

olis

-T

he in

stru

men

ted

grou

p ha

d a

28%

reo

p ra

te c

ompa

red

Med

ical

out

com

e by

reo

p &

op

time

may

be

a re

s-

thes

is, p

rim

ary

dege

nera

tion,

or

seco

ndar

y de

gene

r-w

/ 14%

for

the

noni

nstr

umen

ted

grou

p (p

,0.

03).

Op

pons

ive

indi

cato

r w

/in 5

yrs

of

lum

bar

fusi

ons

w/

atio

n w

ho u

nder

wen

t ins

trum

ente

d or

non

inst

ru-

time

212

vs 1

27 m

in (

p ,

0.00

01)

w/ g

reat

er p

erio

p in

stru

men

tatio

n.m

ente

d fu

sion

. Out

com

e at

5 y

rs w

as d

one

usin

g bl

ood

loss

(p

,0.

01).

func

tiona

l que

stio

nnai

res

alon

g w

ith r

ates

of

RT

E

& r

eop.

*A

LIF

= a

nter

ior

lum

bar

inte

rbod

y fu

sion

; D

PQ =

Dal

las

Pain

Que

stio

nnai

re;

FU =

fol

low

up;

HC

FA=

Hea

lth C

are

Fina

ncin

g A

dmin

istr

atio

n; L

BP

= l

ow-b

ack

pain

; O

DI

= O

swes

try

Dis

abili

ty I

ndex

;PL

F =

pos

tero

late

ral f

usio

n; Q

ALY

= q

ualit

y-ad

just

ed li

fe y

ear;

QO

L=

qua

lity

of li

fe; R

TE

= r

etur

n to

em

ploy

men

t; SF

-36

= S

hort

For

m–3

6.


come questionnaires and return-to-work status. In an over-all assessment, 63% in the surgical group indicated theywere better or much better following treatment comparedwith 29% in the nonsurgical group (p , 0.0001). The netreturn-to-work rate was 39% in the surgical group andonly 23% in the nonsurgical group (p , 0.05). These twostudies suggest that the resumption of employment is aresponsive economic outcome measure for patients withlow-back pain who may be considered surgical candi-dates.

In the study by Franklin and colleagues7 of Workers’Compensation patients, the end of total disability wasmonitored in patients who underwent lumbar fusion be-tween 1986 and 1987. The termination of total disability asan end point occurred in 16% of treated patients at 1 year,32% at 2 years, and 49% at 3 years; however, comparedwith historical controls for Workers’ Compensation pa-tients, the RR of ending total disability was less likelyamong patients treated with lumbar fusion (0.66 at 1 year,0.88 at 2 years, and 0.93 at 3 years) compared with Work-ers’ Compensation patients as a whole. In contrast, Chris-tensen, et al.,1 examined 148 patients who underwent lum-bar fusion over a 3-year period: posterolateral fusion (73cases) or a combination of posterolateral and anterior inter-body fusion procedures (75 cases). Outcome was assessedover 2 years. Overall improvement was greatest in the cir-cumferential treatment group, with a lower reoperation rate(22% compared with 7%, p , 0.009). The return-to-workrate improved in both groups from 24 to 36% with no dif-ference between subgroups. No statistical analyses wereused to assess the overall return-to-work rate. Slosar, etal.,14 reported on 133 patients who underwent circumferen-tial fusion during a 2-year follow-up period. In this group,50 patients (38%) returned to work; 16 (22%) of the 73injured workers resumed work compared with 34 (57%) ofthe 60 patients who were not receiving Workers’ Com-pensation (chi-square test, p , 0.001). These studies indi-cated that return to work and/or termination of disabilitypayment are responsive measures for economic outcomeafter lumbar fusion procedures. They further indicate thatthe presence of a compensable injury is associated with alower rate of return to work.

Cost–Benefit Analysis

Kuntz and colleagues11 undertook a cost-effectivenessanalysis of lumbar fusion by constructing a hypotheticalmodel based on historical reports in prior clinical studies.They examined lumbar laminectomy, laminectomy withnoninstrumented fusion, and laminectomy with instru-mented fusion. Rates of clinical improvement and returnto employment were culled from series reported in the lit-erature as were costs, complication rates, fusion rates, re-operation rates, and the incidence of clinical worsening.Each negative and positive outcome was assigned a rela-tive value pertaining to quality of life, which the authorsadjusted according to hypothetical outcomes.

The authors determined that laminectomy with non-instrumented fusion cost $56,500 per quality-adjustedyear of life compared with laminectomy alone.11 The addi-tion of instrumentation to lumbar fusion procedure cost$3,112,800 per quality-adjusted year of life. The authorsconcluded that laminectomy with noninstrumented fusion

compared favorably with decompression alone; however,improvement in outcome associated with instrumentationwas not well defined enough to accrue a benefit. Theauthors noted that a hypothetical rate of 90% symptomrelief for patients treated with instrumented fusion com-pared with 80% for noninstrumented patients would re-duce the quality-adjusted year of life cost to $82,400.

Tunturi and colleagues15 analyzed 133 consecutive pa-tients who underwent lumbosacral fusion between 1968and 1975. Results were reported for 116 patients in whomthe mean follow-up period was 4.8 years. These authorscalculated the mean expense of the hospital stay and post-operative visits in 1976 dollars. Economic benefits werecalculated based on return-to-employment rates comparedwith the costs of continued disability. The rate of return toemployment was approximately 31%. The mean cost in1976 US dollars for a lumbosacral fusion was $5569. Themean economic benefit in 1976 US dollars for the sameperiod was $16,075. The calculated cost/benefit ratio wastherefore 1:2.9 for lumbosacral arthrodesis. The authorsconcluded that lumbosacral fusion had a positive cost–be-nefit ratio when return-to-employment status and the ter-mination of disability payment were considered as indicesof economic outcome.

Summary

Lumbar fusion may be associated with a high short-term cost, especially if instrumentation is placed; how-ever, there appear to be long-term economic benefits as-sociated with lumbar fusion including resumption ofemployment. To describe the economic impact of lumbarfusion for degenerative disease adequately, it is importantto define the patient population treated with fusion and tocompare efficacy as well as the costs of other treatmentalternatives. Any such analysis should include both short-and long-term costs and benefits.

Key Issues for Future Investigation

The application of valid and reliable outcome measuresin conjunction with a complete short- and long-term eco-nomic analysis will be necessary to assess fully the economic impact of lumbar fusion. To reach meaningfulconclusions, it is imperative to compare the economic out-comes of patients treated with lumbar fusion with those inpatients with similar disease treated without fusion and toinclude all relevant costs. This analysis should include sub-sequent operative and nonoperative medical care, ongoingdisability costs, and the costs of loss of productivity. Mea-sures such as return-to-work status and quality-adjusted lifeyears must be included in to allow the development ofmeaningful data.

References

1. Christensen FB, Hansen ES, Eiskjaer SP, et al: Circumferentiallumbar spinal fusion with Brantigan cage versus posterolateralfusion with titanium Cotrel-Dubousset instrumentation: a pros-pective, randomized clinical study of 146 patients. Spine 27:2674–2683, 2002

2. Christensen FB, Hansen ES, Laursen M, et al: Long-term func-tional outcome of pedicle screw instrumentation as a support


Economic Outcome

651


for posterolateral spinal fusion: randomized clinical study witha 5-year follow-up. Spine 27:1269–1277, 2002

3. Davis H: Increasing rates of cervical and lumbar spine surgeryin the United States, 1979–1990. Spine 19:1117–1124, 1994

4. Deyo RA, Cherkin D, Conrad D, et al: Cost, controversy, crisis:low back pain and the health of the public. Annu Rev PublicHealth 12:141–156, 1991

5. Deyo RA, Cherkin D, Loeser JD, et al: Morbidity and mortali-ty in association with operations on the lumbar spine. The influ-ence of age, diagnosis, and procedure. J Bone Joint Surg Am74:536–543, 1992

6. Deyo RA, Ciol MA, Cherkin DC, et al: Lumbar spinal fusion.A cohort study of complications, reoperations, and resource usein the Medicare population. Spine 18:1463–1470, 1993

7. Franklin GM, Haug J, Heyer NJ, et al: Outcome of lumbarfusion in Washington State workers’ compensation. Spine 19:1897–1904, 1994

8. Fritzell P, Hagg O, Wessberg P, et al: 2001 Volvo AwardWinner in Clinical Studies: Lumbar fusion versus nonsurgicaltreatment for chronic low back pain: a multicenter randomizedcontrolled trial from the Swedish Lumbar Spine Study Group.Spine 26:2521–2534, 2001

9. Katz JN: Lumbar spinal fusion. Surgical rates, costs, and com-plications. Spine 20 (24 Suppl):S78–S83, 1995

10. Katz JN, Lipson SJ, Lew RA, et al: Lumbar laminectomy aloneor with instrumented or noninstrumented arthrodesis in degen-

erative lumbar spinal stenosis. Patient selection, costs, and sur-gical outcomes. Spine 22:1123–1131, 1997

11. Kuntz KM, Snider RK, Weinstein JN, et al: Cost-effectivenessof fusion with and without instrumentation for patients with de-generative spondylolisthesis and spinal stenosis. Spine 25:1132–1139, 2000

12. Malter AD, McNeney B, Loeser JD, et al: 5-year reoperationrates after different types of lumbar spine surgery. Spine 23:814–820, 1998

13. Moller H, Hedlund R: Surgery versus conservative manage-ment in adult isthmic spondylolisthesis—a prospective, ran-domized study: part 1. Spine 25:1711–1715, 2000

14. Slosar PJ, Reynolds JB, Schofferman J, et al: Patient satisfac-tion after circumferential lumbar fusion. Spine 25:722–726,2000

15. Tunturi T, Niemela P, Laurinkari J, et al: Cost-benefit analysisof posterior fusion of the lumbosacral spine. Acta OrthopScand 50:427–432, 1979






Recommendations

Standards. Static lumbar radiographs are not recom-mended as a stand-alone means to assess fusion status fol-lowing lumbar arthrodesis surgery.

Guidelines. 1) Lateral flexion and extension radiogra-phy is recommended as an adjunct to determine the pres-ence of lumbar fusion postoperatively. The lack of motionbetween vertebrae, in the absence of rigid instrumenta-tion, is highly suggestive of successful fusion. 2) Tech-netium-99 bone scanning is not recommended as a meansto assess lumbar fusion.

Options. Several radiographic techniques, including sta-tic radiography, lateral flexion–extension radiography, and/or CT scanning, often in combination, are recommendedas assessment modality options for the noninvasive evalu-ation of symptomatic patients in whom failed lumbar fu-sion is suspected.

Rationale

Lumbar fusion is performed in patients with pain due tolumbar degenerative disease. An outcome measure fre-quently cited in studies evaluating lumbar fusion tech-niques is the “radiographic fusion rate;” however, radio-

graphic fusion is not consistently defined throughout theliterature. The purpose of this review is to examine the lit-erature regarding the ability of various diagnostic tech-niques to assess fusion status after lumbar fusion is per-formed to treat degenerative disease.

Search Criteria

A computerized search of the database of the NationalLibrary of Medicine between 1966 and July 2003 was con-ducted using the search terms “lumbar spine fusion assess-ment,” “lumbar spine pseudoarthrosis,” or “lumbar spinefusion outcome.” The search was restricted to references inthe English language involving humans. This yielded a to-tal of 1076 references. The titles and abstracts of each ofthese references were reviewed. Only papers concernedwith the assessment of fusion status following arthrodesisprocedures for degenerative lumbar disease were included.Additional articles were obtained from the bibliographiesof the selected articles. Forty-five references were identi-fied that provided either direct or supporting evidence rele-vant to the radiographic assessment of lumbar fusion status.Reports involving Class III or better medical evidence arelisted in Table 1. Supportive data are provided by addition-al references listed in the bibliography.


Open surgical exploration is the only method that al-lows direct inspection of fusion integrity. This procedure



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 4: radiographic assessment of fusion



KEY WORDS • lumbar spine • fusion • radiography • treatment outcome •practice guidelines

653

Abbreviations used in this paper: CT = computerized tomogra-phy; NPV = negative predictive value; PPV = positive predictivevalue; RSA = roentgen stereophotogrammetric analysis.


is considered the gold standard of lumbar fusion assess-ment.6,7 It is, therefore, an appropriate benchmark to use inestablishing the accuracy and predictive value of noninva-sive radiolographic studies for the assessment of fusionstatus following attempted lumbar fusion surgery.

Plain Radiographs (static)

Anteroposterior and lateral radiographs can demon-strate a continuous bone mass between adjacent vertebralsegments following lumbar fusion. Because of their rela-tively low cost, widespread availability, and long historyas a means of assessing fusion, plain spinal radiographyremains a common method of assessment of lumbar fu-sion;6 however, the limitations of static plain radiographyas a reliable test for determining the presence or absenceof a solid fusion have been well documented. Brodsky, et al.,3 reported a 64% correlation between preoperativeplain radiographs and surgical exploration in a retrospec-tive study of 214 lumbar fusion exploration procedures inpatients who had undergone prior posterolateral fusion.Plain radiography had an 89% sensitivity and 60% speci-ficity for predicting solid fusion. Radiographs interpretedas demonstrating fusion had a PPV of 76%. Those predict-ing pseudarthrosis had an NPV of 78%. These data indicatea 0.18 likelihood ratio for a false-positive result (chance ofa pseudarthrosis discovered at exploration when radiogra-phy indicates fusion), and a 2.25 likelihood ratio for a neg-ative test result (chance of a fusion discovered at explo-ration when the radiography suggests pseudarthrosis).3 Themedical evidence provided by this review is consideredClass II for the use of plain lumbar radiography comparedwith open surgical exploration to assess fusion because ofthe authors’ selection bias for open exploration.

Similarly, in a retrospective study of 75 patients, Kant andcoworkers11 found a positive correlation between static radi-ography and surgical exploration of lumbar fusion in 68% oftheir patients (sensitivity 85%, specificity 62%, PPV 76%,and NPV 54%). The likelihood ratio for a positive result was0.81, and the likelihood ratio for a negative result was 2.24.Finally, in a study of 49 patients treated with posterolateraland posterior interbody fusion with internal fixation, Blu-menthal and Gill1 compared findings on anteroposterior andlateral radiographs (interpreted by two surgeons and two ra-diologists) with surgical exploration of the fusion mass at thetime of reoperation for hardware removal. They reported a69% agreement between the radiographic diagnosis and sur-gical findings. The accuracy among the four physicians in-terpreting the radiographs ranged from 57 to 77% (false-pos-itive rate 42%, false-negative rate 29%). These authorsconcluded that plain radiography has limited accuracy andvalidity for the assessment of lumbar fusion. Furthermore,they noted significant intra- and interobserver variation, indi-cating a lack of reliability (k 0.4–0.7). Their study providesClass I medical evidence indicating that static radiography isonly accurate in determining fusion status in roughly twothirds of cases. Therefore, static anteroposterior and lateralradiographs are not recommended as a stand-alone assess-ment of the presence of an arthrodesis after lumbar fusionsurgery for degenerative disease.

Flexion–Extension Radiography

In 1948 Cleveland, et al.,6 advocated the use of dynam-

ic lumbar spinal radiography rather than static radiography,for the diagnosis of pseudarthrosis following attemptedlumbar fusion surgery. Other authors have also suggestedthat lateral lumbar flexion–extension radiography allowsfor appropriate assessment of fusion status.4 There has beendisagreement, however, on the number of allowable de-grees of motion at the treated (fused) levels for determiningthe presence or absence of successful bone fusion.16

Brodsky, et al.,3 compared the findings of lumbar flex-ion–extension radiography to surgical exploration in aseries of 175 patients who underwent reoperation for var-ious indications following instrumented and noninstru-mented lumbar fusion. They found a 62% correlation be-tween preoperative flexion–extension radiography andintraoperative findings at exploration (specificity 37%,sensitivity 96%, PPV 70%, and NPV 86%). Their studyprovides Class II medical evidence that the absence ofmotion on flexion–extension x-ray films is highly sugges-tive of a solid fusion. The occurrence of some degree ofmotion at the treated levels, however, does not necessari-ly indicate a pseudarthrosis.

Computerized Tomography Scanning

Since the introduction of CT scanning in the 1970s, thismodality has been used to assess lumbar fusion. Early stud-ies involved axial sequences alone. Brodsky, et al.,3 used 6-mm axial slice CT scans and demonstrated a 57% corre-lation between fusion assessment based on these scans compared with direct surgical exploration in a series of 214operations on 175 patients. Computerized tomographyscanning had a sensitivity of 63%, specificity of 86%, PPVof 72%, and an NPV of 81%. Laasonen and Soini12 con-ducted a retrospective review of 20 patients who underwentCT scanning prior to surgical exploration and found anapproximate 80% correlation between the CT study–baseddiagnosis of fusion and intraoperative diagnosis of fusion.Since the publication of these earlier studies, CT imagingtechnology has advanced. The use of thin-section axial se-quences, improved resolution, and multiplanar imaging ca-pability has enhanced the ability of CT scanning to assesslumbar fusion status. There have been no studies compar-ing these more advanced CT scanning capabilities withdirect surgical exploration. Lang and colleagues14 foundthat the addition of thin-slice and multiplanar CT scanningresulted in a higher rate of detection of pseudarthrosis com-pared with plain radiography. Similarly, Chafetz, et al.,5demonstrated that direct coronal CT scanning may be moresensitive than two-dimensional reconstructed coronal CTimages for the detection of pseudarthrosis. Zinreich and col-leagues21 reported that three-dimensional CT reconstructionmay be more sensitive than two-dimensional CT recon-struction for the detection of pseudarthrosis. Siambanes andMather20 demonstrated that multiplanar CT imaging detect-ed pseudarthrosis in patients who had undergone posteriorlumbar interbody fusion compared with plain radiographythat had suggested a solid fusion. Santos and colleagues18

examined 32 patients who underwent anterior lumbar inter-body fusion with carbon fiber cages. Plain static radiographswere interpreted to demonstrate fusion at 86% of the as-sessed levels. Flexion–extension lumbar radiography sug-gested fusion rates ranging from 74 to 96% in this samegroup of patients, depending on the method used to analyze





Radiographic outcome after fusion

655

TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

rad

iogr

aphi

c as

sess

men

t of

fus

ion*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Blu

men

thal

& G

ill, 1

993

IR

etro

spec

tive

stud

y of

49

patie

nts

inst

rum

ente

d lu

mba

r fu

sion

und

erw

ent e

xplo

ratio

n to

rem

ove

inst

rum

enta

tion.

L

imite

d ac

cura

cy o

f pl

ain

x-ra

y in

ass

essi

ng f

usio

n

AP

& la

t rad

iogr

aphs

com

pare

d w

/ op

find

ings

w/ 6

9% a

gree

men

t. A

ccur

acy

rang

ed a

mon

g th

e ob

serv

ers

from

st

atus

w/ l

ow v

alid

ity (

larg

e in

tra-

& in

tero

bser

ver

57–7

7%. F

alse

–pos

itive

rat

e 42

%; f

alse

–neg

ativ

e ra

te 2

9%.

vari

atio

n).

Boh

nsac

k, e

t al.,

199

9II

Ret

rosp

ectiv

e st

udy

of 4

2 pa

tient

s (4

0 lu

mba

r) o

n ut

ility

of

plan

ar b

one

scin

tigra

phy

(99m

Tc)

to a

sses

s fu

sion

just

B

ased

on

low

sen

sitiv

ity, b

one

scan

not

ade

quat

e to

be

fore

adm

issi

on f

or h

ardw

are

rem

oval

. Bas

ed o

n sc

intig

raph

y da

ta, p

seud

arth

rosi

s w

as s

uspe

cted

in 5

(12

%),

di

agno

se n

onun

ion.

& th

e co

nditi

on w

as c

onfi

rmed

in 4

dur

ing

op (

10%

), 2

dia

gnos

ed &

2 u

ndia

gnos

ed. T

he a

ccur

acy

of th

e m

etho

d w

as 8

8%; s

ensi

tivity

, 50%

; spe

cifi

city

, 93%

; PPV

, 40%

; and

NPV

, 95%

. The

sen

sitiv

ity &

PPV

of b

one

scin

tigra

phy

are

low

for

pos

sibl

e in

stab

ility

aft

er s

pina

l fus

ion.

The

met

hod

is n

ot s

uffi

cien

t to

diag

nose

pse

ud-

arth

rosi

s re

liabl

y af

ter

arth

rode

sis.

Bro

dsky

, et a

l., 1

991

IIR

etro

spec

tive

stud

y of

214

exp

lora

tions

to r

emov

e of

inte

rnal

fix

atio

n de

vice

s, b

atte

ries

, or

for

faile

d-ba

ck s

ur-

Sign

ific

ant i

nacc

urac

y of

pla

in x

-ray

, pol

ytom

ogra

ph,

gery

in 1

75 p

atie

nts

w/ P

LF.

Pla

in x

-ray

s, p

olyt

omog

raph

y, b

endi

ng f

ilms,

&/o

r C

Tsc

ans

corr

elat

ed w

/ sur

-be

ndin

g fi

lms,

& a

xial

CT

in a

sses

sing

fus

ion

gica

l fin

ding

s. S

igni

fica

nt in

accu

racy

fou

nd f

or a

ll m

odal

ities

: pla

in x

-ray

s 36

%, p

olyt

omog

ram

s 41

%, b

endi

ngst

atus

.fi

lms

38%

, axi

al C

T43

% n

onco

rrel

atio

ns. A

xial

CT

had

low

est i

nacc

urac

y (2

2%),

whe

reas

ben

ding

film

s ha

d th

e hi

ghes

t (27

%).

Kan

t, et

al.,

199

5II

Ret

rosp

ectiv

e st

udy

of 7

5 pa

tient

s w

/ ins

trum

ente

d lu

mba

r fu

sion

s. S

ingl

e-bl

inde

d ex

amin

er r

evie

wed

x-r

ays

Lim

ited

accu

racy

of

plai

n x-

rays

.im

med

iate

ly b

efor

e ha

rdw

are

rem

oval

& f

usio

n ex

plor

atio

n: 6

8% c

orre

latio

n bt

wn

radi

ogra

phic

eva

luat

ion

&

intr

aop

obse

rvat

ion.

Sen

sitiv

ity 8

5%, s

peci

fici

ty 6

2%, P

PV76

%, &

NPV

54%

.L

aaso

nen

& S

oini

, 198

9II

Ret

rosp

ectiv

e st

udy

of 4

8 pa

tient

s w

/ per

sist

ent p

ain

afte

r lu

mba

r fu

sion

exa

min

ed u

sing

CT

(6-m

m s

lices

, se-

Mod

erat

e (8

0%)

accu

racy

of

CT

in a

sses

sing

fus

ion.

lect

ive

sagi

ttal r

econ

). 1

57 f

indi

ngs

obse

rved

incl

udin

g: f

ragm

enta

tion

of th

e fu

sion

mas

s (1

6), h

air-

line

pseu

d-ar

thro

sis

(9),

& s

pina

l ste

nose

s (8

). R

eop

in 2

0 pa

tient

s: 2

1 of

27

mai

n le

sion

s de

tect

ed b

y C

Tw

ere

conf

irm

ed;

6 C

Tfi

ndin

gs w

ere

part

ially

or

tota

lly in

corr

ect.

16 (

80%

) of

20

corr

elat

ions

of

CT

& f

usio

n as

sess

men

ts. 2

ca

ses

whe

re C

Tsu

gges

ted

nonu

nion

but

fus

ion

solid

at o

p. 2

cas

es w

here

CT

sugg

este

d un

ion

w/ p

seud

arth

ro-

sis

at o

p.L

arse

n, e

t al.,

199

6II

Pros

pect

ive

stud

y of

25

patie

nts

w/ l

umba

r fu

sion

. All

had

hard

war

e re

mov

al &

fus

ion

insp

ectio

n. S

tudi

es to

rul

e Si

ngle

-obs

erve

r bl

inde

d st

udy

dem

onst

ratin

g no

sig

-ou

t pse

udar

thro

sis

incl

uded

pla

in r

adio

grap

hy, f

lexi

on–e

xten

sion

rad

iogr

aphy

, CT,

& b

one

scin

tigra

phy.

Eac

h ni

fica

nt c

orre

latio

n bt

wn

radi

ogra

phy

& e

xplo

r-

stud

y ev

alua

ted

by b

linde

d ra

diol

ogis

t. A

t exp

lora

tion,

inst

rum

enta

tion

rem

oved

& f

usio

n in

spec

ted.

No

sta-

atio

n.tis

tical

ly s

igni

fica

nt c

orre

latio

n w

as f

ound

bet

wee

n ra

diog

raph

ic &

op

find

ings

.Ja

cobs

on, e

t al.,

199

7II

IU

ltra

eval

uate

d in

10

patie

nts

afte

r po

ster

olat

thor

acic

or

lum

bar

fusi

on w

/in 1

wk

befo

re s

econ

d-lo

ok s

urge

ry.

Cla

ss I

II d

espi

te c

ompa

riso

n w

/ op

beca

use

of la

ck

20 s

ites

eval

uate

d fo

r bo

ne g

raft

, sol

id f

usio

n, c

left

s, f

luid

col

lect

ions

, & h

ardw

are

visi

bilit

y. U

ltra

& o

p fi

nd-

of in

trao

bser

ver

relia

bilit

y da

ta.

ings

com

pare

d. I

n 3

patie

nts,

sta

ndar

d ra

diog

raph

s w

ere

revi

ewed

bef

ore

ultr

a; b

linde

d ul

tra

eval

uatio

n w

as

perf

orm

ed in

the

rem

aini

ng 7

. Ultr

a id

entif

ied

all 1

0 si

tes

of p

seud

arth

rosi

s se

en in

trao

p co

rrec

tly. O

f 10

site

s w

/ sol

id f

usio

n at

sur

gery

, ultr

a de

pict

ed 6

. At 4

site

s (2

pat

ient

s), f

usio

n w

as m

ista

ken

for

or o

bscu

red

by h

ardw

are.

Ove

rall,

sen

sitiv

ity 1

00%

, spe

cifi

city

60%

, & a

ccur

acy

80%

.

*PL

F =

pos

tero

late

ral f

usio

n; r

econ

= r

econ

stru

ctio

n; u

ltra

= u

ltras

onog

raph

y.


the x-ray films. The addition of thin-section helical CTscanning reduced the radiographic fusion rate to 65%. Theauthors concluded that CT scanning is more sensitive thanstatic or flexion–extension lumbar radiography for the de-tection of pseudarthrosis. Shah, et al.,19 reached a similarconclusion in their study of 155 patients who underwentposterior lumbar interbody fusion procedures. They foundthat CT scanning was more sensitive for the detection ofabnormalities than plain radiography. These papers areconsidered to provide Class III medical evidence on theutility of CT scanning for the diagnosis of pseudarthrosisfollowing attempted lumbar fusion.

Technetium-99m Bone Scan

Technetium-99m bone scanning has also been used toassess the integrity of fusion following lumbar arthrodesissurgery. Bohnsack, et al.,2 performed a retrospective studyof 42 patients who underwent lumbar fusion and internalfixation. They obtained 99mTc bone scans before reoperationfor hardware removal. This modality suggested pseudarth-rosis in five patients (12%). Pseudarthrosis was found in-traoperatively in four patients (10%). In two of these fourpatients pseudarthrosis was predicted based on the 99mTcscanning. The accuracy of 99mTc bone scanning was 88%,its sensitivity was 50%, its specificity was 93%, its PPVwas 40%, and its NPV was 95%. This Class II medical evi-dence suggests that 99mTc bone scanning is not sufficientlyreliable to diagnose pseudarthrosis following a lumbar ar-throdesis procedure.2

Roentgen Stereophotogrammetric Analysis

Roentgen stereophotogrammetric analysis is a tech-nique that uses radiopaque 0.8-mm tantalum markers im-planted into each vertebral level incorporated in the fusionat the time of surgery. The details of the technique havebeen described elsewhere.10 Postoperatively, the patientundergoes computerized radiographic assessment inwhich two 40˚ angled roentgen tubes are used. Evaluationis performed with the patient in different positions (forexample, supine and upright) to detect movement. Thetechnique assesses the amount of movement between thefused vertebral bodies in multiple planes. The amount ofallowable movement that determines fusion comparedwith nonunion, however, is not well defined. This modal-ity has been evaluated in patients at several centers. In astudy of 11 patients treated with lumbar fusion, Johnssonand colleagues10 compared the results of RSA with thoseof plain radiography at several postoperative time points.In eight patients in whom plain radiography demonstratedsuccessful fusion, RSA revealed a progressive decrease inintervertebral movement over time with achievement of“rigid fusion” within 3 to 12 months. In a follow-up study,Johnsson, et al.,9 conducted RSA in 12 lumbar fusion patients at multiple postoperative time points. Again,comparative plain radiographs were used and fusion wasconsidered present in all patients. The authors found thatin six patients in whom fusion was considered present ne-gligible movement was observed after 1 month postoper-atively, whereas in others in whom fusion eventuallyoccurred gradual reduction in intervertebral movementwas demonstrated over time. The fact that negligiblemovement was noted so soon after surgery, when fusion

presumably has not yet occurred, is an interesting obser-vation. Pape and associates17 undertook RSA in 10 pa-tients following lumbar arthrodesis. Based on RSA crite-ria, fusion was thought to be present in all patients. Thisfinding was confirmed with open surgical exploration inall cases. Although this report supports the accuracy ofRSA, because fusion was present in all patients it is notpossible to calculate the sensitivity, specificity, PPV, andNPV of RSA compared with exploration from their data.17

Other Techniques

Polytomography has been used to assess lumbar fusionstatus in the pre–CT scanning era, but it has been rarelyused since the widespread introduction of CT scanning inthe 1970s. In their retrospective study of 214 lumbarfusion exploration procedures in patients who had under-gone posterolateral fusion, Brodsky, et al.,3 found only a 59% correlation of fusion status between preoperativepolytomographs and intraoperative findings (sensitivity65%, specificity 84%, PPV 79%, and NPV 73%). Thissingle study provides Class II medical evidence that poly-tomography cannot be reliably used to determine the pres-ence of solid osseous arthrodesis following lumbar fusionprocedures for degenerative disease.

The use of magnetic resonance imaging to assess forpseudarthrosis following lumbar fusion has been exploredby several authors. Lang, et al.,13 maintained that magnet-ic resonance imaging added unique information in casesinvolving lumbar fusion procedures. To date, the impor-tance of this information remains unclear. A single reportof the use of ultrasonography to evaluate fusion status wasalso reviewed.8 Although the results of this study are pro-mising, the ultrasonography technique has not been rigor-ously evaluated.

Summary

The assessment of fusion status with static plain radio-graphy is accurate in approximately two thirds of patientstreated with lumbar fusion when the radiographic resultsare compared with surgical exploration findings. There-fore, static plain radiography is not recommended as astand-alone modality following lumbar fusion procedures.The addition of lateral flexion–extension radiography mayimprove accuracy because the lack of motion betweenfused lumbar segments on lateral views is highly sugges-tive of a solid fusion. Some degree of motion betweensegments may be present even when the spine has fused.The amount of motion allowable across fused segments isnot clear, and the role of internal fixation in limitingmotion has also not been adequately addressed. The addi-tion of multiplanar CT scanning results in the detection ofpseudarthrosis in some patients in whom fusion has beendeemed successful based on plain radiographic criteria.Therefore, CT scanning may be more accurate in thedetermination of fusion status than plain radiography;however, a rigorous comparison of modern CT scanningand surgical exploration has not been performed. It ap-pears that RSA is exquisitely sensitive for the detection ofmotion between vertebral bodies, and the loss of motionbetween treated vertebral segments does appear to indi-cate the presence of fusion. The modality, however, is in-vasive and not widely available. Furthermore, the only




comparison of RSA with surgical exploration providedonly Class III medical evidence supporting the accuracyof RSA. It is recommended that multiple modalities beused for the noninvasive evaluation of symptomatic pa-tients with suspected fusion failure because no radio-graphic gold standard exists.

Key Issues for Further Investigation

Modern CT scanning appears to have superior sensitiv-ity compared with plain radiography for the detection ofpseudarthrosis. A prospective study of CT scanning find-ings prior to surgical exploration for instrumentation re-moval would provide Class I evidence regarding the ac-curacy of the former compared with the gold standard ofsurgical exploration. If preoperative flexion–extension radiography is also used, then the influence of internal fix-ation on the accuracy of flexion–extension radiographycould also be addressed.

References

1. Blumenthal SL, Gill K: Can lumbar spine radiographs accu-rately determine fusion in postoperative patients? Correlationof routine radiographs with a second surgical look at lumbarfusions. Spine 18:1186–1189, 1993

2. Bohnsack M, Gosse F, Ruhmann O, et al: The value of scintig-raphy in the diagnosis of pseudarthrosis after spinal fusionsurgery. J Spinal Disord 12:482–484, 1999

3. Brodsky AE, Kovalsky ES, Khalil MA: Correlation of radio-logic assessment of lumbar spine fusions with surgical explo-ration. Spine 16:S261–S265, 1991

4. Burkus JK, Dorchak JD, Sanders DL: Radiographic assessmentof interbody fusion using recombinant human bone morpho-genetic protein type 2. Spine 28:372–377, 2003

5. Chafetz N, Cann CE, Morris JM, et al: Pseudarthrosis followinglumbar fusion: detection by direct coronal CT scanning. Ra-diology 162:803–805, 1987

6. Cleveland M, Bosworth D, Thompson F: Pseudoarthrosis in thelumbosacral spine. J Bone Joint Surg Am 30:302–312, 1948

7. Hilibrand AS, Dina T: The use of diagnostic imaging to assessspinal arthrodesis. Orthop Clin North Am 29:591–601, 1998

8. Jacobson JA, Starok M, Pathria MN, et al: Pseudarthrosis: USevaluation after posterolateral spinal fusion: work in progress.Radiology 204:853–858, 1997

9. Johnsson R, Axelsson P, Gunnarsson G, et al: Stability of lum-bar fusion with transpedicular fixation determined by roentgenstereophotogammetric analysis. Spine 24:687–690, 1999

10. Johnsson R, Selvik G, Stromqvist B, et al: Mobility of the lowerlumbar spine after posterolateral fusion determined by roentgenstereophotogrammetric analysis. Spine 15:347–350, 1990

11. Kant AP, Daum WJ, Dean SM: Evaluation of lumbar spinefusion. Plain radiographs versus direct surgical exploration andobservation. Spine 20:2313–2317, 1996

12. Laasonen E, Soini J: Low-back pain after lumbar fusion. Surgicaland computed tomographic analysis. Spine 14:210–213, 1989

13. Lang P, Chafetz N, Genant HK, et al: Lumbar spinal fusion.Assessment of functional stability with magnetic resonance im-aging. Spine 15:581–588, 1990

14. Lang P, Genant HK, Chafetz N, et al: Three-dimensional com-puted tomography and multiplanar reformations in the assess-ment of pseudarthrosis in posterior lumbar fusion patients.Spine 13:69–75, 1988

15. Larsen JM, Rimoldi RL, Capen DA, et al: Assessment of pseud-arthrosis in pedicle screw fusion: a prospective study compar-ing plain radiographs, flexion/extension radiographs, CT scan-ning, and bone scintigraphy with operative findings. J SpinalDisord 9:117–120, 1996

16. Mcafee PC, Boden SD, Brantigan JW, et al: Symposium: A crit-ical discrepancy—a criteria of successful arthrodesis followinginterbody spinal fusions. Spine 26:320–334, 2001 (Erratum inSpine 26:1103, 2001)

17. Pape D, Fritsch E, Kelm J, et al: Lumbosacral stability afterconsolidated anteroposterior fusion after instrumentation re-moval determined by roentgen stereophotogrammetric analysisand direct surgical exploration. Spine 27:269–274, 2002

18. Santos ER, Goss DG, Morcom RK, et al: Radiologic assess-ment of interbody fusion using carbon fiber cages. Spine 28:997–1001, 2003

19. Shah RR, Mohammed S, Saifuddin A, et al: Comparison ofplain radiographs with CT scan to evaluate interbody fusion fol-lowing the use of titanium interbody cages and transpedicularinstrumentation. Eur Spine J 12:378–385, 2003

20. Siambanes D, Mather S: Comparison of plain radiographs andCT scans in instrumented posterior lumbar interbody fusion.Orthopedics 21:165–167, 1998

21. Zinreich SF, Long DM, Davis R, et al: Three-dimensional CTimaging in postsurgical “failed back” syndrome. J Comput As-sist Tomogr 14:574–580, 1990




Radiographic outcome after fusion

657


Recommendations

Standards. There is insufficient evidence to recommenda treatment standard.

Guidelines. There is insufficient evidence to recom-mend a treatment guideline.

Options. It is recommended that when performing lum-bar arthrodesis for degenerative lumbar disease, strategiesto achieve successful radiographic fusion should be con-sidered. There appears to be a correlation between suc-cessful fusion and improved clinical outcomes; however,it should be noted that the correlation between fusion sta-tus and clinical outcome is not strong, and in a givenpatient fusion status may be unrelated to clinical outcome.

Rationale

Achieving a solid arthrodesis following a spinal fusionprocedure is generally believed to be an important goal;however, the relationship between successful fusion andclinical outcome has not been fully established. Therefore,the utility of exhaustive radiographic testing to determinefusion status may be questioned. The purpose of this re-view is to examine the literature regarding the relationship

between fusion status and clinical outcome after lumbararthrodesis procedures performed in the treatment of lum-bar spinal degenerative disease.

Search Criteria

A computerized search of the database of the NationalLibrary of Medicine of articles published between 1966and July 2003 was conducted using the search terms “lum-bar spine fusion assessment” or “lumbar spine pseud-arthrosis,” or “lumbar spine fusion outcome.” The searchwas restricted to references in the English language in-volving humans. This yielded a total of 1076 references.The titles and abstracts of each of these references werereviewed. Papers not concerned with the assessment ofpostoperative fusion status or those not focused on adultdegenerative lumbar disease (for example, trauma-relatedfractures, infection, scoliosis, and isthmic spondylolisthe-sis) were discarded. Additional articles were obtained fromthe bibliographies of the selected articles. Thirty-seven ref-erences were identified that provided either direct or sup-porting evidence relevant to the clinical utility of the radi-ographic assessment of lumbar fusion status. These papersand references were obtained and reviewed. Papers pro-viding Class III or better medical evidence regarding therelationship between fusion status and clinical outcomefollowing lumbar arthrodesis procedures for degenerativedisease are listed in Table 1. Supportive data are providedby additional references as listed in the Reference section.


658

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 5: correlation between radiographic and functional outcome

DANIEL K. RESNICK, M.D., TANVIR F. CHOUDHRI, M.D., ANDREW T. DAILEY, M.D., MICHAEL W. GROFF, M.D., LARRY KHOO, M.D., PAUL G. MATZ, M.D., PRAVEEN MUMMANENI, M.D., WILLIAM C. WATTERS III, M.D., JEFFREY WANG, M.D., BEVERLY C. WALTERS, M.D., M.P.H., AND MARK N. HADLEY, M.D.

Department of Neurosurgery, University of Wisconsin, Madison, Wisconsin; Department ofNeurosurgery, Mount Sinai Medical School, New York, New York; Department of Neurosurgery,University of Washington, Seattle, Washington; Department of Neurosurgery, Indiana University,Indianapolis, Indiana; Departments of Orthopedic Surgery and Neurosurgery, University of California at Los Angeles, California; Department of Neurosurgery, University of Alabama at Birmingham,Alabama; Department of Neurosurgery, Emory University, Atlanta, Georgia; Bone and Joint Clinic of Houston, Texas; and Department of Neurosurgery, Brown University, Providence, Rhode Island

KEY WORDS • fusion • lumbar spine • treatment outcome • radiography •practice guidelines


Abbreviations used in this paper: ALIF = anterior lumbar inter-body fusion; AP = anteroposterior; DPQ = Dallas Pain Ques-tionnaire; LBOS = Low Back Pain Outcome; LBPS = Low-BackPain Scale; PLF = posterolateral fusion; VAS = visual analog scale.



One of the integral goals of any lumbar fusion procedureperformed to treat low-back pain due to lumbar degenera-tive disease is the achievement of a solid fusion across thetreated motion segments. Intuitively, one would expect thatpatients who achieve a solid fusion would tend to have bet-ter clinical outcomes compared with those in whom pseud-arthrosis develops; however, several authors have des-cribed patients with pseudarthrosis with favorable clinicaloutcomes and patients with solid osseous unions who havepoor clinical outcomes.3,5 The radiographic assessment oflumbar fusion status is imperfect, consumes healthcare re-sources, and exposes the patient to ionizing radiation. If theclinical results associated with lumbar fusion procedures donot correlate with radiographic findings, one must questionthe utility of exhaustive radiographic study to demonstratefusion. Furthermore, the incorporation of surgical tech-niques and adjuncts designed to increase radiographic fu-sion rates may be inappropriate unless a correlation betweenradiographic and clinical outcomes can be confirmed. Thepurpose of this document is to review the evidence for andagainst such a relationship.

A study correlating clinical outcomes with the results ofopen surgical assessment of lumbar fusion status has notbeen performed; however, studies do exist in which inves-tigators compared various radiographic fusion assessmenttechniques with clinical outcomes. Several studies haveshown correlation between clinical and radiographic out-comes after lumbar fusion. Christensen, et al.,2 studied120 consecutive patients who underwent ALIF. Clinicaloutcome was evaluated 5 to 13 years after surgery byusing the DPQ. At 2 years postoperatively, fusion out-come was assessed using static plain radiography assessedby independent observers. These authors reported com-plete fusion in 52% of patients, questionable fusion in24%, and definitive pseudarthrosis in 24%. Patients withcomplete or questionable union had significantly betterDPQ scores than those with nonunion (p , 0.01). Theauthors concluded that DPQ scores correlated well withradiological outcome. This study is considered to provideClass III medical evidence supporting fusion status as apredictor of functional outcome because the radiographicand clinical follow-up evaluations were obtained at wide-ly separated time points (between 3–11 years apart) andbecause the study relied on static plain radiography to de-termine fusion status.

In 2002, the same group published a prospective ran-domized 2-year follow-up study of 148 patients random-ized to PLF plus pedicle screw fixation or ALIF, PLF, andpedicle screw fixation.1 Clinical outcome was assessedusing the DPQ, the LBPR Scale, and a work status survey.The authors found that patients in both treatment groupsexhibited highly significant improvements in all four cat-egories of quality of life (DPQ) as well as in the back painand leg pain index (LBPR) compared with their preopera-tive status. They identified a significant relationship between fusion status and functional outcome. Patientsjudged to have solid fusions did significantly better thanthose without solid fusions on three of four subsections ofthe DPQ (there was also a nonsignificant improvement onthe social concerns subsection). Because of the reliance onstatic plain radiography to determine fusion status, this

study is thought to provide Class III medical evidencesupporting the importance of fusion status as a predictorof functional outcome.

Vamvanij, et al.,8 reported on 56 consecutive patientstreated with one of four different lumbar fusion proce-dures. Clinical outcome was assessed using a postopera-tive pain survey and an independent clinical assessment.The radiographic outcome was assessed using AP and lat-eral static plain radiography and flexion–extension radi-ography in selected cases in which the static x-ray filmswere thought to be equivocal for fusion success. Patientsin whom successful lumbar fusion was achieved experi-enced better clinical outcomes and had a better chance ofreturning to work. The authors concluded that there was apositive correlation between solid fusion and successfulclinical outcome. This study provides Class III medicalevidence in support of the correlation between radio-graphic and clinical outcome.

Wetzel and colleagues9 prospectively evaluated 74 consecutive patients who underwent lumbar fusion. Out-comes were measured using subjective clinical outcomescores pertaining to pain relief and medication usage. Thepatients were observed at five intervals after surgery dur-ing a minimum 2-year follow-up period (range 24–35months, mean 27 months). Fusion status was evaluatedusing lateral flexion–extension radiography in all cases,with the selective use of other techniques. The authorsnoted a 61% fusion rate. At final follow-up examination,60% of patients had improved back pain and 70% hadimproved leg pain. The presence of radiographic fusioncorrelated positively with a successful clinical outcome (r = 3.3, p = 0.010). In a prospective study of 124 lumbarfusion patients assigned to three different surgical treat-ment groups, Zdeblick10 assessed fusion status by per-forming static and flexion–extension lateral radiographyat 1 year; the clinical outcomes were rated as excellent,good, fair, or poor. They found that patients in the groupswith higher fusion rates had better clinical outcomes.These studies, although prospective (and randomized inthe case of the study by Zdeblick) are considered to pro-vide Class III medical evidence in support of the correla-tion between radiographic and clinical outcome becauseof the use of nonvalidated clinical outcome measures.9

A number of other studies reporting similar trends havefailed to demonstrate a statistically significant correlationbetween clinical and radiographic outcome in patients fol-lowing lumbar arthrodesis surgery. For example, in a ret-rospective review of 83 patients who underwent posteriorlumbar interbody fusion, Diedrich, et al.,4 obtained fol-low-up data in 64 patients. Clinical outcome was assessedusing a Hambly score based on pain intensity, medicationuse, and patient activity. Their results were stratified intoa four-point scale (excellent, good, fair, poor). Assessmentof fusion was performed using AP and lateral plain radi-ography. Using standard radiographic criteria for fusion,the authors identified fusion rates of 51.5% at 12 months,61.4% at 24, 66.7% at 36, and 77.8% at 48 months post-operatively. A comparison of the radiographic fusion ratesand clinical outcomes revealed that in 64% of patientswith excellent or good outcomes radiographic fusion wasachieved. In patients with fair or poor outcomes the rate ofsuccessful fusion was 58% (p value not significant). Theauthors concluded there was slight nonsignificant correla-


Radiographic and functional outcome

659




TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

the

rel

atio

ship

of

clin

ical

and

rad

iogr

aphi

c ou

tcom

es*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Zde

blic

k, 1

993

III

124

lum

bar

fusi

on p

atie

nts

wer

e pr

ospe

ctiv

ely

stud

ied.

Fus

ion

stat

us w

as d

eter

min

ed u

sing

AP

& f

lex–

ext r

adi-

Gro

ups

w/ h

ighe

r fu

sion

rat

es d

id b

ette

r.og

raph

y at

1 y

r. C

linic

al r

esul

ts w

ere

rate

d as

exc

elle

nt, g

ood,

fai

r, or

poo

r.C

hris

tens

en, e

t al.,

199

6II

I12

0 co

nsec

utiv

e pa

tient

s, w

/ clin

ical

out

com

e ev

alua

ted

5–13

yrs

pos

top

usin

g D

PQ. A

t 2 y

rs p

osto

p, r

adio

logi

cal

DPQ

sco

res

corr

elat

ed w

ell w

/ rad

iolo

gica

l out

com

e.ou

tcom

e w

as d

eter

min

ed b

y in

depe

nden

t obs

erve

rs: 5

2% c

ompl

ete

fusi

on, 2

4% q

uest

iona

ble

fusi

on, &

24%

de

fini

tive

PA. P

atie

nts

w/ c

ompl

ete

or q

uest

iona

ble

unio

n ha

d si

gnif

ican

tly b

ette

r re

sults

than

thos

e w

/ non

-un

ion

(p ,

0.01

).Pe

nta

& F

rase

r, 19

97II

I10

3 A

LIF

pat

ient

s (f

rom

a c

onse

cutiv

e se

ries

of

125)

had

clin

ical

(L

BO

S, V

AS,

MSP

Q, Z

DS)

out

com

e as

sess

men

tL

ong-

term

(.

10-y

r) p

rese

nce

of r

adio

logi

cal f

usio

n &

87

patie

nts

also

had

rad

iogr

aphi

c fu

sion

ass

essm

ent (

AP/

lat r

adio

grap

hs w

/ or

w/o

MR

I) .

10 y

rs p

osto

p.

was

not

ass

ocia

ted

w/ t

he c

linic

al o

utco

me.

78%

rat

ed th

emse

lves

as

havi

ng “

com

plet

e re

lief”

or

“a g

ood

deal

of

relie

f,”

but o

nly

34%

had

exc

elle

nt o

r go

od L

BO

S. C

linic

al o

utco

me

was

not

ass

ocia

ted

w/ t

he p

rese

nce

of r

adio

logi

cal f

usio

n &

was

not

infl

uenc

ed

by th

e co

mpe

nsat

ion

stat

us. P

sych

olog

ical

dis

turb

ance

at r

evie

w a

nd r

eop,

how

ever

, wer

e si

gnif

ican

tly c

or-

rela

ted

w/ L

BO

S. C

oncl

usio

ns: A

LIF

out

com

e w

as s

tron

gly

affe

cted

by

psyc

holo

gica

l mak

eup

of p

atie

nt; h

ow-

ever

, the

neg

ativ

e ef

fect

of

com

pens

atio

n ob

serv

ed a

t 2 y

rs s

eem

s to

dis

sipa

te w

/ tim

e &

bec

omes

insi

gnif

ican

t at

10

yrs.

Gre

enou

gh, e

t al.,

199

8II

IPr

ospe

ctiv

e ca

se s

erie

s of

135

pat

ient

s w

ho u

nder

wen

t ins

trum

ente

d PL

F. C

linic

al o

utco

me

was

ass

esse

d us

ing

No

diff

eren

ce in

out

com

e w

as d

emon

stra

ted

btw

n

LB

OS,

rad

iolo

gica

l out

com

e by

AP/

lat p

lain

rad

iogr

aphy

. Fus

ion

rate

was

82%

& c

linic

al im

prov

emen

t rat

epa

tient

s w

/ fus

ion

& th

ose

w/o

reg

ardi

ng L

BO

S (s

elf-

asse

ssm

ent)

onl

y 19

% g

ood

or e

xcel

lent

(L

BO

S). C

ompe

nsat

ion

stat

us &

psy

chol

ogic

al d

istr

ess

wer

e sc

ores

, pat

ient

sat

isfa

ctio

n sc

ore,

or

pain

VA

S.si

gnif

ican

tly a

ssoc

iate

d w

/ out

com

e.V

amva

nij,

et a

l., 1

998

III

56 c

onse

cutiv

e pa

tient

s un

derw

ent 1

of

4 lu

mba

r fu

sion

pro

cedu

res.

Clin

ical

out

com

e w

as a

sses

sed

by p

osto

p pa

inC

orre

latio

n bt

wn

solid

fus

ion

& s

ucce

ssfu

l clin

ical

su

rvey

& in

depe

nden

t clin

ical

exa

m. R

adio

grap

hic

outc

ome

was

ass

esse

d w

/ AP/

lat p

lain

rad

iogr

aphy

& w

/ ou

tcom

e.fl

ex–e

xt v

iew

s in

sel

ecte

d ca

ses

whe

re th

e fo

rmer

was

equ

ivoc

al. C

ases

w/ s

ucce

ssfu

l lum

bar

fusi

on h

ad

bette

r cl

inic

al o

utco

mes

& a

bet

ter

chan

ce o

f w

ork

resu

mpt

ion.

Wet

zel,

et a

l., 1

999

III

74 c

onse

cutiv

e ca

ses

of lu

mba

r fu

sion

. Sta

ndar

d ou

tcom

e sc

ores

on

pain

rel

ief

& m

edic

atio

n us

age

wer

e us

ed.

The

pre

senc

e of

a s

olid

fus

ion

(r =

3.3

, p =

0.0

10)

Patie

nts

wer

e ob

serv

ed p

osto

p at

5 in

terv

als

for

$2

yrs

(ran

ge 2

4–35

mos

, mea

n 27

mos

). F

usio

n st

atus

was

co

rrel

ated

pos

itive

ly w

/ suc

cess

ful c

linic

al o

ut-

base

d on

fle

x–ex

t rad

iogr

aphs

in a

ll ca

ses,

w/ s

elec

tive

use

of o

ther

tech

niqu

es. O

vera

ll fu

sion

rat

e w

as 6

1%.

com

e. T

he p

rese

nce

of P

Aw

as n

egat

ivel

y co

rrel

at-

At f

inal

FU

, 60%

had

impr

oved

bac

k pa

in &

70%

had

impr

oved

leg

pain

. Fus

ion

(r =

3.3

, p =

0.0

10)

corr

elat

ed

ed w

/ a s

ucce

ssfu

l clin

ical

out

com

e.po

sitiv

ely

w/ a

suc

cess

ful c

linic

al o

utco

me;

the

pres

ence

of

PAne

gativ

ely

corr

elat

ed w

/ a s

ucce

ssfu

l clin

ical

ou

tcom

e.D

iedr

ich,

et a

l., 2

001

III

64 p

atie

nts

follo

win

g PL

IF w

ere

eval

uate

d w

/ pla

in r

adio

grap

hy: f

usio

n ra

tes

of 5

1.5%

at 1

2, 6

1.4%

at 2

4, 6

6.7%

N

o si

gnif

ican

t cor

rela

tion

btw

n fu

sion

rat

e &

clin

ical

at

36,

& 7

7.8%

at 4

8 m

os p

osto

p. C

linic

al s

tatu

s w

as e

valu

ated

by

Ham

bly

scor

e (p

ain,

med

icat

ion

use,

&

outc

ome.

activ

ity).

The

clin

ical

out

com

e w

as s

trat

ifie

d in

to 4

-poi

nt s

cale

. Fus

ion

rate

was

64%

in p

atie

nts

w/ e

xcel

lent

or

goo

d ou

tcom

e &

58%

in th

ose

w/ f

air

or p

oor

outc

ome

(NS)

. The

re w

as s

light

NS

corr

elat

ion

btw

n ra

dio-

grap

hic

fusi

on &

pat

ient

-ass

esse

d cl

inic

al o

utco

me.

Chr

istia

nsen

, et a

l., 2

002

III

Pros

pect

ive

2-yr

FU

of

148

patie

nts

rand

omiz

ed to

PL

F 1

PS o

r AL

IF 1

PLF

1PS

. Clin

ical

out

com

e w

as a

s-C

orre

latio

n w

as f

ound

btw

n fu

sion

sta

tus

& f

unc-

sess

ed w

/ DPQ

, LB

PR, &

wor

k st

atus

sur

vey

scal

es. B

oth

grou

ps s

how

ed h

ighl

y si

gnif

ican

t im

prov

emen

t in

tiona

l out

com

e.al

l 4 c

ateg

orie

s of

life

qua

lity

(DPQ

), b

ack

pain

, & le

g pa

in in

dex

(LB

PR)

com

pare

d w

/ pre

op s

tatu

s. T

he c

ir-

cum

fere

ntia

l fus

ion

patie

nts

show

ed a

hig

her

PLF

rate

(92

%)

than

the

PLF

grou

p (8

0%)

(p ,

0.04

). C

ircu

m-

fere

ntia

l lum

bar

fusi

on p

rodu

ced

a hi

gher

fus

ion

rate

w/ t

ende

ncy

tow

ard

bette

r fu

nctio

nal o

utco

me.

*fl

ex–e

xt =

fle

xion

–ext

ensi

on;

FU =

fol

low

up;

MR

I =

mag

netic

res

onan

ce i

mag

ing;

MSP

Q =

Mul

tiple

Som

atic

Per

cept

ion

Que

stio

nnai

re;

NS

= n

ot s

igni

fica

nt;

PA=

pse

udar

thro

sis;

PL

IF =

pos

teri

orlu

mba

r in

terb

ody

fusi

on; P

S =

ped

icle

scr

ew; Z

DS

= Z

ung

Dep

ress

ion

Scal

e.


tion between radiographic fusion and patient-assessed clinical outcome. There were several flaws in this studyrelated to choice of the radiographic assessment tool, theclinical outcome tool, and patient dropout.

In several studies other authors have failed to demon-strate any correlation at all between radiographic fusionstatus and clinical outcome. In a prospective case series of135 patients who underwent instrumentation–augmentedPLF, Greenough and associates6 assessed clinical outcomewith the LBOS and radiographic outcome with static plainradiography. They found an overall fusion rate of 82% butonly a 65% clinical improvement rate based on patientself-assessment. Based on the LBOS, only 19% of the pa-tients made significant improvements. The authors did notidentify a difference between patients with fusion and thosewithout regarding LBOS scores, patient satisfaction scores,or the VAS. This study is considered to provide Class IIImedical evidence refuting the importance of fusion as a de-terminant of outcome based on the study’s use of plain radi-ography alone for the assessment of fusion status. The discrepancy between the LBOS scores and the patient satis-faction ratings illustrates the problems associated with theuse of nonvalidated outcomes measures.

In a long-term outcome study (. 10 years), Penta andFraser7 reported on 103 patients who underwent ALIF(from a consecutive series of 125). Clinical outcome as-sessment involved various validated outcome measuresincluding the LBOS. Eighty-seven patients also under-went fusion assessment with AP and lateral radiography.The authors reported that 78% of patients rated them-selves as having “complete relief” or “a good deal of re-lief,” but only 34% had excellent or good LBOS scores.Their clinical outcomes could not be correlated with thepresence of radiographic fusion. This study also providesClass III medical evidence against a correlation betweenradiographic fusion status and clinical outcome followinglumbar fusion surgery.

Summary

At the present time, there is no Class I or II medical evi-dence to indicate that the appearance of a radiographic fu-sion is significantly associated with improved functionaloutcomes. The majority of Class III medical evidencedoes suggest that successful radiographic fusion is associ-ated with improved clinical outcomes. It is widely ac-knowledged that this relationship is not perfect.

Key Directions for Future Research

A case control study involving categorization of pa-tients based on multiple validated outcome instrument–derived outcome and multimodal radiographic outcomeassessment would provide Class II medical evidence sup-porting or refuting the importance of radiographic fusion.

References


2. Christensen FB, Karlsmose B, Hansen ES, et al: Radiologicaland functional outcome after anterior lumbar interbody spinalfusion. Eur Spine J 5:293–298, 1996

3. DePalma AF, Rothman RH: The nature of pseudarthrosis. ClinOrthop Relat Res 59:113–118, 1968

4. Diedrich O, Perlick L, Schmitt O, et al: Radiographic character-istics on conventional radiographs after posterior lumbar inter-body fusion: comparative study between radiotranslucent andradiopaque cages. J Spinal Disord 14:522–532, 2001

5. Flatley TJ, Derderian H: Closed loop instrumentation of thelumbar spine. Clin Orthop Relat Res 196:273–278, 1985


7. Penta M, Fraser RD: Anterior lumbar interbody fusion. A min-imum 10-year follow-up. Spine 22:2429–2434, 1997

8. Vamvanij V, Fredrickson BE, Thorpe JM, et al: Surgical treat-ment of internal disc disruption: an outcome study of four fu-sion techniques. J Spinal Disord 11:375–382, 1998

9. Wetzel FT, Brustein M, Phillips FM, et al: Hardware failure inan unconstrained lumbar pedicle screw system. A 2-year fol-low-up study. Spine 24:1138–1143, 1999

10. Zdeblick TA: A prospective, randomized study of lumbar fu-sion. Preliminary results. Spine 18:983–991, 1993




Radiographic and functional outcome

661


Recommendations


Guidelines. 1) It is recommended that MR imaging beused as a diagnostic test instead of discography for the ini-tial evaluation of patients with chronic low-back pain. 2) It is recommended that MR imaging–documented discspaces that appear to be normal not be considered for treat-ment as a source of low-back pain. 3) It is recommendedthat lumbar discography not be used as a stand-alone teston which treatment decisions are based for patients withlow-back pain. 4) If discography is performed as a diag-nostic tool to identify the source of a patient’s low-backpain, it is recommended that both a concordant pain re-sponse and morphological abnormalities be present at thepathological level prior to initiating any treatment directedat that level.

Options. 1) It is recommended that discography be re-served for use in patients with equivocal MR imagingfindings, especially at levels adjacent to clearly patholog-ical levels. 2) It is recommended that patients in whom

discography is positive but in whom MR imaging evi-dence of disc degeneration is absent not be consideredcandidates for operative intervention.

Rationale

The successful surgical treatment of patients with low-back pain depends on an accurate diagnosis of the source ofpain. In the absence of gross deformity or neural compres-sion, the diagnosis of “discogenic” low-back pain may beestablished using diagnostic imaging and functional stud-ies. Discography has been used as a diagnostic tool for theevaluation of patients with low-back pain with normal spi-nal alignment and without evidence of neural compression.The purpose of this review is to examine the medical evi-dence in the literature regarding discography as a diagnos-tic test for the localization of the source of low-back pain inthese patients.

Literature Search

The database of the National Library of Medicine wassearched for articles published between 1966 and Novem-ber 2003. Use of the search terms “discography or disco-gram” limited to “human” and “English language” result-ed in 304 matches. The titles and abstracts of these 304abstracts were reviewed and duplicates, technical notes,reviews, and other papers that did not describe the use of


662

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 6: magnetic resonance imaging and discography for patient selection for lumbar fusion



KEY WORDS • lumbar spine • fusion • magnetic resonance imaging • discography •practice guidelines


Abbreviations used in this paper: CT = computerized tomogra-phy; HIZ = high-intensity zone; MR = magnetic resonance; NPV =negative predictive value; PLF = posterolateral fusion; PPV = pos-itive PV.


discography for the diagnosis and management of patientswith low-back pain were discarded. The reference lists ofthe remaining articles were inspected and several morerelevant papers were identified. References consisting ofclinical series of patients managed with discography wereidentified and are briefly described in Table 1. A numberof other references served as background information andare included in the bibliography.


Discography has been used for decades for the diagnosisof lumbar intervertebral disc abnormalities in patients withlow-back pain.17,32 Currently, discography is the only diag-nostic test that has a physiological end point used in theassessment of such patients (that is, the reproduction ofconcordant low-back pain).46,60 Proponents of discographyargue that the technique is more sensitive for the diagnosisof anatomical disc abnormalities and injuries than plainradiography, myelography, or MR imaging.5–7,8–10 Critics ofdiscography claim that the test is not specific because mor-phological abnormalities do not always correlate with clin-ical complaints and because intradiscography pain provo-cation occurs in patients with lumbar pain caused bynonspinal entities.11 Indeed, several studies have demon-strated that severe low-back pain may be elicited by discog-raphy in individuals with no prior complaints of low-backpain.11,13,24 Controversy also exists as to whether discogra-phy adds any diagnostic information to the data providedby MR imaging, a sensitive and specific noninvasive testfor lumbar disease.3,6,9,16,20,21,25–27,33,34,36,41,43,45,47,49–51,57

Several studies have examined the sensitivity, specifici-ty, and predictive value of MR imaging compared with themorphological findings on discography. In a large series ofpatients Bernard2 compared MR imaging and discographyand reported that the PPV of an abnormal MR image for amorphologically abnormal discogram was 92%. The NPVof a normal MR image in the same series was 88%. UsingT2-weighted MR imaging and discography to treat 101 disclevels, Schneiderman and colleagues51 reported that MRimaging was 99% accurate in predicting abnormal mor-phological findings on discography. One group reportedcomplete agreement between abnormal MR imaging find-ings and stage-two or stage-three disc disruption identifiedon CT discography (Dallas Pain Questionnaire).37,46 Sep-arate small studies by Lonergan, et al.,34 and Gibson, et al.,20

noted an approximate 90% concordance rate between ab-normalities identified on MR imaging and discography.Although discography may, on occasion, identify abnor-malities in patients with normal MR imaging findings, thesignificance of these findings is unclear. Current evidenceindicates that MR imaging is a very good imaging tool forthe determination of abnormal disc morphology and that itavoids the expense and invasiveness of discography.19,29,52

For these reasons, lumbar MR imaging is recommended asthe neuroimaging study of choice for the evaluation ofpatients with low-back pain.

The clinical significance of MR imaging– or discog-raphy-identified morphological abnormalities of an in-tervertebral disc has been questioned. Both modalities are sensitive to disc abnormalities. The frequency of discabnormalities identified by discography is quite high inpatients with low-back pain. Grubb, et al.,22 reported that

78% of patients undergoing discography assessment forlow-back pain had morphologically abnormal discs at oneor more levels despite normal plain spine radiography andmyelography. Similarly, Schwarzer and colleagues53 de-scribed abnormal discographic findings in 39% of 92 pa-tients evaluated for low-back pain. Park, et al.,42 also notedabnormal discographic findings in patients whose radio-logical evaluation for low-back pain was otherwise unre-markable. Morphologically abnormal discograms, howev-er, have also been observed in 17 to 37% of asymptomaticpatients.13,24,60

In an attempt to improve the diagnostic utility of dis-cography, Walsh and associates60 required that discogra-phy result in the production of pain identical or very sim-ilar to the patient’s usual pain complaints to be considered“positive.” The authors also required that this pain res-ponse occur in association with demonstrable morpholog-ical abnormalities of the disc space in question. The sever-ity of the patient’s pain, as determined using a visual analogscale as well as observation of patient behavior, must alsobe severe (three of five, or six of 10 on the visual analogscale).14,60 The authors’ description of a “positive” disc-ogram has been adopted by most investigators and authorsas a “concordant” discogram.

Several comparisons between disc morphology and con-cordant pain provocation during discography have beenperformed. These studies have revealed a discrepancybetween morphological disc abnormalities and pain per-ception during discography. Antti-Poika and colleagues1 re-ported that 13% of patients they reviewed reported pain oninjection of morphologically normal discs. Millette andMelanson38 reported that only 37% of patients with abnor-mal disc morphology experienced concordant pain withinjection. Five percent of patients reported pain despite thepresence of normal morphology.40 Sachs, et al.,46 reported a13% incidence of abnormal disc morphology identified by discography in which concordant pain provocation wasabsent in their large series. Saifudin, et al.,48 found that onlyanular tears could be reliably associated with the provoca-tion of pain during discography and that other degenerationpatterns were not necessarily associated with a pain re-sponse during injection. These studies indicate that discmorphology, as assessed by discography, does not ade-quately predict the provocation of symptomatic low-backpain. Therefore, the presence of abnormal discography-documented morphology in the absence of a concordantpain response should not be used to justify intervention atthat disc level.

Abnormal disc morphology identified on MR imaging,including loss of T2 signal intensity, disc space collapse,modic changes, and HIZs, are commonly observed in pa-tients evaluated for low-back pain.9,43 As with discogra-phy, these disc space abnormalities are also seen frequent-ly on imaging studies obtained in asymptomatic patients.12

The correlation of MR imaging abnormalities and painprovocation during discography has been examined inseveral series. Linson and Crowe33 performed a prospec-tive comparison of T2-weighted MR imaging and discog-raphy findings. They found a likelihood ratio of 30 for anabnormal MR image and concordant pain provocationduring discography. In another study, Braithwaite and col-leagues6 reported that modic changes on MR images werea specific, but not necessarily sensitive, predictor of con-


Magnetic resonance imaging and discography

663




TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

MR

im

agin

g an

d di

scog

raph

y fo

r se

lect

ion

of l

umba

r su

rger

y*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

oncl

usio

ns

Hol

t, 19

68II

ID

isco

was

per

form

ed o

n 30

pat

ient

s w

/o h

isto

ry o

f L

BP.

37%

rep

orte

d on

set o

f ba

ck (

no

Dis

co u

nrel

iabl

e fo

r di

agno

sis

of d

isco

geni

c le

akag

e bu

t irr

egul

arity

of

disc

spa

ce n

oted

) or

bac

k &

leg

(lea

kage

of

dye)

pai

n w

/ in-

back

pai

n du

e to

hig

h fa

lse

posi

tive

rate

.je

ctio

n.Si

mm

ons

& S

egil,

197

5II

IPa

infu

l dis

c in

ject

ion

was

use

d as

gui

de f

or o

p (e

ither

sim

ple

disc

ecto

my

or d

isce

ctom

y &

D

isco

can

hel

p to

loca

lize

path

olog

ical

leve

ls

PLF)

in la

rge

pre-

MR

I &

CT

seri

es. M

ost p

atie

nts

did

wel

l.in

pat

ient

s w

/ bac

k pa

in &

rad

icul

opat

hy.

Bro

dsky

& B

inde

r, 19

79II

ID

isco

gram

was

use

d to

sel

ect p

atie

nts

for

op. M

any

w/ n

egat

ive

mye

logr

aphy

had

pos

itive

D

isco

may

be

posi

tive

in th

e fa

ce o

f a

nor

mal

disc

ogra

ms.

mye

logr

am.

Park

, et a

l., 1

979

III

14/4

00 p

atie

nts

w/ b

ack

pain

who

had

abn

orm

al d

isco

gram

s de

spite

nor

mal

pla

in f

ilms

&D

isco

may

dis

clos

e ab

norm

aliti

es in

pat

ient

sin

som

e ca

ses

veno

grap

hy o

r ra

dicu

logr

aphy

.w

/ bac

k pa

in o

ther

wis

e no

rmal

film

s.M

ilette

& M

elan

son,

II

IR

etro

spec

tive

revi

ew o

f la

rge

disc

o se

ries

: 5%

nor

mal

mor

phol

ogy

had

pain

res

pons

e,

Dis

co u

sefu

l for

the

diag

nosi

s of

LB

P; p

ain

1982

37%

abn

orm

al m

orph

olog

y ha

d pa

in r

espo

nse.

resp

onse

is im

port

ant.

John

son

& M

acna

b, 1

985

III

24 p

atie

nts

w/ s

urgi

cally

doc

umen

ted

pseu

dart

hros

is a

t 33

leve

ls w

ere

stud

ied

preo

p w

/ D

isco

app

ears

to b

e us

eful

in d

emon

stra

ting

disc

o. 2

0/29

suc

cess

fully

inje

cted

leve

ls r

esul

ted

in ty

pica

l bac

k pa

in. T

here

wer

e 3

pa

in r

elat

ed to

PA

.fa

lse-

posi

tive

& 9

fal

se-n

egat

ive

disc

ogra

ms.

Gib

son,

et a

l., 1

986

II (

MR

I vs

dis

co m

orph

olog

y)50

dis

cs s

tudi

ed w

/ MR

I &

dis

co. C

onco

rdan

t mor

phom

etri

c fi

ndin

gs w

ere

note

d in

44,

&M

RI

is a

s go

od o

r su

peri

or to

dis

co f

or id

entif

i-in

6 d

iscs

err

ors

wer

e m

ade

by th

e ob

serv

ers.

catio

n of

dis

c de

gene

ratio

n: L

R1

8, L

R2

0.05

.G

rubb

, et a

l., 1

987

III

Dis

co w

as p

erfo

rmed

on

346

disc

s in

108

pat

ient

s. 7

8% h

ad p

ain

repr

oduc

ed a

t 1 o

r m

ore

Dis

co m

ore

sens

itive

than

pla

in f

ilms

or m

yelo

-le

vels

; 37%

had

abn

orm

aliti

es o

n pl

ain

film

s or

mye

logr

ams.

gram

for

eva

luat

ion

of L

BP.

Sach

s, e

t al.,

198

7II

ID

evel

oped

new

cla

ssif

icat

ion

syst

em f

or C

Tdi

sco

& a

pplie

d it

to g

roup

of

59 p

atie

nts.

N

ew c

lass

ific

atio

n of

CT

disc

o ai

ds in

dia

gnos

is:

13%

of

patie

nts

had

posi

tive

CT

disc

o fi

ndin

gs w

/o p

ain

prov

ocat

ion.

LR

11.

46; L

R2

0.16

.Sc

hnei

derm

an, e

t al.,

II

101

disc

leve

ls s

tudi

ed b

y T

2M

RI

& d

isco

grap

hy. M

RI

was

99%

acc

urat

e in

pre

dict

ing

MR

I 99

% a

ccur

ate

in p

redi

ctin

g m

orph

olog

ical

19

87m

orph

olog

ical

dis

cogr

aphi

c re

sults

.di

scog

ram

res

ults

. LR

149

; LR

2 0

.02.

Blu

men

thal

, et a

l, 19

88II

I34

pat

ient

s w

/ pos

itive

dis

co u

nder

wen

t AL

IF. O

f th

ose

who

ach

ieve

d fu

sion

, 73%

had

a

Succ

essf

ul f

usio

n of

dis

co-p

ositi

ve d

iscs

w/

good

clin

ical

res

ult.

AL

IF r

esul

ts in

goo

d re

sults

73%

of

the

time.

Col

houn

, et a

l., 1

988

III

(no

patie

nts

w/ n

orm

al

195

patie

nts

wer

e st

udie

d w

/ dis

co &

182

wen

t on

to s

olid

fus

ion

(of

som

e ty

pe).

Of

thos

e T

reat

men

t of

disc

s w

/ pai

n pr

ovoc

atio

n &

ab-

di

sco

trea

ted;

sel

ectio

nw

/ abn

orm

al &

pai

nful

dis

cs (

137)

, 89%

der

ived

ben

efit

from

fus

ion.

Of

thos

e w

ith a

b-no

rmal

mor

phol

ogy

resu

lts in

suc

cess

ful o

ut-

crite

ria

uncl

ear;

def

initi

onno

rmal

but

non

pain

ful d

iscs

(25

), 5

2% h

ad g

ood

outc

omes

.co

me

89%

of

the

time.

Tre

atm

ent o

f ab

norm

al

of s

ucce

ss u

ncle

ar)

but n

ot p

ainf

ul d

iscs

res

ults

in s

ucce

ss 5

2%

of th

e tim

e.V

anha

rant

a, e

t al.,

198

8II

I81

6 di

scog

ram

s pe

rfor

med

in p

atie

nts

with

mul

tiple

low

-bac

k di

sord

ers.

Man

y ab

norm

al-

Dis

co a

bnor

mal

ities

are

com

mon

in p

atie

nts

ities

see

n.w

/ low

-bac

k di

sord

ers.

Zuc

herm

an, e

t al.,

198

8II

I18

pat

ient

s id

entif

ied

w/ n

orm

al M

RI

& a

bnor

mal

dis

co.

Dis

cos

may

be

abno

rmal

in f

ace

of n

orm

al M

RI.

Van

hara

nta,

et a

l., 1

989

III

790

disc

s st

udie

d w

/ dis

cogr

am. S

core

d by

DPQ

. 87%

of

norm

al d

iscs

wer

e pa

inle

ss. S

light

Prop

ortio

n of

pai

nles

s bu

t deg

ener

ated

dis

cs o

n de

gene

ratio

n w

as a

ssoc

iate

d w

/ pai

n in

33%

of

youn

ger

patie

nts

& 1

2% o

f ol

dest

gro

up.

disc

o in

crea

ses

w/ a

ge.

Thi

s pa

ttern

was

rep

rodu

ced

in m

oder

atel

y &

sev

erel

y de

gene

rate

d di

scs.

Ant

ti-Po

ike,

et a

l., 1

990

II (

disc

ogra

phic

mor

phol

ogy

Abn

orm

al m

orph

olog

y on

dis

cogr

ams

asso

ciat

ed w

/ pai

n pr

ovoc

atio

n 52

.8%

; nor

mal

N

orm

al m

orph

olog

y ha

s N

PVof

87%

for

pai

n vs

pai

n pr

ovoc

atio

n)m

orph

olog

y as

soci

ated

w/ p

ain

prov

ocat

ion

13.2

%. U

se o

f po

stdi

sco

CT

did

not a

ddpr

ovoc

atio

n &

abn

orm

al m

orph

olog

y ha

s a

to d

iagn

ostic

acc

urac

y; m

orph

olog

y PP

V53

% &

NPV

87%

.PP

Vof

53%

for

pai

n pr

ovoc

atio

n. L

R1

3;

LR

20.

34.

Ber

nard

, 199

0II

I (u

tility

of

disc

o); I

I (M

RI

250

patie

nts

stud

ied

w/ d

isco

/CT.

Dis

co a

dded

use

ful i

nfor

mat

ion

93%

of

the

time

(ac-

MR

I co

rrel

ated

w/ d

isco

/CT

89%

of

the

time.

vs d

isco

)co

rdin

g to

rad

iolo

gist

aut

hor)

. PPV

of M

RI

com

pare

d w

/ dis

co w

as 9

2%, N

PVw

as 8

4%.

Lin

son

& C

row

e, 1

990

IIPr

ospe

ctiv

e co

mpa

riso

n bt

wn

T2

MR

I &

dis

co w

as p

erfo

rmed

(94

%co

rrel

atio

n). P

PVM

RI

A

bnor

mal

MR

I ha

s hi

gh li

kelih

hood

for

pai

n-vs

pai

nful

dis

co w

as 9

8% &

NPV

was

88%

.fu

l dis

c on

dis

co (

LR

130

) &

nor

mal

MR

I ha

s an

LR

20.

09.

Wal

sh, e

t al.,

199

0II

I (n

o se

nsiti

vity

abl

e to

be

7 L

BP

patie

nts

& 1

0 vo

lunt

eers

und

erw

ent m

ultil

evel

dis

co. D

isco

was

abn

orm

al 1

7% o

f R

adio

logi

cal r

esul

ts o

f di

sco

are

unre

liabl

e.

repo

rted

)th

e tim

e in

asy

mpt

omat

ic p

atie

nts,

but

no

patie

nt h

ad a

pos

itive

pai

n re

spon

se.

Pain

res

pons

e is

rel

iabl

e fo

r th

e de

term

ina-

tion

of p

ainf

ul d

isc

dise

ase.

Sim

mon

s, e

t al.,

199

1II

(M

RI

vs d

isco

as

gold

16

4 pa

tient

s w

/ LB

Pun

derw

ent d

isco

& M

RI.

Com

pare

d to

pai

n-pr

ovok

ing

disc

o, M

RI

had

MR

I is

ove

rsen

sitiv

e fo

r di

agno

sis

of p

ainf

ul

stan

dard

); I

II (

MR

I or

dis

-an

NPV

of 9

4% &

a P

PVof

58%

.di

sc d

isea

se. A

bnor

mal

dis

co o

ccur

s de

spite

co

vs

pain

ful d

isc

dise

ase)

nega

tive

MR

I 6%

of

the

time.

Con

tinu

ed




665

TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

oncl

usio

ns

Gill

& B

lum

enth

al, 1

992

II53

pat

ient

s un

derw

ent L

5–S1

fus

ion

for

conc

orda

nt p

ain

on d

isco

. Tho

se w

/ con

cord

ant

MR

I pr

edic

ts m

orph

olog

ical

cha

nges

on

disc

opa

in &

abn

orm

al M

RI

did

wel

l 75%

of

the

time;

thos

e w

/ con

cord

ant p

ain

& n

orm

al

wel

l. Pa

tient

s w

/ nor

mal

MR

I &

con

cord

ant

MR

I di

d w

ell 5

0% o

f th

e tim

e. T

he a

utho

rs f

ound

that

MR

I pr

edic

ted

mor

phol

ogic

al

pain

on

disc

o do

rel

ativ

ely

poor

ly f

ollo

win

gch

ange

s on

dis

co 1

00%

of

time.

fusi

on. P

PVof

abn

orm

al M

RI

in th

is s

ettin

gis

74%

(as

opp

osed

to 6

6% o

vera

ll fo

r co

n-co

rdan

t pai

n re

spon

se).

Hor

ton

& D

afta

ri, 1

992

III

63 d

iscs

in 2

5 pa

tient

s st

udie

d w

/ MR

I an

d di

sco.

MR

I si

gnal

pat

tern

s co

rrel

ated

w/ d

isco

Cer

tain

MR

I fi

ndin

gs a

re h

ighl

y as

soci

ated

w/

find

ings

.co

ncor

dant

pai

n pr

ovoc

atio

n &

a n

orm

al im

-ag

e w

as h

ighl

y as

soci

ated

w/ n

o co

ncor

dant

pa

in. T

here

are

inte

rmed

iate

sig

nal c

hara

cter

-is

tics

that

do

not r

elia

bly

pred

ict p

ain

prov

o-ca

tion.

Mae

zaw

a &

Mur

o, 1

992

III

Lar

ge s

erie

s of

dis

co (

1477

). I

mpe

rfec

t rel

atio

nshi

p of

pai

n to

rad

iogr

aphi

c fi

ndin

gs in

D

isc

mor

phol

ogy

& p

ain

resp

onse

not

nec

es-

patie

nts

w/ L

BP

was

fou

nd.

sari

ly r

elat

ed.

Mur

tagh

& A

rrin

gton

, II

IA

utho

rs s

tudi

ed d

iscs

adj

acen

t to

dege

nera

tive

leve

ls to

det

erm

ine

whe

ther

to in

clud

e ad

-M

orph

olog

ical

abn

orm

aliti

es a

re p

rese

nt in

19

92ja

cent

leve

l in

fusi

on. T

hey

foun

d m

orph

olog

ical

dis

cogr

aphi

c ab

norm

aliti

es in

54%

of

~ ha

lf o

f di

scs

adja

cent

to o

ther

deg

ener

ativ

e ad

jace

nt d

iscs

.di

scs.

Bui

rski

& S

ilber

stei

n,

III

MR

I ab

norm

al d

iscs

wer

e ch

arac

teri

zed

in s

ympt

omat

ic &

asy

mpt

omat

ic p

atie

nts.

No

in-

MR

I is

unr

elia

ble

for

the

iden

tific

atio

n of

pai

nful

19

93te

rgro

up d

iffe

renc

es w

ere

seen

in te

rms

of f

requ

ency

or

seve

rity

of

MR

I ch

ange

s. A

ll

disc

s.M

RI

abno

rmal

dis

cs s

ubje

cted

to d

isco

grap

hy w

ere

foun

dto

be

pain

ful.

Kno

x &

Cha

pman

, 199

3II

I22

pat

ient

s w

/ dis

co-p

ositi

ve p

ain

had

AL

IF p

erfo

rmed

bas

ed o

n di

sco.

All

2-le

vel f

usio

ns

AL

IF b

ased

on

disc

ogra

phy

asso

ciat

ed w

/ poo

r di

d po

orly

. Am

ong

sing

le-l

evel

fus

ions

, 35%

wer

e go

od, 1

8% w

ere

fair,

& 4

7% w

ere

poor

.re

sults

.B

righ

tbill

, et a

l., 1

994

III

7 pa

tient

s w

ere

foun

d to

hav

e ab

norm

al d

isco

des

pite

nor

mal

MR

I.N

orm

al M

RI

does

not

exc

lude

abn

orm

al d

isco

.L

oner

agan

, et a

l., 1

994

III

Smal

l ser

ies

of p

atie

nts

wer

e su

bjec

ted

to d

isco

& M

RI.

Ove

rall

conc

orda

nce

for

the

2 C

onco

rdan

ce b

twn

MR

I &

CT

disc

o (m

orph

ol-

mod

aliti

es w

as 9

0%.

ogy)

is ~

90%

.M

onet

a, e

t al.,

199

4II

IC

orre

latio

n w

as n

oted

btw

n di

scog

raph

ic p

atte

rn &

pai

n pr

ovoc

atio

n. O

uter

anu

lar

tear

s A

nula

r di

srup

tion

as s

een

on d

isco

is a

ssoc

i-w

ere

asso

ciat

ed w

/ pai

n; h

owev

er, g

ener

aliz

ed b

ut d

egen

erat

ion

was

not

.at

ed w

/ a p

ain

resp

onse

.W

etze

l, et

al.,

199

4II

I48

pat

ient

s tr

eate

d w

/ fus

ion

base

d on

sym

ptom

atic

dis

co. 4

6% o

f pa

tient

s ha

d an

exc

elle

nt

Sym

ptom

atic

dis

cogr

aphy

pre

dict

s su

rgic

al

or g

ood

(sat

isfa

ctor

y) o

utco

me.

Of

the

23 w

/ a s

olid

art

hrod

esis

, 22

had

satis

fact

ory

resu

lts in

46%

of

patie

nts.

Thi

s re

sult

may

cl

inic

al o

utco

mes

. be

rel

ated

to f

usio

n su

cces

s.R

hyne

, et a

l., 1

995

III

25 p

atie

nts

w/ L

BP

and

conc

orda

nt p

rovo

cativ

e di

sco

wer

e fo

llow

ed n

onop

for

var

ious

M

ost p

atie

nts

w/ d

isco

con

cord

ant b

ack

reas

ons

for

a m

inim

um o

f 3

yrs:

68%

impr

oved

, 8%

sta

yed

the

sam

e, 2

4% w

orse

ned.

pa

in im

prov

e w

/o tr

eatm

ent.

Mos

t pat

ient

s w

ho w

orse

ned

had

sign

ific

ant p

sych

iatr

ic d

isea

se.

Schw

arze

r, et

al.,

199

5II

I92

pat

ient

s w

/ LB

Pst

udie

d w

/ dis

co. P

rovo

catio

n di

sco

was

pos

itive

in 3

9%.

Prov

ocat

ive

disc

o is

pos

itive

in a

larg

e nu

mbe

rof

pat

ient

s w

/ bac

k pa

in.

Blo

ck, e

t al.,

199

6II

Patie

nts

w/ L

BP

wer

e st

udie

d w

ith d

isco

& M

MPI

. Tho

se w

ho r

epor

ted

pain

had

sig

nifi

-Ps

ycho

logi

cal f

acto

rs c

ontr

ibut

ed h

eavi

ly to

ca

ntly

hig

her

hypo

chon

dria

sis

& h

yste

ria

scor

es o

n th

e M

MPI

.di

sco

resu

lts.

Ric

kets

on, e

t al.,

199

6II

I80

dis

cs in

29

patie

nts

stud

ied

w/ M

RI

& d

isco

. No

patie

nt w

/ HIZ

had

a m

orph

olog

ical

ly

Pres

ence

of

HIZ

doe

s no

t nec

essa

rily

pre

dict

no

rmal

dis

c. N

o de

fini

te r

elat

ions

hip

was

fou

nd b

twn

HIZ

& p

ain

resp

onse

(on

ly 7

pa-

pain

ful d

isc:

LR

11.

3; L

R2

0.96

.tie

nts

w/ H

IZ; d

iscr

epan

cy b

twn

tabl

es &

rep

orte

d re

sults

).Sc

helh

as, e

t al.,

199

6II

100

patie

nts

w/ H

IZ d

iscs

& 6

7 pa

tient

s w

/ non

-HIZ

dis

cs s

ubje

cted

to d

isco

grap

hy.

Pres

ence

of

HIZ

hig

hly

corr

elat

ed w

/ pai

n on

87

/100

HIZ

dis

cs a

nd 2

/67

non-

HIZ

dis

cs p

ainf

ul: P

PVH

IZ 8

7%, N

PVno

n-H

IZ 9

7%.

disc

o (L

R1

5.76

; LR

20.

002)

.H

egge

ness

, et a

l., 1

997

III

Ret

rosp

ectiv

e re

view

of

83 p

ostd

isce

ctom

y pa

tient

s w

ho u

nder

wen

t dis

cogr

aphy

was

per

-Pr

evio

usly

op

disc

s w

ere

mor

e fr

eque

ntly

pai

n-fo

rmed

.fu

l tha

n no

nop

disc

s (7

2 vs

38%

). D

ye e

x-tr

avas

atio

n w

as a

ssoc

iate

d w

/ pai

n 75

% o

f th

e tim

e.B

raith

wai

te, e

t al.,

199

8II

(M

RI

vs d

isco

)D

isco

per

form

ed a

t 152

leve

ls in

clud

ing

23 w

/ mod

ic c

hang

es. P

PVfo

r m

odic

cha

nges

M

odic

cha

nges

on

MR

I ar

e re

lativ

ely

spec

ific

, pr

edic

ting

conc

orda

nt p

ain

was

91.

3% &

the

NPV

was

46.

5%.

but n

ot s

ensi

tive

for

the

conc

orda

nce

of p

ain

on d

isco

grap

hy. L

R1

7.6,

LR

2 0

.8. Con

tinu

ed




TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

oncl

usio

ns

Ito,

et a

l., 1

998

II (

MR

I vs

dis

co)

Ret

rosp

ectiv

e co

mpa

riso

n of

MR

I H

IZ v

s pa

infu

l dis

c on

dis

co w

ere

perf

orm

ed.

HIZ

was

sen

sitiv

e fo

r pr

edic

ting

a pa

infu

l dis

c(8

7%)

but n

ot s

peci

fic

(65%

; PPV

43%

).

“Mas

sive

deg

ener

atio

n” o

r “s

ever

e di

sc s

pace

colla

pse”

was

spe

cifi

c fo

r pa

infu

l dis

co. T

he

PPV

for

nonc

onco

rdan

t pai

n or

no

pain

for

ano

rmal

MR

I ap

pear

ance

was

97.

3%.

Saif

uddi

n, e

t al.,

199

8II

Mor

phol

ogy

of d

isco

com

pare

d w

/ pai

n re

spon

se; r

espo

nse

usua

lly a

ssoc

iate

d w

/ iso

late

d Is

olat

ed p

oste

rior

anu

lar

tear

s si

gnif

ican

tly a

s-po

ster

ior

anul

ar te

ar.

soci

ated

w/ c

onco

rdan

t pai

n.Sa

ifud

din,

et a

l., 1

998

IIH

IZ p

rese

nce

was

com

pare

d w

/ dis

co r

esul

ts:s

ensi

tivity

26.

7%, s

peci

fici

ty 9

5.2%

, H

IZ is

hig

hly

spec

ific

& a

ssoc

iate

d w

/ a h

igh

PPV

88.9

%, &

NPV

47%

.lik

elih

ood

for

pain

pro

voca

tion;

LR

16.

8;

LR

20.

7.Sm

ith, e

t al.,

199

8II

(H

IZ v

s pa

infu

l dis

co)

Ret

rosp

ectiv

e an

alys

is o

f pa

tient

s w

/ MR

I &

dis

cogr

am w

/in s

ame

year

. Ak

valu

e of

0.5

7 H

IZ is

a s

peci

fic

but n

ot v

ery

sens

itive

indi

ca-

was

fou

nd b

twn

neur

orad

iolo

gist

s ev

alua

ting

scan

s fo

r H

IZ. S

ensi

tivity

of

HIZ

for

an-

tor

of p

ainf

ul d

isc

disr

uptio

n.ul

ar d

isru

ptio

n w

as o

nly

25%

, but

spe

cifi

city

was

99%

. Sen

sitiv

ity f

or p

ain

resp

onse

w

as 2

3% &

spe

cifi

city

90%

.C

arra

gee,

et a

l., 1

999

II

24 d

iscs

in 8

sel

ecte

d pa

tient

s w

/ non

spin

al L

BP

wer

e in

ject

ed: c

onco

rdan

t pai

n el

icite

d in

Lum

bar

disc

o ca

nnot

rel

iabl

y di

ffer

entia

te th

e 8

disc

s &

4 p

atie

nts

had

seve

re p

ain

in 1

dis

c &

no

pain

in o

ther

s (m

et c

rite

ria

for

op).

sour

ce o

f L

BP

due

to a

hig

h–fa

lse

posi

tive

rate

(50

%):

LR

1 0

.72,

LR

2 0

.72.

Der

by, e

t al.,

199

9II

IR

etro

spec

tive

com

pari

son

of p

atie

nts

w/ p

ositi

ve d

isco

who

und

erw

ent d

iffe

rent

fus

ion

Patie

nts

w/ “

chem

ical

ly s

ensi

tive”

dis

cs d

o be

t-pr

oced

ures

.te

r w

/ an

inte

rbod

y fu

sion

com

pare

d w

/ PL

F.M

ilette

, et a

l., 1

999

IIR

etro

spec

tive

com

pari

son

of M

RI

find

ings

& d

isco

fin

ding

s w

as p

erfo

rmed

.L

oss

of d

isc

spac

e he

ight

, abn

orm

al c

entr

al s

ig-

nal i

nten

sity

, pro

trus

ions

& d

isc

bulg

es p

re-

dict

ed S

tage

2 o

r 3

disr

uptio

ns o

n di

sco;

m

ost o

f th

ese

wer

e pa

infu

l.R

anki

ne, e

t al.,

199

9II

IO

bser

vatio

nal s

tudy

of

83 p

atie

nts

w/ b

ack

&/o

r le

g pa

in. 4

5% h

ad a

n H

IZ n

oted

on

MR

I.H

IZ is

a c

omm

on f

indi

ng in

pat

ient

s w

/ LB

P.

Car

rage

e, e

t al.,

200

0II

26 p

atie

nts

w/o

bac

k pa

in f

ollo

wed

for

1 y

r af

ter

disc

ogra

phy.

No

patie

nt w

/o a

som

atiz

a-D

isco

doe

s no

t cau

se c

hron

ic L

BP

in p

atie

nts

tion

diso

rder

suf

fere

d ba

ck p

ain

as a

res

ult o

f di

sco.

w/o

som

atiz

atio

n di

sord

ers.

Pai

n co

rrel

ates

w

/ som

atiz

atio

n (p

,0.

03).

Car

rage

e, e

t al.,

200

0II

2 se

lect

ed g

roup

s of

pat

ient

s st

atus

pos

tdis

cect

omy

w/ o

r w

/o b

ack

pain

. Dis

co w

as p

ositi

ve

Lum

bar

disc

o ha

s a

high

–fal

se p

ositi

ve r

ate

inin

40%

of

asym

ptom

atic

gro

up &

63%

of

sym

ptom

atic

gro

up.

patie

nts

w/o

bac

k pa

in (

40%

): L

R1

1.05

, L

R2

0.93

.C

arra

gee,

et a

l., 2

000

IISe

lect

ed p

opul

atio

ns o

f L

BP

& n

on-L

BP

stud

ied

w/ M

RI

& d

isco

. LB

Ppa

tient

s ha

d si

g-H

IZ is

too

nons

peci

fic

for

clin

ical

use

in L

BP;

ni

fica

ntly

hig

her

rate

of

HIZ

, but

24%

of

asym

ptom

atic

pat

ient

s al

so h

ad H

IZ.

it m

ay p

redi

ct d

isco

gram

pai

n: L

R1

14.6

, L

R2

0.60

.C

arra

gee,

et a

l., 2

000

III

Fals

e-po

sitiv

e ra

te o

f di

sco

in n

onco

mpe

nsat

ed p

atie

nt w

as lo

w (

10%

), b

ut m

uch

high

er in

Po

sitiv

e di

scog

ram

rat

es c

orre

late

d cl

osel

y w

/co

mpe

nsat

ed p

atie

nts

& th

ose

w/ s

omat

izat

ion

diso

rder

s.di

sabi

lity

clai

ms,

som

atiz

atio

n, &

anu

lar

dis-

rupt

ion.

L

am, e

t al.,

200

0II

92 H

IZs

wer

e id

entif

ied

in 7

3 pa

tient

s un

derg

oing

wor

kup

for

fusi

on f

or L

BP.

Blin

ded

HIZ

goo

d pr

edic

tor

of p

ainf

ul d

isco

.co

mpa

riso

n bt

wn

MR

I pr

esen

ce o

f H

IZ &

pai

n re

spon

se. P

PVfo

r H

IZ f

or p

ain

resp

onse

was

87%

.Sa

ndhu

, et a

l., 2

000

III

Com

pare

d m

odic

cha

nges

& r

esul

ts o

f di

scog

raph

y in

53

patie

nts

w/ L

BP.

Not

all

leve

ls

Mod

ic c

hang

es o

n M

RI

do n

ot p

redi

ct p

ain

unde

rwen

t dis

co. A

utho

rs f

ound

no

rela

tions

hip

btw

n m

odic

cha

nges

& p

ain

prov

oca-

prov

ocat

ion

w/ d

isco

.tio

n on

dis

co.

Slip

man

, et a

l., 2

001

III

40 c

aref

ully

sel

ecte

d di

scog

ram

s ev

alua

ted

to d

eter

min

e re

latio

n of

sid

e of

anu

lar

tear

&

Side

of

anul

ar te

ar n

ot r

elat

ed to

sid

e of

bac

k si

de o

f sy

mpt

oms

in p

atie

nts

w/ c

onco

rdan

t bac

k pa

in. N

o re

latio

nshi

p fo

und.

or le

g pa

in.

Wei

shau

pt, e

t al.,

200

1II

(M

RI

vs d

isco

)Pr

ospe

ctiv

e st

udy

of 5

0 pa

tient

s w

/ LB

Pw

ho u

nder

wen

t MR

I &

dis

co. N

orm

al M

RI

had

MR

I go

od to

ol f

or p

redi

ctin

g pa

in p

rovo

catio

n N

PVfo

r pa

in p

rovo

catio

n of

98%

. Onl

y m

oder

ate-

to-s

ever

e en

dpla

te c

hang

es h

ad a

w

/ dis

co.

high

PPV

(100

%).

*A

LIF

= a

nter

ior

lum

bar

inte

rbod

y fu

sion

; di

sco

= d

isco

grap

hy;

DPQ

= D

alla

s Pa

in Q

uest

ionn

aire

; L

BP

= l

ow-b

ack

pain

; L

R =

lik

elih

ood

ratio

; M

MPI

= M

inne

sota

Mul

tipha

sic

Pers

onal

ity I

nven

tory

;PA

= p

seud

arth

rosi

s.


cordant pain provocation during discography. Weishaupt,et al.,61 found that moderate-to-severe endplate changespredicted a concordant pain response 100% of the time. Incontrast, Sandhu, et al.,49 did not identify a significantrelationship between modic changes on MR imaging andconcordant pain responses during discography. Ito andcolleagues26 found that the absence of an HIZ had a stronglikelihood (negative ratio 0.08) of predicting the absenceof a pain response. Conversely, Saifuddin, et al.,47 report-ed that the presence of an HIZ on MR imaging was spe-cific (96%) and predictive (likelihood ratio 6.8) of a con-cordant pain response during discography. Schelhas, etal.,50 reported similar findings in that the presence (posi-tive likelihood 5.8) or absence (negative likelihood 0.002)of an HIZ on MR imaging was predictive of the presenceor absence, respectively, of concordant pain during dis-cography.

Lam and colleagues31 performed a prospective blindedevaluation of HIZs identified on MR imaging comparedwith discography. They found an 87% PPV of the HIZ forthe provocation of pain with discography and reportedsensitivity and specificity values of 81 and 79%, respec-tively. Ricketson and colleagues45 identified a significantassociation between the presence of an HIZ and concor-dant back pain during discography; however, only sevenHIZs were noted. Several other studies confirmed the highNPV (94–100%) of a normal MR image for the produc-tion of a concordant pain response during discography.25,26,

47,50,55,61 Although there are conflicting reports, the majori-ty of evidence reported in the literature indicates that cer-tain MR imaging findings, particularly the presence of anHIZ, are closely correlated with the provocation of disco-graphic concordant pain in patients with low-back pain. Itis also apparent that a concordant pain response is ex-tremely uncommon in the presence of normal MR imag-ing findings.

The knowledge of the relative ability of one imagingstudy (such as MR imaging) to predict the results ofanother diagnostic test (such as provocative discography)is useful for the selection of diagnostic tests; however, thetrue litmus test is the ability of the diagnostic test to pre-dict the outcome of treatment based on the results of thetest. In the low-back pain population, fusion is often per-formed to treat patients with recalcitrant low-back pain.The next relevant question concerns the ability of discog-raphy or MR imaging to predict the outcome after lumbarfusion. If discography (or MR imaging) were to have anaccuracy of 100% in terms of diagnosing the source of apatient’s low-back pain and if successful fusion of thepathological interspace diagnosed using discography (orMR image) were 100% effective for the treatment of low-back pain, then every patient with a positive discogram (orMR image) and a successful fusion would be expected toexperience relief of low-back pain. Conversely, a patientwith a negative discogram (or MR image) would not ex-perience pain relief despite a successful fusion.

To address this issue, Gill and Blumenthal21 reported onthe outcomes of 53 patients who underwent L5–S1 fusion,based primarily on concordant pain provocation duringdiscography. They found that patients with concordant painand abnormal MR imaging findings did well approximate-ly 75% of the time. This success rate was compared withresults obtained by the same authors in a group of patients

similarly treated based on concordant pain on discographybut in whom MR imaging was normal. Only half of theselatter patients experienced a favorable result. There was atrend for an abnormal MR imaging study to predict func-tional outcome following surgery (p , 0.10). Colhoun andcolleagues15 reported an 89% favorable result followingfusion in patients with abnormal disc morphology and aconcordant pain response compared with a 52% favorablerate in patients with abnormal disc morphology alone. Bothof these studies provide Class III medical evidence sug-gesting that both anatomical abnormality and a concordantpain response together are required for a discogram to havea PPV for fusion outcome after lumbar surgery.

Other authors have provided more sobering reports ofoutcomes following lumbar fusion when discography alonehas been used as a diagnostic tool. Wetzel, et al.,62 andKnox and Chapman30 each described surgical series inwhich patient selection was dependent primarily on discog-raphy. The results of both of these series are disappointing,with successful outcome rates of 35 to 46%. These resultsare particularly troubling given the findings by Rhyne, etal.,44 that the majority (68%) of patients with discographicconcordant pain in their experience improved without sur-gical treatment during a 3-year follow-up period. The fu-sion rates and techniques may have influenced the overallresults. In the series by Wetzel, et al.,62 for example, in themajority of cases believed to represent a successful fusionoutcomes were satisfactory. Some authors argue that thetechniques used to achieve fusion are important. For exam-ple, Derby and colleagues18 have suggested that the elimi-nation of motion at the pathological disc space through theuse of interbody implants is important for adequate relief of discogenic pain. This hypothesis is partly based on theobservation that discography can elicit pain at disc spaceswithin a solidly fused segment following PLF.28 Conse-quently, although acceptable results following surgical treat-ment of discography-diagnosed low-back pain have beenreported, the best medical evidence suggests that treatmentof a disc in a patient with low-back pain, a positive disco-gram, and a normal MR imaging study is not likely to influ-ence favorably the natural history of the pain. Discographyis not, therefore, recommended for the evaluation of pa-tients with normal MR imaging examinations of the lumbarspine.

Discography has been used as an adjunct for the study ofdiscs associated with equivocal MR imaging findings, par-ticularly those adjacent to clearly pathological interspacesconsidered for fusion. Discs that are morphologically ab-normal but painless at discography may be excluded fromthe fusion construct.40 Discography may also have a role inthe diagnosis of painful pseudarthrosis, although the litera-ture on this is scant.28 Provocation of pain at disc levels thatare morphologically normal on MR imaging is a contra-indication for surgical (or other invasive) intervention.Discography-provoked pain at multiple disc levels in a pa-tient with equivocal morphological findings on discogra-phy or MR imaging should raise a significant red flag forthe presence of factors reported to be associated with poorsurgical outcomes following lumbar fusion surgery.4,11

Summary

Discography is an exquisitely sensitive but not specific



667


diagnostic test for the diagnosis of discogenic low-backpain. The restriction of the definition of a positive disco-graphic study to one that elicits concordant pain from amorphologically abnormal disc improves the definition’saccuracy. Fusion surgery based on discography alone,however, is not reliably associated with clinical success.Therefore, discography is not recommended as a stand-alone test for treatment decisions in patients with low-back pain. Magnetic resonance imaging is a sensitive andnoninvasive test for the presence of degenerative disc dis-ease. Discography should not be attempted in patientswith normal lumbar MR images. Discography appears tohave a role in the evaluation of patients with low-backpain, but it is best limited to the evaluation of abnormalinterspaces identified on MR imaging, the investigation of adjacent-level disc disease, and as a means to rule outcases of nonorganic pain from surgical consideration.

Directions for Future Research

A large cohort series comparing the results of discogra-phy and MR imaging for predicting the success of surgi-cal intervention via a standardized protocol would be avaluable addition to the literature. These data would pro-vide at least Class II evidence for the value of either imag-ing technique for predicting the response of a patient to agiven treatment strategy.

References

1. Antti-Poika I, Soini J, Tallroth K, et al: Clinical relevance ofdiscography combined with CT scanning. A study of 100 pa-tients. J Bone Joint Surg Br 72:480–485, 1990

2. Bernard TN Jr: Lumbar discography followed by computed to-mography. Refining the diagnosis of low-back pain. Spine 15:690–707, 1990

3. Birney TJ, White JJ Jr, Berens D, et al: Comparison of MRI anddiscography in the diagnosis of lumbar degenerative disc dis-ease. J Spinal Disord 5:417–423, 1992

4. Block AR, Vanharanta H, Ohnmeiss DD, et al: Discographicpain report. Influence of psychological factors. Spine 21:334–338, 1996

5. Blumenthal SL, Baker J, Dossett A, et al: The role of anteriorlumbar fusion for internal disc disruption. Spine 13:566–569,1988

6. Braithwaite I, White J, Saifuddin A, et al: Vertebral end-plate(Modic) changes on lumbar spine MRI: correlation with painreproduction at lumbar discography. Eur Spine J 7:363–368,1998

7. Brightbill TC, Pile N, Eichelberger RP, et al: Normal magneticresonance imaging and abnormal discography in lumbar discdisruption. Spine 19:1075–1077, 1994

8. Brodsky AE, Binder WF: Lumbar discography. Its value in di-agnosis and treatment of lumbar disc lesions. Spine 4:110–120,1979

9. Buirski G, Silberstein M: The symptomatic lumbar disc in pa-tients with low-back pain. Magnetic resonance imaging appear-ances in both a symptomatic and control population. Spine18:1808–1811, 1993

10. Carragee EJ, Chen Y, Tanner CM, et al: Can discography causelong-term back symptoms in previously asymptomatic sub-jects? Spine 25:1803–1808, 2000

11. Carragee EJ, Chen Y, Tanner CM, et al: Provocative discogra-phy in patients after limited lumbar discectomy: a controlled,randomized study of pain response in symptomatic and asymp-tomatic subjects. Spine 25:3065–3071, 2000

12. Carragee EJ, Paragioudakis SJ, Khurana S: 2000 Volvo Awardwinner in clinical studies: Lumbar high-intensity zone and dis-cography in subjects without low back problems. Spine 25:2987–2992, 2000

13. Carragee EJ, Tanner CM, Khurana S, et al: The rates of false-positive lumbar discography in select patients without low backsymptoms. Spine 25:1373–1381, 2000

14. Carragee EJ, Tanner CM, Yang B, et al: False-positive findingson lumbar discography. Reliability of subjective concordanceassessment during provocative disc injection. Spine 24:2542–2547, 1999

15. Colhoun E, McCall IW, Williams L, et al: Provocation discog-raphy as a guide to planning operations on the spine. J BoneJoint Surg Br 70:267–271, 1988

16. Collins CD, Stack JP, O’Connell DJ, et al: The role of discogra-phy in lumbar disc disease: a comparative study of magnetic reso-nance imaging and discography. Clin Radiol 42:252–257, 1990

17. Collis JS Jr, Gardner W: Lumbar discography. An analysis ofone thousand cases. J Neurosurg 19:452–461, 1962

18. Derby R, Howard MW, Grant JM, et al: The ability of pressure-controlled discography to predict surgical and nonsurgical out-comes. Spine 24:364–372, 1999

19. Fraser RD, Osti OL, Vernon-Roberts B: Discitis after discogra-phy. J Bone Joint Surg Br 69:26–35, 1987

20. Gibson MJ, Buckley J, Mawhinney R, et al: Magnetic reso-nance imaging and discography in the diagnosis of disc degen-eration. A comparative study of 50 discs. J Bone Joint Surg Br68:369–373, 1986

21. Gill K, Blumenthal SL: Functional results after anterior lumbarfusion at L5–S1 in patients with normal and abnormal MRIscans. Spine 17:940–942, 1992

22. Grubb SA, Lipscomb HJ, Guilford WB: The relative value oflumbar roentgenograms, metrizamide myelography, and discog-raphy in the assessment of patients with chronic low-back syn-drome. Spine 12:282–286, 1987

23. Heggeness MH, Watters WC III, Gray PM Jr: Discography oflumbar discs after surgical treatment for disc herniation. Spine22:1606–1609, 1997

24. Holt EP Jr: The question of lumbar discography. J Bone JointSurg Am 50:720–726, 1968

25. Horton WC, Daftari TK: Which disc as visualized by magneticresonance imaging is actually a source of pain? A correlationbetween magnetic resonance imaging and discography. Spine17 (6 Suppl):S164–S171, 1992

26. Ito M, Incorvaia KM, Yu SF, et al: Predictive signs of disco-genic lumbar pain on magnetic resonance imaging with discog-raphy correlation. Spine 23:1252–1260, 1998

27. Jackson RP, Becker GJ, Jacobs RR, et al: The neuroradio-graphic diagnosis of lumbar herniated nucleus pulposus: I. Acomparison of computed tomography (CT), myelography, CT-myelography, discography, and CT-discography. Spine 14:1356–1361, 1989 (Erratum in Spine 15:59, 1990)

28. Johnson RG, Macnab I: Localization of symptomatic lumbarpseudarthroses by use of discography. Clin Orthop Relat Res197:164–170, 1985

29. Junila J, Niinimaki T, Tervonen O: Epidural abscess after lum-bar discography. A case report. Spine 22:2191–2193, 1997

30. Knox BD, Chapman TM: Anterior lumbar interbody fusion fordiscogram concordant pain. J Spinal Disord 6:242–244, 1993

31. Lam KS, Carlin D, Mulholland RC: Lumbar disc high-intensityzone: the value and significance of provocative discography inthe determination of the discogenic pain source. Eur Spine J9:36–41, 2000

32. Lindblom K: Technique and results of diagnostic disc punctureand injection (discography) in the lumbar region. Acta OrthopScand 20:315–326, 1951

33. Linson MA, Crowe CH: Comparison of magnetic resonanceimaging and lumbar discography in the diagnosis of disc degen-eration. Clin Orthop Relat Res 250:160–163, 1990




34. Loneragan R, Khangure MS, McCormick C, et al: Comparisonof magnetic resonance imaging and computed tomographicdiscography in the assessment of lumbar disc degeneration.Australas Radiol 38:6–9, 1994

35. Maezawa S, Muro T: Pain provocation at lumbar discographyas analyzed by computed tomography/discography. Spine 17:1309–1315, 1992

36. McCutcheon ME, Thompson WC III: CT scanning of lumbar dis-cography. A useful diagnostic adjunct. Spine 11:257–259, 1986

37. Milette PC, Fontaine S, Lepanto L, et al: Differentiating lumbardisc protrusions, disc bulges, and discs with normal contour butabnormal signal intensity. Magnetic resonance imaging withdiscographic correlations. Spine 24:44–53, 1999

38. Milette PC, Melanson D: A reappraisal of lumbar discography.J Can Assoc Radiol 33:176–182, 1982

39. Moneta GB, Videman T, Kaivanto K, et al: Reported pain dur-ing lumbar discography as a function of anular ruptures anddisc degeneration. A re-analysis of 833 discograms. Spine 19:1968–1974, 1994

40. Murtagh FR, Arrington JA: Computer tomographically guideddiscography as a determinant of normal disc level before fu-sion. Spine 17:826–830, 1992

41. Osti OL, Fraser RD: MRI and discography of annular tears andintervertebral disc degeneration. A prospective clinical compar-ison. J Bone Joint Surg Br 74:431–435, 1992 (erratum in JBone Joint Surg Br 74:793, 1992)

42. Park WM, McCall IW, O’Brien JP, et al: Fissuring of the pos-terior annulus fibrosus in the lumbar spine. Br J Radiol 52:382–387, 1979

43. Rankine JJ, Gill KP, Hutchinson CE, et al: The clinical signifi-cance of the high-intensity zone on lumbar spine magnetic res-onance imaging. Spine 24:1913–1920, 1999

44. Rhyne A, Smith S, Wood K, et al: Outcome of unoperated dis-cogram positive low back pain. Spine 20:1997–2001, 1995

45. Ricketson R, Simmons JW, Hauser BO: The prolapsed inter-vertebral disc. The high-intensity zone with discography corre-lation. Spine 21:2758–2762, 1996

46. Sachs BL, Vanharanta H, Spivey MA, et al: Dallas discogramdescription. A new classification of CT/discography in low-back disorders. Spine 12:287–294, 1987

47. Saifuddin A, Braithwaite I, White J, et al: The value of lumbarspine magnetic resonance imaging in the demonstration of anu-lar tears. Spine 23:453–457, 1998

48. Saifuddin A, Emanuel R, White J, et al: An analysis of radiat-ing pain at lumbar discography. Eur Spine J 7:358–362, 1998

49. Sandhu HS, Sanchez-Caso LP, Parvataneni HK, et al: Associ-ation between findings of provocative discography and verte-bral endplate signal changes as seen on MRI. J Spinal Disord13:438–443, 2000

50. Schellhas KP, Pollei SR, Gundry CR, et al: Lumbar disc high-

intensity zone. Correlation of magnetic resonance imaging anddiscography. Spine 21:79–86, 1996

51. Schneiderman G, Flannigan B, Kingston S, et al: Magnetic res-onance imaging in the diagnosis of disc degeneration: correla-tion with discography. Spine 12:276–281, 1987

52. Schreck RI, Manion WL, Kambin P, et al: Nucleus pulposus pul-monary embolism. A case report. Spine 20:2463–2466, 1995

53. Schwarzer AC, Aprill CN, Derby R, et al: The prevalence andclinical features of internal disc disruption in patients withchronic low back pain. Spine 20:1878–1883, 1995

54. Simmons EH, Segil CM: An evaluation of discography in thelocalization of symptomatic levels in discogenic disease of thespine. Clin Orthop Relat Res 108:57–69, 1975

55. Simmons JW, Emery SF, McMillin JN, et al: Awake discogra-phy. A comparison study with magnetic resonance imaging.Spine 16 (6 Suppl):S216–S221, 1991

56. Slipman CW, Patel RK, Zhang L, et al: Side of symptomatic an-nular tear and site of low back pain: is there a correlation?Spine 26:E165–E169, 2001

57. Smith BM, Hurwitz EL, Solsberg D, et al: Interobserver relia-bility of detecting lumbar intervertebral disc high-intensity zoneon magnetic resonance imaging and association of high-inten-sity zone with pain and anular disruption. Spine 23:2074–2080,1998

58. Vanharanta H, Guyer RD, Ohnmeiss DD, et al: Disc deteriora-tion in low-back syndromes. A prospective, multi-center CT/discography study. Spine 13:1349–1351, 1988

59. Vanharanta H, Sachs BL, Ohnmeiss DD, et al: Pain provocationand disc deterioration by age. A CT/discography study in a low-back pain population. Spine 14:420–423, 1989

60. Walsh TR, Weinstein JN, Spratt KF, et al: Lumbar discographyin normal subjects. A controlled, prospective study. J BoneJoint Surg Am 72:1081–1088, 1990

61. Weishaupt D, Zanetti M, Hodler J, et al: Painful lumbar disk de-rangement: relevance of endplate abnormalities at MR imaging.Radiology 218:420–427, 2001

62. Wetzel FT, LaRocca SH, Lowery GL, et al: The treatment oflumbar spinal pain syndromes diagnosed by discography. Lum-bar arthrodesis. Spine 19:792–800, 1994

63. Zucherman J, Derby R, Hsu K, et al: Normal magnetic resonanceimaging with abnormal discography. Spine 13:1355–1359, 1988

Manuscript received December 7, 2004.Accepted in final form April 11, 2005.Address reprint requests to: Daniel K. Resnick, M.D., Department

of Neurological Surgery, University of Wisconsin Medical School,K4/834 Clinical Science Center, 600 Highland Avenue, Madison,Wisconsin 53792. email: [email protected].



669


Recommendations

Standards. Lumbar fusion is recommended as a treat-ment for carefully selected patients with disabling low-back pain due to one- or two-level degenerative diseasewithout stenosis or spondylolisthesis.

Guidelines. There is insufficient evidence available tosupport a treatment guideline.

Options. An intensive course of physical therapy andcognitive therapy is recommended as a treatment optionfor patients with low-back pain in whom conventionalmedical management has failed.

Rationale

Lumbar spinal fusion procedures are presently beingprovided as a treatment for patients with low-back paindue to lumbar degenerative disease without stenosis orspondylolisthesis. These procedures are associated withsignificant cost and the potential for complications. Therehas been considerable debate regarding the role, if any, oflumbar fusion for the treatment of patients with low-backpain without deformity or neurological deficit. The pur-

pose of this review is to evaluate the published literatureregarding the use of lumbar fusion in this patient popu-lation.

Literature Search

The database of the National Library of Medicine wassearched using the search terms “spinal fusion and random-ized clinical trial,” “lumbar fusion and randomized clini-cal trial,” “spinal fusion and outcomes,” “lumbar fusion andoutcomes,” and “lumbar fusion and physical therapy.” The Cochrane Central Register of Controlled Trials wassearched using the search term “spinal fusion.” Referencelists from relevant papers as well as from the Cochrane Re-view4 were reviewed, and all randomized clinical trialscomparing lumbar fusion with nonoperative managementwere identified. Both of these trials are identified in Table 1.A number of case series, cohort studies, and studies evalu-ating different fusion techniques were also identified andprovide supportive scientific evidence.

Scientific Background

In 1999, Gibson, et al.,4 published an evidence-basedreview of the literature regarding the surgical treatment oflumbar spondylosis by using the protocol developed bythe Cochrane Review. These authors identified 26 RCTsdealing with surgical management of lumbar spondylosis


670

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 7: intractable low-back pain without stenosis orspondylolisthesis



KEY WORDS • fusion • lumbar spine • spondylosis • practice guidelines •treatment outcome


Abbreviations used in this paper: GFS = General Function Scale;ODI = Oswestry Disability Index; PLF = posterolateral fusion;RCT = randomized controlled trial; VAS = visual analog scale.


including disc herniation and degeneration. None of thetrials identified specifically involved the issue of lumbarfusion compared with conservative management of pa-tients with low-back pain. Because of the absence of ClassI medical evidence regarding the efficacy of lumbar fu-sion, the authors concluded that “There is no scientificevidence on the effectiveness of any form of surgical de-compression or fusion for degenerative lumbar spondylo-sis compared with the natural history, placebo, or conser-vative management.”

In 2001, Fritzell and colleagues3 published the results ofa multicenter RCT from the Swedish Lumbar Spine StudyGroup. In their study, 294 patients with disabling backpain thought to be surgical candidates were randomized to conservative care (physical therapy supplemented witheducation and other pain-relieving technologies at the dis-cretion of the treating physician), or one of three surgicaltreatments. Patients were required to have suffered fromlow-back pain for at least 2 years and were required tohave radiographic and clinical evidence of spondylosis atL4–5, L5–S1, or both levels. Surgery was performed byone of 26 participating surgeons and consisted of eitherPLF, PLF supplemented with pedicle screw fixation, orinterbody fusion supplemented with PLF and pediclescrew fixation. The groups were comparable in all mea-sured demographic variables with the exception of a high-er incidence of medical comorbidity in the surgical group.Patients were followed for 2 years with intermediate eval-uations at 6 months and at 1 year following treatment.Results were assessed using multiple well-validated out-come measures including pain VASs, the ODI for low-back pain, the Million VAS, the GFS, Work Status, a pa-tient satisfaction survey, and an independent functionalassessment by a second spine surgeon.3

Follow up was achieved in 98% of the patients enrolledin the study. Appropriate statistical analysis was performedbased on the type of data derived from the different out-come measures. The surgical group did statistically signif-icantly better than the conservatively treated group in termsof pain relief, degree of disability as measured by the ODI,Million, and GFS, return-to-work status, and degree of sat-isfaction reported by the patients and by the independentobserver. The authors’ statistical analysis was rigorous andincluded “intention to treat” as well as “worst case” sce-narios. All primary outcome measures evaluated in thestudy were statistically significantly better in patients ran-

domized to the surgical group compared with the nonsurgi-cal group.3 This study provides Class I medical evidencethat demonstrates that lumbar fusion is associated with bet-ter outcomes than conservative care for appropriately se-lected patients with disabling low-back pain. Considerationfor surgery should be reserved for those patients with per-sistent pain thought to arise from one or two motion seg-ments despite the best medical management available tothe patient.

The study by Fritzell and colleagues,3 despite its rigor-ous design and robust results, has been criticized by pro-ponents of various specific therapies. Mooney,7 for exam-ple, commented that the study was unfairly biased againstconservative care because a trial of the same type of ther-apy had already failed in the enrolled patients prior toentry into the study.

In 2003, Brox, et al.,2 conducted a smaller randomizedstudy evaluating the relative efficacy of instrumented PLFcompared with a specific protocol of cognitive interven-tion and physical therapy. In their study, patients were re-quired to have suffered low-back pain for 1 year and havean ODI score of 30 to 100 and radiographic evidence ofdegeneration at L4–5, L5–S1, or both. The primary out-come measure used was a modified ODI (modified for theNorwegian population). This particular outcome measurewas previously studied in the target population and foundto be reliable, although the authors specifically stated thatup to a 12-point difference on the scale may be attributedto random error.1,5,6 Secondary outcome measures includ-ed pain VASs, daily use of medication, GFS, the WaddelFear Avoidance Belief Questionnaire, and a patient satis-faction score. Outcomes were assessed by physical thera-pists or rehabilitation physicians at 1 year following initi-ation of treatment.

Patients randomized to the surgical arm were treatedwith instrumented PLF. Postoperative physical therapy wasvariably used at the discretion of the surgeon and was notcodified. The patients enrolled in the physical therapy armunderwent a program specifically designed for patientswith low-back pain that was thought to be more effectivethan standard conservative care based on a previous study.1This more comprehensive program included significantcognitive therapy designed to lower patient fear as well assupervised physical therapy averaging 25 hours per weekfor 8 weeks. Because of the intensity of the program, mostpatients stayed at the treatment center in patient hotels. This


Fusion for low-back pain

671

TABLE 1Summary of studies involving patients with intractable, chronic low-back pain*

Authors & Year Class Description Comment

Fritzell, et al., 2001 I RCT of fusion vs conservative care in patients w/ CLBP. All Lumbar fusion is an effective treat-3 fusion groups fared better on all outcome measures. ment for CLBP in patients who fail

conservative measures.Brox, et al., 2003 III RCT of intensive physiotherapy vs instrumented PLF in pa- Intensive physiotherapy may be effec-

tients w/ CLBP. No significant difference noted in pri- tive alternative to instrumented PLF mary outcome measure; however, patients in op group did in patients w/ CLBP who fail con-better in terms of leg & back pain. Patients in physiother- servative measures.apy group did better on psychological assessments & in flexibility testing. Small sample size precludes definitive conclusions.

* CLBP = chronic low-back pain.


intensive course was followed by a home program of theexercises prescribed in the supervised portion. In addition,patients in the physical therapy group were offered individ-ual consultations, lessons, group therapy sessions, and par-ticipation in peer-led discussion groups.

One hundred twenty-one consecutive patients were eval-uated and 57 were excluded for a variety of reasons. Six-ty-four patients were randomized, 37 to surgery and 27 tophysical therapy. There were more men randomized to thesurgical group; otherwise, the groups were comparable. The1-year follow-up rate was 97%. Both groups improved sig-nificantly from baseline on all outcome measures. The im-provement in the primary outcome measure, the modifiedODI, in the surgical group was 15.6 and the improvement inthe physical therapy group was 13.3. The authors found anuncorrected “mean difference in change” of 2.3 points fa-voring patients in the surgical group. This difference in-creased to 2.7 when corrections for preoperative variables(such as sex) were made. There were very large confidenceintervals noted in all outcome measures assessed. The dif-ference in the degree of improvement on the modified ODIbetween the surgical and physical therapy group was notfound to be statistically significant. The large confidenceintervals noted are due to the small number of patients en-rolled in the study. Therefore, the study is almost certainlyunderpowered to detect a significant effect if one in factexists. The results obtained using secondary outcome mea-sures were also reported. In the patients in the surgical groupstatistically significantly more relief of lower-limb pain,more improvement in back pain, less emotional distress, andhigher overall success ratings were observed, as measuredby both the patient and the independent observer. The phys-ical therapy group scored better on fear avoidance as well asin fingertip–floor distance. Nonsignificant trends were alsoseen in favor of the physical therapy group in terms of theGFS and life satisfaction score.2

The authors concluded that there was equivalence be-tween their program of physical therapy and lumbar fu-sion in the treatment of low-back pain. Given the smallsize of the study groups, the absence of a true controlgroup, and the very large confidence intervals, the resultsare considered to provide Class III medical evidence con-cerning the relative efficacy of lumbar fusion comparedwith intensive physical and cognitive therapy. The factthat the surgical group in the study by Brox, et al.,2 tend-ed to do better on the primary outcome measure couldalternatively be interpreted as Class III medical evidencesupporting the use of lumbar fusion, rather than the vigor-ous course of physical therapy described by the authors.

No other randomized studies were identified that pro-vided a direct comparison between the surgical and non-surgical management of patients with low-back pain.Numerous case series and comparisons of surgical tech-niques have been published that provide supporting ClassIII medical evidence for the use of various types of lum-bar fusion in the treatment of patients with low-back painrecalcitrant to medical management. These papers are dis-cussed in detail elsewhere.

Summary

Class I medical evidence exists in support of the use of

lumbar fusion as a treatment standard for carefully selectedpatients with low-back pain intractable to the best medicalmanagement. There is Class III medical evidence that sug-gests that a course of intensive cognitive and physical ther-apy may be an efficacious treatment option for the treat-ment of patients with chronic disabling low-back pain.

Key Issues for Further Research

Multiple modalities may be applied in the treatment oflow-back pain. To demonstrate convincingly the efficacy ofany treatment modality, surgical or other, an appropriatecontrol group must be compared with the treatment group.The size of the study must allow for adequate statisticalpower to minimize the possibility of a beta error. In thestudy of lumbar fusion, it has been demonstrated that sam-ple sizes of 50 to 70 patients per group were adequate todifferentiate treated patients from conservatively managedcontrols. In comparing techniques thought to be moreequivalent (such as different types of efficacious treat-ments), larger numbers of cases are required to demonstratea difference between groups. Researchers are cautionedthat in the context of small sample size, the absence of astatistically significant benefit does not necessarily indicatethe absence of a clinically relevant benefit.

References

1. Brox IJ, Friis A, Holm I: [Patients with chronic degenerativespinal disease: Can conservative treatment reduce the waitinglist for surgery?] Tidsskr Nor Logeforen 119:1784–1787,1999 (Nor)

2. Brox IJ, Sorenson R, Friis A, et al: Randomized clinical trial oflumbar instrumented fusion and cognitive intervention and ex-ercises in patients with chronic low back pain and disc degen-eration. Spine 28:1913–1921, 2003

3. Fritzell P, Hagg O, Wessberg P, et al: 2001 Volvo Award Win-ner in Clinical Studies: Lumbar fusion versus nonsurgical treat-ment for chronic low back pain: a multicenter randomized con-trolled trial from the Swedish Lumbar Spine Study Group.Spine 26:2521–2534, 2001

4. Gibson JNA, Grant IC, Waddell G: The Cochrane review ofsurgery for lumbar disc prolapse and degenerative lumbar spon-dylosis. Spine 24:1820–1832, 1999

5. Hagg O, Fritzell P, Romberg K: The General Function Score: auseful tool for the measurement of physical disability. Validityand reliability. Eur Spine J 10:203–210, 2001

6. Holm I, Friis A, Storheim K, et al: Measuring self-reportedfunctional status and pain in patients with chronic low backpain by postal questionnaires: a reliability study. Spine 28:828–833, 2003

7. Mooney V: Point of view. Spine 26:2532–2533, 2001






Recommendations



Options. 1) Lumbar spinal fusion is not recommendedas routine treatment following primary disc excision inpatients with a herniated lumbar disc causing radiculopa-thy. 2) Lumbar spinal fusion is recommended as a poten-tial surgical adjunct in patients with a herniated disc inwhom there is evidence of preoperative lumbar spinaldeformity or instability. 3) Lumbar spinal fusion is rec-ommended as a potential surgical adjunct in patients withsignificant chronic axial low-back pain associated withradiculopathy due to a herniated lumbar disc. 4) Reopera-tive discectomy is recommended as a treatment option inpatients with a recurrent lumbar disc herniation. 5) Re-operative discectomy combined with fusion is recom-mended as a treatment option in patients with a recurrentdisc herniation associated with lumbar instability, defor-mity, or chronic axial low-back pain.

Rationale

Spinal fusion is a commonly performed procedure, often

conducted following a decompressive procedure. In cases oflumbar disc herniation, the primary problem is usually lim-ited to radicular pain due to nerve compression. Typical-ly, patients with a symptomatic herniated disc refractory tomedical management undergo discectomy without fusion.Spinal fusion has, however, been used as a treatment for pa-tients with primary and recurrent disc herniations. The pur-pose of this review is to examine the medical evidence con-cerning the role of lumbar fusion in the operative treatmentof patients with radiculopathy and back pain caused by aherniated lumbar intervertebral disc.

Search Criteria

A computerized search of the database of the NationalLibrary of Medicine from 1966 to November 2003 wasconducted using the search terms “spinal fusion and discherniation,” “lumbar disc herniation and surgery and out-come,” and “lumbar disc herniation and fusion.” Thesearch was restricted to the English language. This yield-ed a total of 389 references. The titles and abstracts ofeach of these references were reviewed, and papers notconcerned with the use of fusion with lumbar disc hernia-tions were discarded. References were identified that pro-vided either direct or supporting evidence relevant to theuse of fusion as a treatment for lumbar disc herniations.These papers were pulled and reviewed, and relevant ref-erences from their bibliographies were identified. Rele-



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 8: lumbar fusion for disc herniation and radiculopathy


Department of Neurosurgery, University of Wisconsin, Madison, Wisconsin; Department ofNeurosurgery, Mount Sinai Medical School, New York, New York; Department of Neurosurgery,University of Washington, Seattle, Washington; Department of Neurosurgery, Indiana University,Indianapolis, Indiana; Departments of Orthopedic Surgery and Neurosurgery, University ofCalifornia at Los Angeles, California; Department of Neurosurgery, University of Alabama atBirmingham, Alabama; Department of Neurosurgery, Emory University, Atlanta, Georgia; Bone and Joint Clinic of Houston, Texas; and Department of Neurosurgery, Brown University, Providence, Rhode Island

KEY WORDS • fusion • lumbar spine • herniated nucleus pulposus • radiculopathy •practice guidelines

673

Abbreviation used in this paper: PLF = posterolateral fusion.


vant papers providing Class III or better evidence are sum-marized in Table 1. Significant supportive data are pro-vided by other references listed in the bibliography.


Patients with a primary disc herniation typically reportradicular pain as their main symptom. In these patients,surgical treatment typically involves a partial discectomyand decompression of the nerve root without the additionof a fusion. The intervertebral disc is a primary stabilizerof the functional spinal unit and decreases the biomechan-ical forces transmitted to the adjacent vertebral endplates.Injury to the intervertebral disc can, potentially, lead tosegmental spinal instability, which may result in chroniclow-back pain. Yorimitsu and colleagues25 reported that74.6% of patients followed for 10 years after discectomysuffered from residual low-back pain. Thirteen percent oftheir patients reported severe low-back pain. Similarly,Loupasis, et al.,18 reported that 28% of patients who theytreated with discectomy continued to complain of signifi-cant back or leg pain 7 to 10 years after surgery. Dvorakand associates6 found that 23% of patients complained of“constant heavy” back pain between 4 and 17 years fol-lowing discectomy. Several authors have evaluated theaddition of fusion at the time of initial discectomy as ameans to improve patient outcome.

Takeshima, et al.,23 performed a prospective study of 95patients they treated with surgery for a primary disc her-niation. Forty-four patients underwent discectomy alone;51 underwent discectomy and fusion. Clinical outcomeswere assessed approximately 7 years following surgeryusing the Japanese Orthopaedic Association system. Pa-tients with a greater than 50% improvement in symptomswere considered to have an excellent or good outcome. In73% of the discectomy-only group an excellent or goodscore was achieved, compared with 82% of the discecto-my plus fusion group. This difference was not statistical-ly significant (p = 0.31). The patients who underwent fu-sion had longer surgical times, greater blood loss, longerhospital stays, and an increased overall treatment cost.There was, however, a lower disc recurrence rate amongpatients who had undergone discectomy plus fusion (0%compared with 11%).23 This study provides Class III med-ical evidence (small sample size) that the routine additionof a noninstrumented PLF does not improve functionaloutcome in patients treated surgically for a lumbar discherniation.

Donceel and Du Bois5 described a series of 3956 patientstreated for a lumbar disc herniation with either discectomyalone (3670 patients) or discectomy and fusion (286 pa-tients). The authors used return to work 1 year followingsurgery as an outcome measure. They found that 70% ofthe discectomy-only group were able to resume their pre-operative work level at the 1-year follow up compared withonly 40% of the discectomy/fusion group. They noted thatthe fusion group tended to have more significant symptomsand more complex preoperative histories. This retrospec-tive review provides Class III medical evidence suggestingthat discectomy combined with fusion does not improveoutcomes in patients compared with discectomy alonewhen treating lumbar disc herniation.

Young26 performed a retrospective review of a largeseries of patients who underwent surgery for a lumbar discherniation at the Mayo clinic. During a 40-year period,450 patients underwent discectomy and noninstrumentedPLF and 555 underwent discectomy alone. Patients werefollowed for a mean of 8 years. Young observed that thefusion group had superior long-term relief of sciatica(73%) and lumbago (68%) compared with the discecto-my-alone group (48% relief of sciatica and 52% relief oflumbago). He reported a 95% patient satisfaction rate inthe fusion group and an 84% satisfaction rate in the disc-ectomy-alone group. Selection criteria cited for the com-bined operation included patients with spondylolisthesis,spondylolysis, localized degenerative arthritis, partial sa-cralization, scoliosis, fractures, facet joint degeneration,six lumbar vertebrae, congenital anomalies, and recurrentdisc herniation.26 The medical evidence provided by thisreport is considered Class III because of the retrospectivenature of the review, dissimilar patient groups, the use ofnonvalidated outcome measures, and high patient dropout.

One proposed rationale for the addition of fusion to aprimary discectomy is the prevention of late-onset insta-bility and associated chronic low-back pain. Kotilainenand Valtonen17 found that 22% of their patients developedclinical and radiographic signs of lumbar spinal instabili-ty following lumbar microdiscectomy. Kotilainen16 per-formed a follow-up examination 5 years later in 39 of thepatients in whom clinical and radiographic instability afterprimary disc excision developed. He concluded that pa-tients who experience instability after lumbar discectomydid not do well, with only 38% of these patients able towork. The author hypothesized that if lumbar instabilitycould be identified preoperatively, or if the surgeon couldidentify patients at risk for the postoperative instability,these patients might be better treated with fusion at thetime of discectomy. In a series of 520 patients with herni-ations treated by discectomy alone during an 18-year peri-od, Cauchoix, et al.,2 observed only 31 patients (5.9%) inwhom signs or symptoms of mechanical lumbar instabili-ty subsequently developed and who required fusion. Al-though the follow-up duration of their cohort was unclear,these authors concluded that lumbar instability followingdiscectomy was rare and thus did not warrant routine fusionat the time of the primary operation. Padua and colleagues21

studied 150 patients who underwent primary lumbar disc-ectomy. Ten to 15 years following surgery, patients wereexamined clinically and radiographically (flexion–exten-sion lateral lumbar x-ray films). Thirty patients displayedradiographic signs of instability, yet only nine patients werebelieved to be symptomatic. These series provide Class IIImedical evidence indicating that postoperative spinal in-stability may occur after lumbar discectomy and that theoccurrence of instability is associated with a greater like-lihood of a poor outcome. The incidence of symptomaticlumbar spinal instability is relatively low.

A second rationale for adjunctive fusion is in the treat-ment of patients with lumbar disc herniation suffering ra-diculopathy and significant axial back pain or patientswho performs heavy, manual labor. Advocates of spinalarthrodesis in these circumstances point out that eventhough there is no evidence of “true” segmental lumbar in-stability, there is often significant lumbar pain or “fa-tigue.” This may prevent the full return to manual labor in





Fusion for disc herniation

675

TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

the

rel

atio

nshi

p of

cli

nica

l an

d ra

diog

raph

ic o

utco

mes

*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

You

ng, 1

962

III

Lar

ge r

etro

spec

tive

stud

y: 4

50 p

atie

nts

disc

ect o

nly

& 5

55 d

isce

ct 1

noni

nstr

umen

ted

PLF.

Fus

ion

grou

p 68

%

Con

clud

ed th

at e

ach

patie

nt m

ust b

e ev

alua

ted

indi

- re

lief

of b

ack

pain

, 73%

of

scia

tica,

dis

cect

gro

up 4

8% r

elie

f of

bac

k pa

in, 5

2% o

f sc

iatic

a on

ly. O

vera

ll po

or

vidu

ally

. For

cas

es w

/ spo

ndyl

olis

thes

is, i

nsta

bili-

resu

lt in

4%

of

com

bine

d gr

oup

& 1

3% d

isce

ct g

roup

. Ove

rall

patie

nt s

atis

fact

ion

favo

red

fusi

on g

roup

(95

%

ty, o

r R

DH

, adv

ocat

ed c

ombi

ned

op.

vs 8

4% in

non

fusi

on g

roup

). A

ll R

DH

s (u

ncle

ar n

o.)

did

bette

r w

/ fus

ion.

Cau

choi

x, e

t al.,

197

8II

IR

etro

spec

tive

pape

r (2

gro

ups)

: 60

patie

nts

w/ r

ecur

rent

sx

afte

r pr

ior

disc

ect;

how

ever

, dra

wn

from

mix

ed s

ourc

es.

Aut

hors

con

clud

e fr

om th

is la

rge

coho

rt th

at o

nly

No

clea

r in

dica

tions

for

cho

ices

of

op. A

utho

rs a

lso

repo

rt o

n th

eir

own

coho

rt o

f 52

0 pa

tient

s ov

er 1

8 yr

s sm

all p

erce

ntag

e he

rnia

ted

disc

req

uire

fus

ion.

w

/ 5.9

% (

31)

need

ing

reop

(re

peat

dis

cect

, sca

r re

leas

e, r

edo

lam

inot

/fac

etec

t). O

nly

9 (1

.7%

) re

quir

ed f

usio

n Fl

awed

by

no s

tate

men

t of

how

man

y w

ere

fol-

op b

y ra

diog

raph

ic/c

linic

al (

uncl

ear)

cri

teri

a.lo

wed

up

& lo

st to

FU

.E

ie, 1

978

III

Ret

rosp

ectiv

e st

udy

of 2

59 p

atie

nts:

119

dis

cect

onl

y, 6

8 di

scec

t 1in

situ

non

inst

rum

ente

d PL

F. A

t 6 m

os: 8

9%

Con

clud

ed e

quiv

alen

cy a

t ear

ly p

erio

d, b

ut f

usio

n di

scec

t, 88

% f

usio

n sa

tisfi

ed. A

t 6 y

rs: 7

6% v

s 85

% (

not s

igni

fica

nt).

Muc

h hi

gher

pai

n re

curr

ence

in d

isce

ct

prov

ided

mor

e st

able

rel

ief

of p

ain

& m

aint

enan

ce

grou

p in

2–5

yrs

(27

% d

isce

ct, 1

5% f

usio

n). A

bilit

y to

mai

ntai

n w

ork

at p

reop

sta

tus

(79%

dis

cect

, 86%

fu-

of w

ork

func

tion

over

pro

long

ed ti

me.

si

on)

at 6

–7 y

rs. N

o p

valu

es c

ited.

Inou

e, e

t al.,

198

4II

IR

etro

spec

tive

stud

y of

350

pat

ient

s. L

ong

FU o

f 8.

5 yr

s; 6

4–81

% r

elie

f of

pri

mar

y sx

, 94.

3% f

usio

n ra

te. N

o N

ot a

ble

to d

raw

any

goo

d co

nclu

sion

s ex

cept

AL

IF

com

pari

son

arm

of

disc

ect,

etc.

Pri

mar

y A

LIF

for

her

niat

ed d

isc

as s

tudy

gro

up.

yiel

ds g

ood

relie

f of

pri

mar

y sx

, whi

ch a

ppea

rsst

able

ove

r tim

e w

/ pro

long

ed F

U.

Mat

suna

ga, e

t al.,

199

3II

IR

etro

spec

tive

stud

y of

82

patie

nts

(mic

rodi

scec

t 30,

per

c di

scec

t 51,

fus

ion

29).

Rou

ghly

sim

ilar

dem

ogra

phic

s M

anua

l lab

orer

s/ac

tive

athl

etes

do

bette

r af

ter

spin

alw

/ slig

htly

sho

rter

sx

in p

erc

disc

ect g

roup

. Unc

lear

tech

niqu

e of

pre

op “

inst

abili

ty”

asse

ssm

ent t

o de

term

ine

fusi

on. U

ncle

ar s

elec

tion

crite

ria

for

this

. Lon

ger

need

for

op.

3–7

-yr

FU. R

etur

n to

wor

k at

1 y

r (7

5% d

isce

ct b

ut 2

2% c

ould

not

sus

tain

wor

k [5

3% in

end

],

reco

very

aft

er s

pina

l fus

ion.

89%

in s

pina

l fus

ion

grou

p, 5

8% p

erc

disc

ect)

. Tim

e to

ret

urn

to w

ork

(9 w

ks p

erc

disc

ect,

15 w

ks m

icro

disc

-ec

t, 25

wks

fus

ion)

. Con

cept

of

“lum

bar

fatig

ue”

as in

abili

ty o

f m

any

disc

ect p

atie

nts

to r

etur

n to

for

mer

lin

e of

wor

k.

Don

ceel

& D

u B

ois,

199

8II

IR

etro

spec

tive—

very

larg

e; 3

956

case

s (1

26 p

erc,

354

4 m

icro

disc

, 286

dis

cect

1fu

sion

: p ,

0.00

01).

1 yr

FU

. L

arge

coh

ort w

/ cle

ar p

oore

r ou

tcom

es w

/ fus

ion

Fi

tnes

s to

res

ume

wor

k at

1 y

r (7

0% b

oth

disc

ects

, 40%

fus

ion

grou

p); p

oore

st o

vera

ll ou

tcom

es in

fus

ion

grou

pal

thou

gh n

o cl

ear

op in

dica

tions

; no

diff

eren

ces

(ten

ded

to h

ave

mor

e co

mpl

ex h

isto

ries

& lo

nger

dur

atio

n).

btw

n di

scec

t tec

hniq

ues.

Kot

ilain

en, 1

998

III

39 p

atie

nts

w/ c

linic

allu

mba

r in

stab

ility

aft

er m

icro

disc

ect w

ere

eval

uate

d fo

r th

eir

long

-ter

m o

utco

me.

The

re

Thi

s is

a g

roup

of

patie

nts

that

had

inst

abili

ty a

fter

w

ere

21 (

54%

) m

ales

& 1

8 (4

6%)

fem

ales

(m

ean

age

55 y

rs).

All

had

unde

rgon

e op

for

a v

irgi

n si

ngle

-lev

elpr

imar

y m

icro

disc

. Onl

y 38

% w

ere

able

to

lum

bar

disc

her

niat

ion

btw

n 19

85–1

989

& w

ere

eval

uate

d fo

r lu

mba

r in

stab

ility

in 1

991.

Sig

ns &

sxs

of

seg-

wor

k &

5%

wen

t on

to f

usio

n in

the

shor

t FU

. m

enta

l ins

tabi

lity

wer

e th

en d

etec

ted

in a

ll pa

tient

s, w

/ the

sym

ptom

of

“app

rehe

nsio

n” p

ositi

ve in

30.

Dur

ing

The

se p

atie

nts

did

not d

o w

ell.

Iden

tifyi

ng p

atie

nts

the

FU, 2

(5%

) ha

d un

derg

one

lum

bar

spon

dylo

desi

s. A

t the

pre

sent

inve

stig

atio

n, b

oth

gave

the

info

rmat

ion

that

w

ho m

ay d

evel

op in

stab

ility

is e

ssen

tial.

thei

r L

BP

& s

ciat

ica

had

dim

inis

hed

com

pare

d w

/ the

pre

disc

ect;

both

wer

e re

tired

. The

sx

of “

appr

ehen

sion

” T

hose

w/o

inst

abili

ty a

re n

ot d

iscu

ssed

in th

is

was

neg

ativ

e in

bot

h. O

f th

e re

mai

ning

37,

LB

Pha

d co

mpl

etel

y re

cove

red

in 4

(11

%)

& d

imin

ishe

d in

23

(62%

),st

udy.

whe

reas

in 9

(24

%),

LB

Pha

d re

mai

ned

unch

ange

d &

bec

ome

wor

se in

1 (

3%).

Scia

tica

had

com

plet

ely

reco

vere

d in

4 (

11%

) &

dim

inis

hed

in 2

3 (6

2%),

whe

reas

in 7

(19

%),

scia

tica

had

rem

aine

dun

chan

ged

& b

ecam

e w

orse

in3

(8%

). O

nly

14 (

38%

) of

thes

e w

ere

able

to w

ork;

how

ever

, bas

ed o

n th

e O

DI,

the

over

all o

utco

me

in A

DL

had

sign

ific

antly

impr

oved

in a

ll 37

sin

ce 1

991

(p =

0.0

1). T

he s

x of

app

rehe

nsio

n w

as n

ow p

ositi

ve in

26.

Asi

gnif

i-ca

nt c

orre

latio

n w

as o

bser

ved

btw

n th

e po

sitiv

ity o

f th

is te

st &

per

sist

ence

of

LB

P(p

= 0

.02)

& a

poo

r ou

tcom

ein

AD

L(p

,0.

0001

). C

omfi

rmin

g ea

rlie

r ob

serv

atio

ns, t

he f

indi

ngs

supp

ort t

he c

once

pt th

at p

atie

nts

w/ p

osto

p lu

mba

r in

stab

ility

hav

e a

poor

pro

gnos

is. F

urth

er s

tudi

es a

rene

eded

to d

efin

e th

e op

timal

trea

tmen

t for

this

pro

b-le

mat

ic g

roup

.E

ule,

et a

l., 1

999

III

Patie

nts

w/ L

BP

& s

teno

sis

or M

DH

s un

derw

ent d

ecom

pres

sion

w/o

fus

ion.

152

pat

ient

s w

/ lum

bar

spin

al s

ten-

Thi

s is

a m

ixed

pop

ulat

ion

of p

atie

nts

w/ s

teno

sis

&/o

ros

is o

r M

DH

s w

ere

surg

ical

ly tr

eate

d w

/ bila

t par

tial l

amin

ect.

Shor

t-te

rm F

U w

as a

vaila

ble

for

138

patie

nts

(2

MD

Hs

w/ L

BP

that

res

pond

ed w

ell t

o de

com

pres

-w

ks–2

mos

) an

d lo

ng-t

erm

FU

was

ava

ilabl

e fo

r 88

pat

ient

s (1

–6 y

rs, m

ean

3.5

yrs)

. Clin

ical

out

com

e w

as d

e-si

on a

lone

. For

LB

P, s

impl

e de

com

pres

sion

w/o

term

ined

by

char

t rev

iew

& s

tand

ardi

zed

ques

tionn

aire

. Pre

op &

pos

top

back

& le

g pa

in, a

mbu

latio

n, e

mpl

oym

ent

fusi

on f

or M

DH

s ha

s re

ason

ably

goo

d re

sults

. For

stat

us, &

sat

isfa

ctio

n w

ere

asse

ssed

. Ove

rall

impr

ovem

ent w

as n

oted

in 8

8% o

f th

e sp

inal

ste

nosi

s pa

tient

s &

prim

ary

disc

her

niat

ions

, thi

s w

ould

arg

ue a

gain

st91

% o

f th

e M

DH

s at

long

-ter

m F

U. M

ean

hosp

ital s

tays

for

the

spin

al s

teno

sis

grou

p &

the

MD

H g

roup

wer

efu

sion

.3.

7 &

2.8

day

s, r

espe

ctiv

ely.

Onl

y 2

patie

nts

had

unde

rgon

e su

bseq

uent

lum

bar

fusi

ons

afte

r bi

lat p

artia

l lam

-in

ect.

Con

tinu

ed




TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Take

shim

a, e

t al.,

200

0II

IA

pros

pect

ive

stud

y ev

alua

ting

clin

ical

& r

adio

grap

hic

resu

lts in

95

patie

nts

w/ l

umba

r D

H. T

o ev

alua

te r

esul

ts o

fPr

ospe

ctiv

e st

udy

of d

isce

ct v

s di

scec

t & f

usio

n fo

r

disc

ect,

w/ &

w/o

PL

F, 4

4 un

derw

ent d

isce

ct, &

51

unde

rwen

t dis

cect

& f

usio

n. S

ympt

oms

wer

e ev

alua

ted

prim

ary

DH

s. A

lthou

gh th

e re

sults

wer

e be

tter

inus

ing

the

JOA

syst

em. C

linic

al o

utco

me

was

exc

elle

nt o

r go

od in

73%

of

the

nonf

usio

n gr

oup

& 8

2% in

the

the

fusi

on g

roup

, thi

s di

ffer

ence

was

not

sta

tistic

al-

fusi

on g

roup

(p

= 0

.31)

. The

pos

top

redu

ctio

n in

LB

Ppo

stop

was

.in

the

fusi

on g

roup

. The

rat

e of

RD

H a

t ly

sig

nifi

cant

. The

re w

as le

ss b

ack

pain

& le

ss

the

op le

vel i

n th

e no

nfus

ion

grou

p in

crea

sed,

but

intr

aop

bloo

d lo

ss, o

p tim

e, L

OS,

& to

tal c

ost o

f op

wer

e re

curr

ence

in th

e fu

sion

gro

up, b

ut th

e fu

sion

gro

upal

l sig

nifi

cant

ly le

ss in

the

disc

ect-

alon

e th

an in

the

fusi

on g

roup

. The

rad

iolo

gica

l ana

lysi

s re

veal

ed th

at th

e ha

d in

crea

sed

intr

aop

bloo

d lo

ss, o

p tim

e, L

OS,

&

DH

at t

he le

vel o

f di

scec

t & P

LF

in th

e fu

sion

gro

up d

ecre

ased

with

tim

e, a

s in

the

nonf

usio

n gr

oup.

Cha

nges

to

tal c

ost o

f pr

oced

ure.

in D

H &

spi

nal m

otio

n w

ere

not r

elat

ed to

the

clin

ical

res

ults

. Alth

ough

ther

e is

con

trov

ersy

reg

ardi

ng th

e pr

os

& c

ons

of f

usio

n in

ass

ocia

tion

w/ d

isce

ct, t

here

is s

eldo

m a

n in

dica

tion

for

prim

ary

fusi

on f

or lu

mba

r D

H.

Chi

tnav

is, e

t al.,

200

1II

IA

utho

rs f

ollo

wed

thei

r 1s

t 50

patie

nts

for

a m

ax o

f 5

yrs

& a

min

of

6 m

os a

fter

impl

ant o

f th

e C

FCs.

Pat

ient

s Pa

tient

s w

/ RD

Hs

alon

e on

ly w

ere

excl

uded

. Pat

ient

s

in w

hom

MR

I de

mon

stra

ted

“sim

ple”

rec

urre

nt h

erni

atio

n di

d no

t und

ergo

PL

IF. O

p w

as p

erfo

rmed

in p

atie

nts

w/ d

egen

erat

ive

chan

ges

& L

BP

wer

e tr

eate

d su

c-w

/ sx

of n

eura

l com

pres

sion

, ten

sion

sig

ns, &

bac

k pa

in w

/ foc

al d

isc

dege

nera

tion

& n

erve

roo

t dis

tort

ion

cess

fully

w/ p

oste

rior

fus

ion

& d

id w

ell.

Thi

s is

depi

cted

on

MR

Ico

mpa

tible

w/ c

linic

al s

igns

. In

40 p

atie

nts

(80%

) PS

s w

ere

not u

sed.

Clin

ical

out

com

e w

as

agai

n a

good

trea

tmen

t opt

ion

w/ p

atie

nts

w/ L

BP

asse

ssed

usi

ng th

e Pr

olo

Eco

nom

ic S

cale

; fus

ion

outc

ome

was

ass

esse

d us

ing

an e

stab

lishe

d cl

assi

fica

tion.

&

RD

Hs.

Sx in

46

patie

nts

(92%

) im

prov

ed, &

giv

en th

eir

outc

omes

45

(90%

) w

ould

und

ergo

the

sam

e op

aga

in. 2

/3

of p

atie

nts

expe

rien

ced

good

or

exce

llent

out

com

es (

Prol

o sc

ore

$8)

at e

arly

and

late

FU

. No

diff

eren

ce in

clin

ical

out

com

e bt

wn

thos

e w

/ PSs

& th

ose

w/o

(p

= 0

.83,

Man

n–W

hitn

ey U

-tes

t). T

he f

usio

n ra

te a

t 2 y

rsw

as 9

5%. T

here

wer

e m

in c

ompl

icat

ions

, & n

o pa

tient

s fa

red

wor

se a

fter

sur

gery

. No

addi

tiona

l op

trea

tmen

tof

the

fuse

d in

terv

erte

bral

spa

ce.

Vis

hteh

& D

ickm

an, 2

001

III

To d

emon

stra

te th

e fe

asib

ility

of

ant l

umba

r m

icro

disc

ect i

n pa

tient

s w

/ rec

urre

nt, s

eque

ster

ed lu

mba

r D

Hs.

Btw

n 6

patie

nts

w/ s

eque

ster

ed R

HD

s un

derw

ent A

LIF

19

97 &

199

9, 6

pat

ient

s un

derw

ent a

mus

cle-

spar

ing

“min

ilapa

roto

my”

& s

ubse

quen

t mic

rosc

opic

ant

lum

bar

w/ t

itani

um o

r TA

BD

s. T

he m

ain

prin

cipl

e is

that

mic

rodi

sc &

fra

gmen

tect

omy

for

recu

rren

t lum

bar

disc

ext

rusi

ons

at L

5–S1

(4)

or

L4–

5 (2

). A

cont

rala

t dis

-an

t dis

cect

s ar

e su

ffic

ient

for

neu

ral d

ecom

pres

-tr

actio

n pl

ug p

erm

itted

ipsi

lat d

isce

ct u

nder

mic

rosc

ope.

Ext

rude

d di

sc f

ragm

ents

wer

e ex

cise

d by

ope

ning

the

sion

.PL

L. I

nter

body

fus

ion

was

per

form

ed b

y pl

acin

g cy

lindr

ical

TT

Cs

(4)

or T

AB

Ds

(2).

The

re w

ere

no c

ompl

i-ca

tions

, & b

lood

loss

was

min

. FU

MR

I re

veal

ed c

ompl

ete

rese

ctio

nof

all

hern

iate

d di

sc m

ater

ial.

Plai

n fi

lms

reve

aled

exc

elle

nt in

terb

ody

cage

pos

ition

. Rad

icul

ar p

ain

& n

euro

logi

cal d

efic

its r

esol

ved

in a

ll 6

patie

nts

(mea

n FU

14

mos

). A

nt lu

mba

r m

icro

disc

ect w

/ int

erbo

dy f

usio

npr

ovid

es a

via

ble

alte

rnat

ive

for

lum

bar

DH

s.H

uang

& C

hen,

200

3II

I28

pat

ient

s w

/ RD

Hs

w/ L

BP

& d

egen

erat

ive

spon

dylo

listh

esis

und

erw

ent d

ecom

pres

sive

PL

F (s

ingl

e T

TC

& P

S Fo

r R

DH

s w

/ evi

denc

e of

spo

ndyl

olis

thes

is, f

usio

nsu

pple

men

tatio

n). F

U p

erio

d w

as 8

–39

mos

(m

ean

14.4

mos

). C

linic

al o

utco

mes

wer

e as

sess

ed. D

ynam

ic r

a-yi

elds

goo

d re

sults

.di

ogra

phs

for

fusi

on m

ass

inte

rpre

ted

by a

n in

depe

nden

t rad

iolo

gist

. Ove

rall,

92.

86%

wer

e sa

tisfi

ed w

/ the

ir

post

op c

ondi

tions

. Rad

iogr

aphi

c fu

sion

rat

e w

as 8

2.14

%. F

ibro

us u

nion

was

not

ed in

5. N

o ca

ge m

igra

tion

was

obs

erve

d. 1

cas

e of

dur

al l

acer

atio

n w

/o c

linic

al s

eque

lae,

1 p

atie

nt w

/ tra

nsie

nt p

ares

thes

ia r

ecov

ered

w/in

2

wks

, 1 w

/ tra

nsie

nt b

ladd

er a

tony

reco

vere

d w

/in 3

day

s. O

vera

ll, c

ompl

icat

ions

wer

e ne

glig

ible

& n

o pa

tient

ssu

stai

ned

a m

otor

def

icit

or p

erm

anen

t com

plic

atio

n. T

his

PLIF

res

ulte

d in

sat

isfa

ctor

y ou

tcom

e w

/in s

hort

-te

rm o

r lo

ng-t

erm

FU

.

*A

DL

= a

ctiv

ities

of

daily

livi

ng; A

LIF

= a

nter

ior

lum

bar

inte

rbod

y fu

sion

; CFC

= c

arbo

n fi

ber

cage

; DH

= d

isc

hern

iatio

n; d

isce

ct =

dis

cect

omy;

fac

etec

t = f

acet

ecto

my;

FU

= f

ollo

w u

p; J

OA

= J

apan

ese

Ort

hopa

edic

Ass

ocia

tion;

lam

inec

t = la

min

ecto

my;

LB

P=

low

-bac

k pa

in; L

OS

= le

ngth

of

stay

; MD

H =

mid

line

DH

; MR

I =

mag

netic

res

onan

ce im

agin

g; O

DI

= O

swes

try

Dis

abili

ty I

ndex

; per

c =

per

cu-

tane

ous;

PL

IF =

pos

teri

or L

IF; P

LL

= p

oste

rior

long

itudi

nal l

igam

ent;

PS =

ped

icle

scr

ew; R

DH

= r

ecur

rent

DH

; sx

= s

ympt

om(s

); T

AB

D =

thre

aded

allo

graf

t bon

e do

wel

; TT

C =

thre

aded

tita

nium

cag

e.


a large portion of patients despite treatment with discecto-my alone. Inoue, et al.,14 reported the use of anterior lum-bar discectomy and primary fusion in the treatment of alumbar disc herniation in 350 patients followed for longerthan 8.5 years. They observed a 75% good outcome ratewith regard to primary symptoms. They concluded thatfusion provides sustained improvement in these selectpatients compared with historical results of series of sim-ilar patients treated with discectomy alone. Eie8 examined259 patients with a herniated disc who underwent one oftwo treatments: discectomy alone (119) or discectomy andnoninstrumented PLF (68); the author observed equiva-lent rates of good outcome between the two treatmentgroups during the first few months after surgery (89 and88%, respectively). The author, however, found that thesatisfaction rates reported by the discectomy-alone groupdeteriorated over time compared with the satisfaction ratereported by the discectomy/fusion group. At 6 years post-surgery, 76% of the discectomy-alone group reported sat-isfaction compared with 85% of the discectomy/fusiongroup. Additionally the discectomy-only patients reporteda significantly higher incidence of pain recurrence (27%of cases) compared with the discectomy/fusion group(15% of patients) (p = 0.001). Manual laborers with sig-nificant preoperative axial back pain were especially like-ly to suffer recurrences of pain when treated with disc-ectomy alone. Seventy-nine percent of the discectomypatients and 86% of the discectomy/fusion patients main-tained their preoperative work status at the 6-year followup.8 These papers provide Class III medical evidence insupport of the use of fusion at the time of discectomy,especially in manual laborers or those with significant pre-operative low-back pain.

Matsunaga, et al.,19 performed a retrospective study of80 manual laborers and athletes who underwent eitheropen or percutaneous discectomy (51) or via open discec-tomy combined with fusion (29). Their primary outcomemeasure was return to work or participation in athletics.At 1 year they observed that 54% of the discectomy groupand 89% of the discectomy/fusion group were able toreturn to and maintain preoperative work or athletic activ-ities. They found that although discectomy-treated pa-tients returned to work earlier (12 weeks) than the discec-tomy/fusion-treated group (25 weeks), 22% of the formercould not maintain their previous activity level because ofso-called lumbar fatigue. These authors concluded that theaddition of fusion should be considered in manual labor-ers and active athletes because it appeared to provide abetter chance of return to and maintenance of a preopera-tive level of function.19 This paper is considered to provideClass III medical evidence in support of the use of PLF atthe time of discectomy for patients involved in heavy la-bor and athletics.

No study was identified that compared outcomes afterreoperative discectomy compared with reoperative discec-tomy/fusion in patients with a recurrent disc herniation.There are a number of case series describing outcomesafter either reoperative discectomy or reoperative discec-tomy combined with fusion. For example, Suk, et al.,22

reported that patient outcomes following reoperative dis-cectomy were satisfactory and similar to those in patientstreated with primary disc excision. Cinotti, et al.,4 report-ed an overall good outcome rate of 85% and a return-to-

work rate of 81% among 26 patients who had undergonereoperative discectomy. Their results were similar to thoseidentified among a cohort of 50 other patients who under-went primary disc excision during the same time period(88% good outcome and 84% return to work). Ozgen, etal.,20 performed reoperative decompressions in 114 pa-tients, including reoperative discectomies in 89 patientswith a recurrent disc herniation. Good outcomes weredemonstrated in 69% of patients. Several other authorsdescribe similar findings.1,7,11,12,15 These series indicate thatpatients improve following reoperative discectomy forrecurrent disc herniation and provide Class III medicalevidence in support of this practice.

Other authors have described the results of reoperativedecompression and supplemental fusion for patients withrecurrent lumbar disc herniation. Glassman and col-leagues10 used the 36-item Short Form to perform a pro-spective study of patients with recurrent herniated discsundergoing reoperative discectomy and fusion. They des-cribed significant improvement in physical function, socialfunction, and bodily pain 1 year after surgery. Huang andChen13 described a series of 28 patients with a recurrentdisc herniation who also experienced low-back pain andspondylolisthesis, who underwent posterior decompressivesurgery and interbody fusion. During a follow-up periodranging from 8 to 39 months (mean 14 months), 93% of thepatients were satisfied with their condition and 82% wereconsidered to have achieved radiographic fusion. Chit-navis and associates3 also reported on patients with recur-rent disc herniations and symptoms of back pain or signsof instability who underwent posterior decompression andinterbody fusion. Patients with recurrent disc herniationwithout low-back pain or instability were excluded. Of the50 patients with 6 months to 5 years of follow-up data,92% improved and 90% were very satisfied with their out-come. The fusion rate was 95% and the complication ratewas low. Vishteh and Dickman24 presented a small seriesof five patients with recurrent sequestered disc herniationstreated by anterior lumbar discectomy and interbody fu-sion alone. The results were very good in all five patients:100% fusion rate and relief of leg pain. These series pro-vide Class III medical evidence in support of performinga fusion at the time of reoperative discectomy, particular-ly in patients with associated deformity, instability, orchronic axial back pain.

Summary

There is no convincing medical evidence to support theroutine use of lumbar fusion at the time of a primary lum-bar disc excision. There is conflicting Class III medicalevidence regarding the potential benefit of the addition offusion in this circumstance. Therefore, the definite in-crease in cost and complications associated with the use offusion are not justified. Patients with preoperative lumbarinstability may benefit from fusion at the time of lumbardiscectomy; however, the incidence of such instabilityappears to be very low (, 5%) in the general lumbar discherniation population. Patients who suffer from chroniclow-back pain, or are heavy laborers or athletes with axiallow-back pain, in addition to radicular symptoms may al-so be candidates for fusion at the time of lumbar disc exci-sion. Patients with a recurrent disc herniation have been


Fusion for disc herniation

677


treated successfully with both reoperative discectomy andreoperative discectomy combined with fusion. In patientswith a recurrent lumbar disc herniation with associatedspinal deformity, instability, or associated chronic low-back pain, consideration of fusion in addition to reopera-tive discectomy is recommended.


A direct comparison between reoperative discectomyand reoperative discectomy/fusion would provide neededinformation regarding the role for fusion in this patientpopulation. A detailed analysis of patients who do and donot suffer from postoperative chronic low-back pain (casecontrol study) could potentially provide Class II medicalevidence to support the use of fusion in the subgroup ofpatients in whom there is likely to be a benefit.

References

1. Baba H, Chen Q, Kamitani K, et al: Revision surgery for lum-bar disc herniation. An analysis of 45 patients. Int Orthop 19:98–102, 1995

2. Cauchoix J, Ficat C, Girard B: Repeat surgery after disc exci-sion. Spine 3:256–259, 1978

3. Chitnavis B, Barbagallo G, Selway R, et al: Posterior lumbar in-terbody fusion for revision disc surgery: review of 50 cases inwhich carbon fiber cages were implanted. J Neurosurg (Spine2) 95:190–195, 2001

4. Cinotti G, Roysam GS, Eisenstein SM, et al: Ipsilateral recur-rent lumbar disc herniation. A prospective, controlled study. JBone Joint Surg Br 80:825–832, 1998

5. Donceel P, Du Bois M: Fitness for work after surgery for lum-bar disc herniation: a retrospective study. Eur Spine J 7:29–35,1998

6. Dvorak J, Gauchat MH, Valach L: The outcome of surgery forlumbar disc herniation. I. A 4–17 years’ follow-up with empha-sis on somatic aspects. Spine 13:1418–1422, 1988

7. Ebeling U, Kalbarcyk H, Reulen HJ: Microsurgical reoperationfollowing lumbar disc surgery. Timing, surgical findings, andoutcome in 92 patients. J Neurosurg 70:397–404, 1989

8. Eie N: Comparison of the results in patients operated upon forruptured lumbar discs with and without spinal fusion. ActaNeurochir 41:107–113, 1978

9. Eule JM, Breeze R, Kindt GW: Bilateral partial laminectomy: atreatment for lumbar spinal stenosis and midline disc hernia-tion. Surg Neurol 52:329–338, 1999

10. Glassman SD, Minkow RE, Dimar JR, et al: Effect of prior lumbar discectomy on outcome of lumbar fusion: a prospec-tive analysis using the SF-36 measure. J Spinal Disord 11:383–388, 1998

11. Haglund MM, Moore AJ, Marsh H, et al: Outcome after repeatlumbar microdiscectomy. Br J Neurosurg 9:487–495, 1995

12. Herron L: Recurrent lumbar disc herniation: results of repeat laminectomy and discectomy. J Spinal Disord 7:161–166, 1994

13. Huang KF, Chen TY: Clinical results of a single central inter-body fusion cage and transpedicle screws fixation for recurrentherniated lumbar disc and low-grade spondylolisthesis. ChangGung Med J 26:170–177, 2003

14. Inoue S, Watanabe T, Hirose A, et al: Anterior discectomy andinterbody fusion for lumbar disc herniation. A review of 350cases. Clin Orthop Relat Res 183:22–31, 1984

15. Jonsson B, Stromqvist B: Repeat decompression of lumbarnerve roots. A prospective two-year evaluation. J Bone JointSurg Br 75:894–897, 1993

16. Kotilainen E: Long-term outcome of patients suffering fromclinical instability after microsurgical treatment of lumbar discherniation. Acta Neurochir 140:120–125, 1998

17. Kotilainen E, Valtonen S: Clinical instability of the lumbarspine after microdiscectomy. Acta Neurochir 125:120–126,1993

18. Loupasis GA, Stamos K, Katonis PG, et al: Seven-to 20-yearoutcome of lumbar discectomy. Spine 24:2313–2317, 1999

19. Matsunaga S, Sakou T, Taketomi E, et al: Comparison of oper-ative results of lumbar disc herniation in manual laborers andathletes. Spine 18:2222–2226, 1993

20. Ozgen S, Naderi S, Ozek MM, et al: Findings and outcome ofrevision lumbar disc surgery. J Spinal Disord 12:287–292,1999

21. Padua R, Padua S, Romanini E, et al: Ten- to 15-year outcomeof surgery for lumbar disc herniation: radiographic instabilityand clinical findings. Eur Spine J 8:70–74, 1999

22. Suk KS, Lee HM, Moon SH, et al: Recurrent lumbar disc her-niation: results of operative management. Spine 26:672–676,2001

23. Takeshima T, Kambara K, Miyata S, et al: Clinical and radio-graphic evaluation of disc excision for lumbar disc herniationwith and without posterolateral fusion. Spine 25:450–456,2000

24. Vishteh AG, Dickman CA: Anterior lumbar microdiscectomyand interbody fusion for the treatment of recurrent disc hernia-tion. Neurosurgery 48:334–338, 2001

25. Yorimitsu E, Chiba K, Toyama Y, et al: Long-term outcomes ofstandard discectomy for lumbar disc herniation: a follow-upstudy of more than 10 years. Spine 26:652–657, 2001

26. Young HH: Posterior fusion of vertebrae in treatment for pro-truded intervertebral disk. J Neurosurg 19:314–318, 1962

Manuscript received December 7, 2004..Accepted in final form February 18, 2005.Address reprint requests to: Daniel K. Resnick, M.D., Depart-





Recommendations

Guidelines. The performance of a lumbar PLF is recom-mended for patients with lumbar stenosis and associateddegenerative spondylolisthesis who require decompres-sion. There is insufficient evidence to recommend a treat-ment guideline.

Options. Pedicle screw fixation as an adjunct to lumbarPLF should be considered as a treatment option in patientswith lumbar stenosis and spondylolisthesis in cases inwhich there is preoperative evidence of spinal instabilityor kyphosis at the level of the spondylolisthesis or wheniatrogenic instability is anticipated.

Rationale

Patients with lumbar stenosis often present with con-comitant degenerative spondylolisthesis. Decompression

alone in this population may result in deformity progres-sion. Lumbar PLF has been used as a means to preventpostoperative deformity progression and to improve func-tional outcome after decompressive surgery in this popu-lation. The purpose of this review is to examine the liter-ature concerning the role of fusion after decompressionsurgery in patients with degenerative spondylolisthesisand stenosis.

Search Criteria

A computerized search of the database of the NationalLibrary of Medicine from 1966 to August 2003 was con-ducted using the search terms “fusion and degenerativespondylolisthesis.” The search was restricted to theEnglish language and yielded 217 references. The titlesand abstracts of each of these references were reviewed,and papers not concerned with the use of fusion with de-generative spondylolisthesis were discarded. Eighty-fivereferences were identified that provided either direct orsupporting evidence relevant to the use of fusion for de-generative lumbar spondylolisthesis. These papers werereviewed, and relevant references from the bibliographieswere identified. All papers providing Class III or bettermedical evidence are summarized in Table 1. Additional



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 9: fusion in patients with stenosis and spondylolisthesis



KEY WORDS • lumbar spine • stenosis • spondylolisthesis • fusion •practice guidelines

679

Abbreviations used in this paper: DPQ = Dallas Pain Question-naire; JOA = Japanese Orthopaedic Association; PLF = posterolat-eral fusion; SF-36 = Short Form–36; VAS = visual analog scale.




TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

lum

bar

inte

rbod

y fu

sion

in

pati

ents

wit

h de

gene

rati

ve l

umba

r di

seas

e*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Dal

l & R

owe,

198

5II

I17

pat

ient

s w

/ DS

had

deco

mpr

essi

ons:

1 gr

oup

w/ l

amin

ect &

fac

etec

t & th

e ot

her

w/ l

amin

ect w

/ for

amin

ot. A

t 20

-mo

FU, 5

/6 w

/ fac

etec

t wer

e be

tter

& o

nly

4/11

w/ f

oram

inot

wer

e be

tter.

Ove

rall

poor

res

ults

led

auth

ors

to r

ecom

men

d ex

tens

ive

deco

mpr

essi

ons

as b

ette

r th

an la

min

ect w

/ jus

t for

amin

ot.

Feff

er, e

t al.,

198

5II

I19

pat

ient

s w

/ DS

unde

rwen

t dec

ompr

essi

on w

/ & w

/o f

usio

n:8

patie

nts

w/ d

ecom

pres

sion

/fus

ion

& 1

1 w

/ de-

Fusi

on-t

reat

ed p

atie

nts

did

bette

r th

an th

ose

rece

iv-

com

pres

sion

alo

ne. O

f th

e 8

w/ f

usio

n, 5

had

goo

d re

sults

, 3 w

ere

fair

, & 0

wer

e po

or. O

f th

e 11

w/o

fus

ion,

in

g de

com

pres

sion

alo

ne.

5 ha

d go

od r

esul

ts, 3

wer

e fa

ir, &

3 w

ere

poor

. 4 p

atie

nts

w/o

fus

ion

suff

ered

gro

ss in

stab

ility

. L

omba

rdi,

et a

l., 1

985

III

47 p

atie

nts

w/ D

S. T

reat

men

t gro

ups

of w

ide

deco

mpr

essi

on w

/ fac

etec

tom

y, d

ecom

pres

sion

, & d

ecom

pres

sion

L

arge

ben

efit

to a

ddin

g fu

sion

to d

ecom

pres

sion

.w

/ fus

ion.

Goo

d-to

-exc

elle

nt r

esul

ts w

ere

33, 8

9, &

90%

, res

pect

ivel

y. T

he a

dditi

on o

f a

fusi

on w

as b

enef

icia

l.K

aned

a, e

t al.,

198

6II

I54

pat

ient

s w

/ uns

tabl

e D

S tr

eate

d by

dec

ompr

essi

on &

fus

ion.

96%

fus

ion

rate

w/ e

xcel

lent

clin

ical

res

ults

.In

oue,

et a

l., 1

988

III

36 p

atie

nts

w/ D

S tr

eate

d w

/ ant

erio

r in

terb

ody

fusi

on. A

ll fu

sed

& d

id w

ell c

linic

ally

. C

ase

seri

es—

no f

orm

al o

utco

mes

mea

sure

s.Fu

jiya,

et a

l., 1

990

III

40 p

atie

nts

w/ u

nsta

ble

DS

unde

rwen

t dec

ompr

essi

on &

fus

ion.

1 p

atie

nt s

uffe

red

a no

nuni

on. A

t a m

ean

FU

of 2

6 m

os, 7

5% h

ad e

xcel

lent

, 20%

goo

d, 5

% f

air,

& 0

% p

oor

resu

lts.

Taka

hash

i, et

al.,

199

0II

I34

pat

ient

s ha

ving

ant

erio

r de

com

pres

sion

& in

terb

ody

fusi

on f

or D

S: 7

6% w

/ sat

isfa

ctor

y re

sults

at 1

0-yr

FU

.Su

rviv

orsh

ip a

naly

sis

of f

usio

ns f

or D

S.H

erko

witz

& K

urz,

199

1I

50 p

atie

nts

w/ D

S &

ste

nosi

s (2

5 w

/ fus

ion

& 2

5 w

/o f

usio

n). I

n a

mea

n FU

of

3 yr

s, p

atie

nts

w/ a

fus

ion

did

Goo

d st

udy

dem

onst

ratin

g di

rect

ben

efit

of f

usio

n be

tter.

com

pare

d w

/ no

fusi

on.

Her

ron,

et a

l., 1

991

III

24 p

atie

nts

w/ D

S ha

ving

dec

ompr

essi

on o

nly

w/o

fus

ion

at m

ean

FU o

f 34

mos

: 20

w/ g

ood

resu

lts, 3

w/ f

air,

Smal

l no.

of

patie

nts

& s

hort

FU

.&

1 w

/ poo

r.Po

stac

chin

i, et

al.,

199

1II

I32

pat

ient

s w

/ DS

trea

ted

w/ d

ecom

pres

sion

& s

ome

w/ v

ario

us f

orm

s of

dec

ompr

essi

on w

/ or

w/o

fus

ion.

Ove

r-Sm

all n

os. s

how

tren

d fo

r be

tter

resu

lts w

/ fus

ion,

al

l, 17

w/ f

usio

n &

15

w/o

. W/ f

usio

n, a

ll 17

had

goo

d or

exc

elle

nt r

esul

ts, w

here

as in

thos

e w

/o f

usio

n, 1

0ha

d bu

t not

sta

tistic

ally

dif

fere

nt.

good

or

exce

llent

res

ults

& 5

had

fai

r or

poo

r re

sults

. Aut

hors

sta

ted

no s

tatis

tical

dif

fere

nce

btw

n fu

sion

&

nonf

usio

n &

rec

omm

ende

d fu

sion

for

uns

tabl

e si

tuat

ions

.C

hang

, et a

l., 1

993

III

85 p

atie

nts

w/ s

pond

ylol

isth

esis

(57

w/ D

S) u

nder

wen

t op.

Pat

ient

s w

/ ins

trum

enta

tion

had

bette

r fu

sion

rat

es &

57

pat

ient

s w

/ DS

had

good

fus

ion-

rela

ted

resu

lts.

clin

ical

out

com

es (

60%

w/ c

ompl

ete

relie

f of

pai

n).

Sand

erso

n &

Woo

d, 1

993

III

31 p

atie

nts

w/ d

ecom

pres

sion

w/o

fus

ion

(19

w/ D

S). O

f th

e po

or r

esul

ts, 3

wer

e in

pat

ient

s w

/ DS.

In s

ubse

t of

19 p

atie

nts

w/ D

S un

derg

oing

dec

om-

pres

sion

alo

ne: 3

/19

w/ p

oor

resu

lts &

16

w/

good

res

ults

.So

ini,

et a

l., 1

993

III

51 p

atie

nts

(13

w/ D

S) h

avin

g de

com

pres

sion

&fu

sion

. Ins

trum

enta

tion

resu

lted

in h

ighe

r fu

sion

s, b

ut a

ll di

d w

ell.

Smal

l ser

ies

of p

atie

nts

doin

g w

ell w

/ fus

ion.

Axe

lsso

n, e

t al.,

199

4II

I71

pat

ient

s w

/ spi

nal f

usio

n (s

ubse

t of

43 w

/ DS)

. 29/

43 h

ad g

ood

resu

lts. I

n th

e pa

tient

s w

/ DS,

a h

eale

d fu

sion

43

pat

ient

s w

/ DS

in w

hom

goo

d ou

tcom

e w

as a

sso-

corr

elat

ed w

/ a g

ood

resu

lt, b

ut n

ot s

o w

/ the

oth

er d

iagn

oses

.ci

ated

w/ h

eale

d fu

sion

.L

ee, 1

994

III

52 p

atie

nts

w/ u

nsta

ble

DS

trea

ted

w/ r

educ

tion

& f

usio

n (n

o fo

rmal

dec

ompr

essi

on).

1-y

r m

in F

U. 8

9% h

ad s

a-U

nsta

ble

DS,

bet

ter

w/ d

ecom

pres

sion

& f

usio

n.

tisfa

ctor

y ou

tcom

es w

/ reg

ard

to b

ack

pain

. The

rem

aini

ng w

/o s

atis

fact

ory

outc

ome

had

PA. F

or le

g pa

in, 9

3%

W/o

suc

cess

ful f

usio

n, u

nsat

isfa

ctor

y re

sults

in

satis

fact

ory

outc

ome.

Rem

aini

ng u

nsat

isfa

ctor

y ha

d go

od o

utco

mes

aft

er f

orm

al d

ecom

pres

sion

.te

rms

of b

ack

pain

.M

arkw

alde

r, et

al.,

199

5II

IPr

ospe

ctiv

e st

udy

of 1

00 p

atie

nts

w/ D

S ha

ving

dec

ompr

essi

on &

fus

ion.

95

patie

nts

w/ g

ood

or e

xcel

lent

res

ults

.G

ood

resu

lts.

Ger

tzbe

in, e

t al.,

199

6II

IC

ircu

mfe

rent

ial f

usio

n fo

r va

riou

s co

nditi

ons.

67

patie

nts

avai

labl

e fo

r 2-

yr F

U. 1

5.2%

had

DS

w/ s

teno

sis.

97%

In

sub

grou

p of

pat

ient

s w

/ DS,

mor

e re

turn

to w

ork;

su

cces

sful

fus

ion.

18%

ret

urne

d to

eith

er li

ghte

r w

ork

or jo

b re

trai

ning

, & 2

3% w

ere

not w

orki

ng. T

hus,

77%

m

ore

favo

rabl

e su

bgro

up in

ove

rall

favo

rabl

e w

ere

perf

orm

ing

the

sam

e or

ligh

ter

leve

ls o

f ac

tivity

& 2

3% w

ere

not w

orki

ng. I

n di

agno

stic

cat

egor

ies,

the

stud

y.no

s. n

ot w

orki

ng w

ere

as f

ollo

ws:

DD

D, 2

5%; P

A, 3

3%; &

SS,

8%

. Pai

n si

gnif

ican

tly r

educ

ed o

n V

AS

from

7.

1–2.

1 (b

ack)

& f

rom

5.8

–1.5

(le

g) (

p ,

0.00

6 &

0.0

001,

res

pect

ivel

y).

Han

ley,

199

6II

I20

pat

ient

s w

/ DS

& s

teno

sis

unde

rgoi

ng d

ecom

pres

sion

& f

usio

n. A

t a m

in 2

-yr

FU, s

atis

fact

ory

resu

lts w

ere

foun

d in

17

(85%

).Fi

schg

rund

, et a

l., 1

997

II76

pat

ient

s w

/ sym

ptom

atic

ste

nosi

s as

soci

ated

w/ l

umba

r D

S w

ere

pros

pect

ivel

y st

udie

d. A

ll un

derw

ent p

oste

-Pr

ospe

ctiv

e st

udy,

lum

bar

DS,

dec

ompr

essi

on &

ri

or d

ecom

pres

sion

& p

oste

rola

t int

ertr

ansv

erse

pro

cess

art

hrod

esis

. Pat

ient

s ra

ndom

ized

to a

seg

men

tal t

rans

-fu

sion

w/ o

r w

/o f

ixat

ion.

Im

prov

emen

t w/

pedi

cula

r in

stru

men

ted

or n

onin

stru

men

ted

grou

p. 6

7 w

ere

avai

labl

e fo

r a

2-yr

FU

. Clin

ical

out

com

e w

as e

x-de

com

pres

sion

& f

usio

n.ce

llent

or

good

in 7

6% w

/ ins

trum

enta

tion

& in

85%

of

thos

e w

/o in

stru

men

tatio

n (p

= 0

.45)

. Suc

cess

ful a

r-th

rode

sis

in 8

2% o

f in

stru

men

ted

vs 4

5% o

f no

nins

trum

ente

d ca

ses

(p =

0.0

015)

. Ove

rall,

suc

cess

ful f

usio

n di

d no

t inf

luen

ce o

utco

me

(p =

0.4

35).

In

patie

nts

unde

rgoi

ng 1

-lev

el P

LF

for

DS

& s

pina

l ste

nosi

s, P

Ss m

ay le

ad

to a

hig

her

fusi

on r

ate,

but

clin

ical

out

com

e sh

ows

no im

prov

emen

t in

back

& lo

wer

-lim

b pa

in. C

linic

al o

ut-

com

e 78

% w

/o in

stru

men

tatio

n &

85%

w/ i

nstr

umen

tatio

n (g

ood

or e

xcel

lent

).K

atz,

et a

l., 1

997

IIPr

ospe

ctiv

e m

ultic

ente

r ob

serv

atio

nal s

tudy

of

272

patie

nts

havi

ng o

p fo

r de

gene

rativ

e lu

mba

r st

enos

is. 1

23 id

en-

Subg

roup

had

DS

or s

colio

sis

& a

non

stat

istic

ally

tif

ied

as h

avin

g D

S or

sco

liosi

s. T

he 1

st s

et o

f an

alys

es u

sed

all p

atie

nts

in th

e co

hort

; the

2nd

was

res

tric

ted

tosi

gnif

ican

t tre

nd o

f be

tter

resu

lts in

pat

ient

s w

/ non

-pa

tient

s w

/ typ

ical

indi

catio

ns f

or a

rthr

odes

is:D

S of

,5

mm

&/o

r sc

olio

sis

of ,

15˚.

Pat

ient

s un

derg

oing

ar-

inst

rum

ente

d fu

sion

. Oth

er o

utco

mes

sim

ilar

btw

nth

rode

sis

wer

e si

gnif

ican

tly y

oung

er &

mos

tly f

emal

e. 9

7% w

ere

cauc

asia

n. T

he 3

gro

ups

had

sim

ilar

leve

lsno

nfus

ion,

fus

ion

w/ i

nstr

umen

tatio

n, &

fus

ion

Con

tinu

ed



Stenosis and spondylolisthesis

681

TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Kat

z, e

t al.,

con

tinu

edof

edu

catio

n. P

atie

nts

who

rec

eive

d ar

thro

desi

s ha

d si

gnif

ican

tly f

ewer

leve

ls d

ecom

pres

sed

& g

reat

er f

requ

ency

w

/o in

stru

men

tatio

n.of

DS.

Of

the

272

patie

nts,

105

(39

%)

met

ent

ry c

rite

ria

for

the

coho

rt r

estr

icte

d to

$ 5

mm

DS

(93

patie

nts)

, &

/or

15˚

scol

iosi

s (1

9 pa

tient

s). 7

pat

ient

s ha

d bo

th D

S .

5 m

m &

sco

liosi

s .

15˚.

In

this

res

tric

ted

coho

rt,

52 p

atie

nts

(50%

) un

derw

ent l

amin

ect w

/o a

rthr

odes

is, 2

6 (2

5%)

noni

nstr

umen

ted

arth

rode

sis,

& 2

7 (2

6%)

inst

rum

ente

d ar

thro

desi

s. O

n av

erag

e, th

ose

not u

nder

goin

g fu

sion

wer

e 8

yrs

olde

r th

an th

ose

unde

rgoi

ng a

rth-

rode

sis

& h

ad s

igni

fica

ntly

mor

e le

vels

dec

ompr

esse

d. O

ther

bas

elin

e va

riab

les

wer

e si

mila

r in

the

3 gr

oups

. U

niva

riat

e an

alys

es s

how

ed th

at a

t 6-m

o FU

, the

non

inst

rum

ente

d gr

oup

had

sign

ific

antly

gre

ater

impr

ovem

ent

in b

ack

pain

(p

= 0

.009

) &

no

sign

ific

ant d

iffe

renc

es in

impr

ovem

ent i

n w

alki

ng, S

IPsc

ore,

or

satis

fact

ion.

The

24

-mo

FU s

how

ed g

reat

er r

educ

tion

in b

ack

pain

am

ong

patie

nts

rece

ivin

g no

nins

trum

ente

d fu

sion

, but

this

tr

end

was

not

sta

tistic

ally

sig

nifi

cant

(p

= 0

.20)

. Oth

er o

utco

mes

wer

e si

mila

r am

ong

the

3 gr

oups

. Lin

ear

regr

essi

on m

odel

s in

this

res

tric

ted

coho

rt s

how

ed n

o si

gnif

ican

t ass

ocia

tion

btw

n tr

eatm

ent g

roup

& im

prov

e-m

ent i

n ba

ck p

ain

or in

wal

king

at 6

& 2

4 m

os. T

he p

val

ues

for

thes

e te

sts

wer

e al

l $ 0

.20.

Tho

mse

n, e

t al.,

199

7II

Pros

pect

ive

rand

omiz

ed s

tudy

of

fusi

on &

dec

ompr

essi

on w

/ & w

/o in

stru

men

tatio

n. 1

30 p

atie

nts

rand

omly

allo

-St

udy

had

subg

roup

of

patie

nts

w/ D

S &

ste

nosi

s,

cate

d to

rec

eive

no

inst

rum

enta

tion

(66)

or

Cot

rel-

Dub

ouss

et in

stru

men

tatio

n (6

4) in

PL

F. V

aria

bles

wer

e re

-bu

t not

all.

gist

ered

at t

he ti

me

of o

p &

at 1

& 2

yrs

pos

top.

No

func

tiona

l dif

fere

nces

btw

n gr

oups

w/ &

w/o

inst

rum

enta

-tio

n, a

lthou

gh f

usio

n ra

te im

prov

ed w

/ ins

trum

enta

tion

(68

vs 8

5%).

Neu

ral c

ompr

essi

on d

id b

ette

r—st

atis

tic-

ally

sig

nifi

cant

.B

ooth

, et a

l., 1

999

III

41 p

atie

nts

w/ D

S un

derw

ent d

ecom

pres

sion

& f

usio

n w

/ min

5-y

r FU

. 36

(88%

) re

turn

ed F

U q

uest

ionn

aire

s.

Lon

g-te

rm F

U o

f su

cces

sful

res

ults

w/ f

usio

n.83

% s

atis

fact

ion

w/ o

p &

86%

w/ s

igni

fica

ntly

impr

oved

bac

k &

leg

pain

. No

nonu

nion

s, b

ut w

orse

out

com

es

w/ a

djac

ent-

leve

l pro

blem

s &

. 4

med

ical

com

orbi

ditie

s.M

ochi

da, e

t al.,

199

9II

I10

2 pa

tient

s w

/ DS

& s

teno

sis

had

deco

mpr

essi

on &

fus

ion

(Gro

up I

: non

rigi

d in

stru

men

tatio

n, 3

3 pa

tient

s;

Goo

d im

prov

emen

t, hi

gher

fus

ion

rate

, hig

her

suc-

Gro

up I

I: r

igid

inst

rum

enta

tion,

34

patie

nts;

Gro

up I

II: n

o in

stru

men

tatio

n, 3

5 pa

tient

s; G

roup

I,

2 no

nuni

ons,

ce

ss r

ate—

rigi

d in

stru

men

tatio

n be

tter.

82%

suc

cess

rat

e; G

roup

II,

1 n

onun

ion,

91%

suc

cess

rat

e; G

roup

III

, 5

nonu

nion

s, 7

1% s

ucce

ss r

ate.

Nor

k, e

t al.,

199

9II

I30

pat

ient

s w

/ DS

had

deco

mpr

essi

on &

fus

ion

w/ S

F-36

out

com

es d

ata.

93%

wer

e sa

tisfi

ed w

/ out

com

es. I

nter

-Pa

tient

s tr

eate

d w

/ dec

ompr

essi

on &

fus

ion

for

DS

vi

ewed

at F

U a

mea

n of

37.

1 m

os p

osto

p (r

ange

24–

55 m

os).

Pat

ient

s im

prov

ed s

igni

fica

ntly

in a

bilit

y to

ha

d im

prov

ed f

unct

iona

l out

com

es w

hen

mea

-pe

rfor

m h

eavy

& li

ght a

ctiv

ities

, par

ticip

ate

in s

ocia

l act

iviti

es, s

it, &

sle

ep (

p ,

0.00

1) &

als

o im

prov

ed in

su

red

by a

dis

ease

-spe

cifi

c qu

estio

nnai

re &

wid

e-pa

in, d

epre

ssio

n, &

med

icat

ion

usag

e (p

,0.

0001

). S

F-36

dat

a sh

owed

sig

nifi

cant

ly b

ette

r ov

eral

l ass

essm

ent

ly u

sed

inst

rum

ents

.of

hea

lth in

all

cate

gori

es th

an th

at in

a p

ublis

hed

coho

rt o

f pa

tient

s w

/ LB

P. N

o di

ffer

ence

obs

erve

d in

7 o

f 8

cate

gori

es c

ompa

red

w/ t

he g

ener

al p

opul

atio

n. F

usio

n ra

te w

as 9

3% a

t a m

ean

of 1

28 d

ays.

3 p

atie

nts

requ

ired

reo

p: 2

for

PA

& 1

for

a d

eep

infe

ctio

n. A

poor

er o

utco

me,

sco

red

by th

e SF

-36,

was

ass

ocia

ted

w/

grea

ter

preo

p st

enos

is (

p ,

0.05

) or

occ

urre

nce

of a

com

plic

atio

n (p

,0.

05).

Park

, et a

l., 1

999

III

Rev

iew

of

32 p

atie

nts

w/ d

ecom

pres

sion

& f

usio

n. 2

1 w

/ DS

& 1

1 w

/ fai

led-

back

syn

drom

e. 2

rad

iogr

aphi

c no

n-W

eak

clin

ical

out

com

es m

easu

rem

ents

.un

ions

, but

ove

rall

succ

ess

rate

96%

for

1-l

evel

& 8

3% f

or 2

-lev

el. N

o fo

rmal

clin

ical

eva

luat

ion

othe

r th

an a

ll di

d w

ell.

Zha

o, e

t al.,

200

0II

IR

etro

spec

tive

revi

ew o

f 25

pat

ient

s w

/ DS

(Gro

up I

, 13,

sin

gle

BA

K p

oste

rior

fus

ion)

, (G

roup

II,

12,

2 B

AK

pos-

Fusi

on f

or D

S su

cces

sful

ly tr

eate

d pr

oble

ms.

Low

te

rior

fus

ion)

. Gro

up I

(11

goo

d or

exc

elle

nt, 2

fai

r, 0

poor

), G

roup

II

(9 g

ood

or e

xcel

lent

, 1 f

air,

2 po

or).

One

PA

rate

s w

/ low

poo

r ou

tcom

es.

patie

nt in

eac

h gr

oup

did

not f

use.

Ido

& U

rush

idan

i, 20

01II

IR

etro

spec

tive

revi

ew o

f 10

pat

ient

s w

/ lon

g-te

rm 5

–10-

yr F

U, D

S w

/ dec

ompr

essi

on &

fus

ion.

All

w/ g

ood

clin

i-Sm

all c

ase

seri

es w

/ lon

g-te

rm F

U.

cal r

esul

ts.

Kim

ura,

et a

l., 2

001

III

Ret

rosp

ectiv

e re

view

of

57 p

atie

nts

w/ L

4–5

DS

& s

teno

sis

havi

ng d

ecom

pres

sion

& f

usio

ns. C

ompa

riso

n w

/ &

Goo

d im

prov

emen

ts. A

dditi

on o

f PS

s im

prov

ed o

ut-

w/o

PS

inst

rum

enta

tion,

Gro

ups

A&

B. G

roup

A: 7

2% im

prov

emen

t rat

e, 2

1/29

goo

d or

exc

elle

nt, w

/ 72%

co

mes

.sa

tisfa

ctio

n ra

te. G

roup

B: 8

2% im

prov

emen

t, 23

/28

good

or

exce

llent

, 82%

sat

isfa

ctio

n ra

te. N

onun

ions

, G

roup

A(5

), G

roup

B (

2).

Kin

oshi

ta, e

t al.,

200

1II

I51

pat

ient

s un

derg

oing

mic

rode

com

pres

sion

s fo

r D

S &

ste

nosi

s w

/o f

usio

n. 7

8% g

ood-

to-e

xcel

lent

res

ults

w/o

fus

ion.

Patie

nts

doin

g re

ason

ably

wel

l w/o

fusi

on a

t 4.7

-yr F

U.

Bed

nar,

2002

III

Ret

rosp

ectiv

e re

view

of

54 p

atie

nts

w/ D

S &

ste

nosi

s tr

eate

d w

/ “m

icro

deco

mpr

essi

on”

& in

stru

men

ted

fusi

on.

Goo

d re

sults

in m

ajor

ity o

f pa

tient

s.2-

yr F

U. 1

non

unio

n, 1

6% lo

st s

ome

redu

ctio

n. O

DI

dem

onst

rate

d 77

% h

ad s

ubst

antia

l ben

efic

ial e

ffec

t of

op.

Kaw

akam

i et a

l., 2

002

III

Ret

rosp

ectiv

e re

view

of

47 p

atie

nts

w/ D

S &

ste

nosi

s ha

ving

dec

ompr

essi

on &

fus

ion.

Sag

ittal

bal

ance

mea

sure

d Fu

sion

-tre

ated

pat

ient

s w

/ DS

impr

oved

. Alig

nmen

t &

slip

s in

mm

. 1–3

leve

ls, a

utog

enou

s bo

ne, s

ome

had

inst

rum

enta

tion;

2 n

onun

ions

. Out

com

es d

epen

dent

on

infl

uenc

ed f

inal

out

com

es.

fina

l lor

dosi

s &

alig

nmen

t. O

vera

ll, J

OA

scor

e im

prov

ed f

rom

12.

6 to

21.

7 w

/ a r

ecov

ery

rate

of

55.1

%.

Park

, et a

l., 2

002

III

Ret

rosp

ectiv

e re

view

of

99 p

atie

nts

(82

w/ s

ingl

e-le

vel D

S w

/ ste

nosi

s or

dis

c he

rnia

tion)

. All

DS

patie

nts

heal

ed.

Mos

t pat

ient

s w

/ 1-l

evel

DS

& s

teno

sis:

100

% f

usio

n 70

(85

%)

of 1

-lev

el h

ad V

AS

pain

sco

res

. 5

0% f

or a

goo

d or

exc

elle

nt r

esul

t. ra

te &

goo

d cl

inic

al o

utco

me.

*D

DD

= d

egen

erat

ive

disc

dis

ease

; D

S =

deg

ener

ativ

e sp

ondy

lolis

thes

is;

face

tect

= f

acet

ecto

my;

for

amin

ot =

for

amin

otom

y; F

U =

fol

low

up;

lam

inec

t =

lam

inec

tom

y; L

BP

= l

ow-b

ack

pain

; O

DI

=O

swes

try

Dis

abili

ty I

ndex

; PA

= p

seud

arth

rosi

s; P

S =

ped

icle

scr

ew; S

IP=

Sic

knes

s Im

pact

Pro

file

; SS

= s

pond

yliti

c sp

ondy

lolis

thes

is.


supportive data are provided in references listed in the bi-bliography.


Several authors have reported the results of studies thatdemonstrate a beneficial effect when PLF is conductedfollowing decompression in patients with lumbar stenosisassociated with degenerative spondylolisthesis. Herko-witz and Kurz12 presented a series of 50 patients with va-rying degrees of degenerative spondylolisthesis and steno-sis who were treated with decompressive surgery. Patientswere alternately assigned to decompression alone (25) ordecompression combined with noninstrumented PLF (25).At a mean follow-up interval of 3 years, 96% of the pa-tients treated with a fusion/decompression reported an ex-cellent or good outcome compared with 44% of thosetreated with decompression alone. Patients who under-went fusion also reported statistically significantly lessback and leg pain on a VAS (p = 0.01, and p = 0.002,respectively). Poor results were generally associated withprogression of spinal deformity in the decompression-alone group. This study is considered to provide Class IImedical evidence in support of the use of fusion at thetime of decompression to improve functional outcome inpatients with lumbar stenosis and spondylolisthesis.

Lombardi, et al.,23 published a retrospective review of47 patients with lumbar stenosis and degenerative spondy-lolisthesis who underwent decompression. They dividedpatients into three treatment groups: decompression withfacetectomy, decompression with preservation of the facetjoints, and decompression combined with PLF. The de-compression/facetectomy group had the worst outcomes;only 33% reported good or excellent results. In the de-compression/preserved facet joint group 80% reportedgood or excellent results. In the decompression/PLF group90% reported good or excellent results. This paper pro-vides Class III medical evidence suggesting that iatrogenicinstability is associated with poor outcomes and that theaddition of a PLF improves outcomes after decompression.Several smaller studies have also provided evidence for thebeneficial effect of a PLF following decompression forlumbar stenosis and preexisting spondylolisthesis. Feffer, etal.,7 reported their results in a series of 19 patients withstenosis and degenerative spondylolisthesis. Eight patientsunderwent decompression and fusion and 11 decompres-sion alone. The patients who underwent decompression/fusion reported better results compared with those whounderwent decompression alone. Postacchini and colleag-ues29 described results obtained in a series of 32 patientswith stenosis and degenerative spondylolisthesis. Seven-teen patients were treated with decompression and fusionand 15 underwent decompression alone. Again, the de-compression/fusion group reported a higher incidence ofgood or excellent outcomes than the decompression-alonegroup. These series provide corroborating Class III med-ical evidence in support of the use of PLF in addition todecompression in patients with stenosis and degenerativespondylolisthesis.

Katz, et al.,17 reported the results of a retrospective mul-ticenter observational trial involving 272 patients withlumbar stenosis. Seventy-one percent of these patients

were treated with laminectomy alone and 29% were treat-ed with instrumented or noninstrumented fusion in addi-tion to decompression. At the 6-month follow up, theyobserved that fusion-treated patients reported better out-comes with respect to back pain scores (p , 0.004) andwalking tolerance (p , 0.05). The patients treated withfusion/decompression continued to enjoy improved out-comes at the 24-month follow-up visit (p , 0.01 low-backpain, p , 0.09 walk tolerance). In the subset of patientswith stenosis and spondylolisthesis or scoliosis, the statis-tically significant benefit of fusion was even stronger andremained stable over time (p , 0.0001). In a separateprospective nonrandomized observational trial, Katz, etal.,18 reported a similar benefit of adding fusion in 199patients with lumbar stenosis and degenerative spondy-lolisthesis of whom 61 were treated with fusion (31 non-instrumented and 30 instrumented). These observationalstudies provide Class III medical evidence supporting abeneficial effect of fusion in patients with stenosis andspinal deformity.

Further supporting evidence for the addition of fusionto decompression in patients with stenosis and spondy-lolisthesis is derived from a number of case series report-ing excellent results when using various lumbar fusiontechniques. For example, Zhao, et al.,34 studied 25 patientswith degenerative spondylolisthesis and stenosis whowere treated with decompressive procedures and concom-itant fusion involving either one or two interbody fusioncages. The results in both treatment groups were excellentwith fusion occurring in greater than 90% of patients ineach group, and good or excellent outcomes being report-ed in greater than 90% of patients in each group. Park, etal.,28 performed an intrafacet fusion supported by pediclescrew instrumentation in 99 patients who suffered fromstenosis or recurrent disc herniation and spondylolisthesis.Of the 82 patients with stenosis and degenerative spondy-lolisthesis, fusion was achieved in all and 85% reportedgood or excellent results. Bednar2 reviewed a series of 54patients with degenerative spondylolisthesis and stenosiswho underwent microdecompressive surgery and fusioninvolving placement of spinal instrumentation. Overall,77% of patients reported substantial benefit from the sur-gery. Booth, et al.,3 published a long-term follow-up studyof 41 patients who underwent decompression and PLF fordegenerative spondylolisthesis and stenosis. Eighty-threepercent of patients were satisfied with the results of surgery,and 86% reported significant functional improvement.Nork and colleagues26 reported a series of 30 patients treat-ed with decompression and fusion for degenerative spon-dylolisthesis and stenosis. Outcomes were measured usingthe SF-36 as well as a patient satisfaction score. Ninety-three percent of these patients were satisfied with theiroutcomes. Patients reported statistically significant im-provements in their ability to perform heavy and light ac-tivities, participate in social activities, sit, and sleep. Pa-tients treated with decompression/fusion also notedimprovements in pain, depression, and medication usage.The SF-36 data showed a significant improvement aftersurgery compared with baseline and also compared favor-ably with a matched cohort of conservatively treatedpatients with low-back pain. These and other series1,14,19,32

provide corroborative Class III medical evidence in sup-port of the use of fusion following decompression in pa-




tients with lumbar stenosis and degenerative spondylolis-thesis.

Several authors have published their experience in thesurgical management of patients with stenosis and spon-dylolisthesis treated with decompression with or withoutfusion. The results are variable and all studies involvednonvalidated outcome measures. Kinoshita, et al.,21 re-ported their results in 51 patients with degenerative spon-dylolisthesis who underwent decompression without fu-sion. At a mean follow-up duration of 4.7 years, 78%reported good or excellent results. Herron and Man-gelsdorf13 reported the results obtained in 24 patients withdegenerative spondylolisthesis after decompression with-out fusion. At the 34-month follow-up interval, 20 patientsreported good, three reported fair, and one patient report-ed a poor result. In contrast, Dall and Rowe6 reported poorresults in 17 patients with degenerative spondylolisthesistreated with decompression alone (six underwent exten-sive facetectomies and 11 laminectomy and less aggres-sive foraminotomy). Only nine of 17 patients reportedgood results. Overall, the preponderance of the medicalevidence appears to favor the use of lumbar fusion fol-lowing decompression in patients with stenosis andspondylolisthesis, particularly in those patients whorequire a more extensive decompression.

A number of authors have examined the role of pediclescrew fixation as an adjunct to PLF following decompres-sion in this patient population. Bridwell, et al.,4 performeda prospective study of 44 patients with claudication symp-toms and stenosis due to degenerative spondylolisthesis.Most patients were randomized into one of three groups:decompression alone (Group I); decompression and non-instrumented fusion (Group II); and decompression andinstrumented fusion (Group III). All patients with preop-erative instability were automatically placed into GroupIII. Outcomes were assessed using a satisfaction scale ap-proximately 3 years following surgery. The authors notedimproved radiographic and functional outcomes amongpatients in Group III compared with the other two treat-ment groups. This paper provides Class III medical evidence supporting the role of instrumented fusion forpatients undergoing decompression for stenosis with de-generative spondylolisthesis.

Fischgrund and colleagues8 performed a prospectiveclinical trial of 68 patients with stenosis and degenerativespondylolisthesis who were randomized into one of twogroups: decompression and PLF in one group and decom-pression and PLF supplemented with pedicle screw fixa-tion in the other. Fusion status was assessed using plain anddynamic radiography, and clinical outcomes were assessedusing a VAS for pain as well as a patient satisfaction scale.The patients treated with pedicle screw fixation had a sta-tistically significantly higher fusion rate (83%) than thenoninstrumented group (45%). Both groups demonstratedsignificant score improvements on the VAS for both backand leg pain (p , 0.001), and the majority of patients inboth groups reported their outcomes as good or excellent(78% in the instrumented group and 85% in the noninstru-mented group). This paper provides Class I medical evi-dence that pedicle screw fixation, as an adjunct to decom-pression and PLF, improves fusion success, and Class IIImedical evidence (due to the nonvalidated patient satisfac-tion score and inadequate sample size) suggesting that

pedicle screw fixation does not improve functional out-come following PLF in this patient population.

Thomsen, et al.,33 performed a randomized controlledclinical trial of 130 patients with spondylolisthesis whounderwent lumbar fusion for low-back pain. Patients wererandomized to instrumented (pedicle screw fixation) andnoninstrumented PLF groups. Overall, there was no sig-nificant difference in functional outcome (measured bythe DPQ); however, in the decompression/fusion group,patients who were treated with instrumented PLF scoredbetter than those treated with noninstrumented PLF on theactivities of daily living subsection of the DPQ. Althoughthis paper describes a randomized controlled trial withvalidated outcome measures, the small sample size ofpatients who underwent decompression and the fact that asignificant difference was noted on only one subsection ofthe DPQ, the medical evidence provided by this reportsupporting the role for pedicle screw fixation as an adjunctto PLF following decompression in patients is consideredClass III. Several other small comparative studies provideClass III medical evidence supporting the use of pediclescrew fixation as an adjunct to fusion; however, thesestudies are small, retrospective, and use nonvalidated out-come measures.5,31

Kimura, et al.,20 reported the results of a retrospectivereview of 57 patients with degenerative lumbar spondy-lolisthesis and stenosis treated with decompression andfusion with and without pedicle screw instrumentation.They reported approximately similar intergroup clinicaloutcomes but noted that patients with “excessive motion”or kyphosis associated with spondylolisthesis did betterwith pedicle screw fixation. This paper provides Class IIImedical evidence suggesting that the routine use of pedi-cle screw instrumentation does not improve functionaloutcome; however, it also provides Class III evidence sup-porting the use of pedicle screw fixation in patients withkyphosis or excessive motion at the site of the degenera-tive spondylolisthesis. Kawakami, et al.,19 also reported anassociation between preoperative instability, kyphosis,and failure of noninstrumented fusion. They found thatpatients with preoperative lumbar kyphosis who under-went reduction and pedicle screw instrumentation–aug-mented fusion did better than patients who were treatedwith noninstrumented fusion. The authors used the JOAand the VAS scores as outcome measures. Improvementsin both scores were reported for the patients treated withinstrumentation. Because of the retrospective nature of thestudy, this paper provides Class III medical evidence sup-porting the role of pedicle screw fixation as an adjunct tofusion in the presence of preoperative kyphosis. Mochida,et al.,25 also addressed the issue of instability and kypho-sis as a selection factor for the use of rigid pedicle screwfixation as an adjunct to lumbar PLF. These authors per-formed a prospective study (cases alternately assigned) ofpatients treated with a semirigid instrumentation systemcompared with those undergoing rigid pedicle screw fixa-tion. They compared the results obtained with both de-vices with a historical cohort of patients treated with non-instrumented PLF following decompression for stenosisassociated with spondylolisthesis. They found that theaddition of rigid pedicle screw fixation improved out-comes (as measured using the JOA score) compared withtheir historical controls. In 91% of patients with screw fix-



683


ation the JOA score was greater than 12, compared with71% of patients treated by noninstrumented fusion; meanscores were 13.2 and 11.2, respectively. The improvementwas most significant when patients exhibited greater than11˚ of motion on preoperative flexion–extension x-rayfilms. These three studies all provide Class III medicalevidence supporting the use of pedicle screw fixation inpatients undergoing decompression and PLF for stenosisassociated with spondylolisthesis and either kyphosis orinstability.

Summary

The best medical evidence available in the literature con-firms the utility of fusion for improving patient outcomesfollowing decompression for stenosis associated with spon-dylolisthesis. The majority of evidence from other studiescomparing outcomes after decompression alone or decom-pression combined with PLF in patients with stenosis andspondylolisthesis also favors the performance of PLF. Themedical evidence regarding the use of pedicle screw fixa-tion in this patient population is rated as Class III and isinconsistent. A consistent benefit associated with the use ofpedicle screw fixation has been reported in patients withpreoperative instability or kyphosis. Iatrogenic instabilityfollowing decompression is associated with poor outcomesand may also be treated with PLF involving supplementalinstrumentation. The precise definition of instability or ky-phosis has varied among researchers and requires furtherstudy.


A prospective comparison of outcomes using validatedoutcome measures would add significantly to the evidencesupporting or refuting the benefit of fusion following de-compression for stenosis associated with spondylolisthe-sis. A prospective multicenter study exploring the role ofpedicle screw fixation for patients with or without preop-erative instability or kyphosis and involving standardizeddefinitions would provide Class I or II evidence support-ing or refuting the role of pedicle screw fixation in thispatient population and would help standardize the defini-tion of instability applied in this patient population.

References

1. Axelsson P, Johnsson R, Stromqvist B, et al: Posterolateral lum-bar fusion. Outcome of 71 consecutive operations after 4 (2–7)years. Acta Orthop Scand 65:309–314, 1994

2. Bednar DA: Surgical management of lumbar degenerative spinalstenosis with spondylolisthesis via posterior reduction with min-imal laminectomy. J Spinal Disord Tech 15:105–109, 2002

3. Booth KC, Bridwell KH, Eisenberg BA, et al: Minimum 5-yearresults of degenerative spondylolisthesis treated with decompres-sion and instrumented posterior fusion. Spine 24:1721–1727,1999

4. Bridwell KH, Sedgewick TA, O’Brien MF, et al: The role offusion and instrumentation in the treatment of degenerativespondylolisthesis with spinal stenosis. J Spinal Disord 6:461–472, 1993

5. Chang P, Seow KH, Tan SK: Comparison of the results ofspinal fusion for spondylolisthesis in patients who are instru-mented with patients who are not. Singapore Med J 34:511–514, 1993

6. Dall BE, Rowe DE: Degenerative spondylolisthesis. Its surgicalmanagement. Spine 10:668–672, 1985

7. Feffer HL, Wiesel SW, Cuckler JM, et al: Degenerative spondy-lolisthesis. To fuse or not to fuse. Spine 10:287–289, 1985

8. Fischgrund JS, Mackay M, Herkowitz HN, et al: 1997 VolvoAward winner in clinical studies. Degenerative lumbar spondy-lolisthesis with spinal stenosis: a prospective, randomized studycomparing decompressive laminectomy and arthrodesis withand without spinal instrumentation. Spine 22:2807–2812, 1997

9. Fujiya M, Saita M, Kaneda K, et al: Clinical study on stabilityof combined distraction and compression rod instrumentationwith posterolateral fusion for unstable degenerative spondy-lolisthesis. Spine 15:1216–1222, 1990

10. Gertzbein SD, Betz R, Clements D, et al: Semirigid instrumen-tation in the management of lumbar spinal conditions combinedwith circumferential fusion. A multicenter study. Spine 21:1918–1926, 1996

11. Hanley EN Jr: Indications for fusion in the lumbar spine. BullHosp Jt Dis 55:154–157, 1996

12. Herkowitz HN, Kurz LT: Degenerative lumbar spondylolisthe-sis with spinal stenosis. A prospective study comparing decom-pression with decompression and intertransverse process ar-throdesis. J Bone Joint Surg Am 73:802–808, 1991

13. Herron L, Mangelsdorf C: Lumbar spinal stenosis: results ofsurgical treatment. J Spinal Disord 4:23–33, 1991

14. Ido K, Urushidani H: Radiographic evaluation of posterolaterallumbar fusion for degenerative spondylolisthesis: long-termfollow-up of more than 10 years vs. midterm follow-up of 2–5years. Neurosurg Rev 24:195–199, 2001

15. Inoue S, Watanabe T, Goto S, et al: Degenerative spondylolis-thesis. Pathophysiology and results of anterior interbody fusion.Clin Orthop Relat Res 227:90–98, 1988

16. Kaneda K, Kazama H, Satoh S, et al: Follow-up study of medi-al facetectomies and posterolateral fusion with instrumentationin unstable degenerative spondylolisthesis. Clin Orthop RelatRes 203:159–167, 1986

17. Katz J, Lipson S, Lew R, et al: Lumbar laminectomy alone orwith instrumented or noninstrumented arthrodesis in degenera-tive lumbar spinal stenosis. Patient selection, costs, and surgicaloutcomes. Spine 22:1123–1131, 1997

18. Katz J, Stucki G, Lipson S, et al: Predictors of surgical outcomein degenerative lumbar spinal stenosis. Spine 24:2229–2233,1999

19. Kawakami M, Tamaki T, Ando M, et al: Lumbar sagittal bal-ance influences the clinical outcome after decompression andposterolateral spinal fusion for degenerative lumbar spondy-lolisthesis. Spine 27:59–64, 2002

20. Kimura I, Shingu H, Murata M, et al: Lumbar posterolateralfusion alone or with transpedicular instrumentation in L4–L5degenerative spondylolisthesis. J Spinal Disord 14:301–310,2001

21. Kinoshita T, Ohki I, Roth KR, et al: Results of degenerativespondylolisthesis treated with posterior decompression alonevia a new surgical approach. J Neurosurg (Spine 1) 95:11–16,2001

22. Lee TC: Reduction and stabilization without laminectomy forunstable degenerative spondylolisthesis: a preliminary report.Neurosurgery 35:1072–1076, 1994

23. Lombardi JS, Wiltse LL, Reynolds J, et al: Treatment of degen-erative spondylolisthesis. Spine 10:821–827, 1985

24. Markwalder TM, Dubach R, Braun M: Soft system stabilizationof the lumbar spine as an alternative surgical modality to lum-bar arthrodesis in the facet syndrome. Preliminary results. ActaNeurochir (Wien) 134:1–4, 1995

25. Mochida J, Suzuki K, Chiba M: How to stabilize a single levellesion of degenerative lumbar spondylolisthesis. Clin OrthopRelat Res 368:126–134, 1999

26. Nork SE, Hu SS, Workman KL, et al: Patient outcomes after




decompression and instrumented posterior spinal fusion fordegenerative spondylolisthesis. Spine 24:561–569, 1999

27. Park YK, Chung HS: Instrumented facet fusion for the degen-erative lumbar disorders. Acta Neurochir (Wien) 141:915–920, 1999

28. Park YK, Kim JH, Oh JI, et al: Facet fusion in the lumbosacralspine: a 2-year follow-up study. Neurosurgery 51:88–96, 2002

29. Postacchini F, Cinotti G, Perugia D: Degenerative lumbar spon-dylolisthesis. II. Surgical treatment. Ital J Orthop Traumatol17:467–477, 1991

30. Sanderson PL, Wood PL: Surgery for lumbar spinal stenosis inold people. J Bone Joint Surg Br 75:392–397, 1993

31. Soini J, Laine T, Pohjolainen T, et al: Spondylodesis augment-ed by transpedicular fixation in the treatment of olisthetic anddegenerative conditions of the lumbar spine. Clin Orthop Re-lat Res 297:111–116, 1993

32. Takahashi K, Kitahara H, Yamagata M, et al: Long-term resultsof anterior interbody fusion for treatment of degenerativespondylolisthesis. Spine 15:1211–1215, 1990

33. Thomsen K, Christensen FB, Eiskjaer SP, et al: 1997 VolvoAward winner in clinical studies. The effect of pedicle screwinstrumentation on functional outcome and fusion rates in pos-terolateral lumbar spinal fusion: a prospective, randomized cli-nical study. Spine 22:2813–2822, 1997

34. Zhao J, Hai Y, Ordway NR, et al: Posterior lumbar interbodyfusion using posterolateral placement of a single cylindricalthreaded cage. Spine 25:425–430, 2000

Manuscript received December 7, 2004.Accepted in final form April 11, 2005.Address reprint requests to: Daniel K. Resnick, M.D., Depart-




685


Recommendations


Guidelines. There is insufficient evidence to recommenda treatment guideline.

Options. 1) In situ posterolateral lumbar fusion is notrecommended as a treatment option in patients with lum-bar stenosis in whom there is no evidence of preexistingspinal instability or likely iatrogenic instability due tofacetectomy. 2) In situ lumbar PLF is recommended as atreatment option in addition to decompression in patientswith lumbar stenosis without deformity in whom there isevidence of spinal instability. 3) The addition of pediclescrew instrumentation is not recommended in conjunctionwith PLF following decompression for lumbar stenosis inpatients without spinal deformity or instability.

Rationale

The surgical management of patients with lumbar ste-nosis without spondylolisthesis has traditionally involvedposterior decompressive procedures including laminecto-

my or laminotomy and judicious use of partial medial fa-cetectomies and foraminotomies, with or without discec-tomy. In a subset of patients who have undergone lumbarlaminectomy progressive vertebral displacement and slip-page at or adjacent to the decompressed levels will bedemonstrated. This has led some surgeons to recommendthe routine use of PLF, with or without placement of in-strumentation, to achieve an in situ arthrodesis followingdecompression. Given the additional blood loss and fu-sion-related risk, its use as an adjunct in patients with lum-bar stenosis without deformity remains controversial. Thepurpose of this review is to examine the literature con-cerning the use of PLF after decompression in patientswith lumbar stenosis without deformity. The followingtwo key questions are examined. 1) Is there evidence thatthe addition of PLF improves outcome compared withdecompression alone in patients with lumbar stenosiswithout deformity? If so, which patients with lumbarstenosis are likely to benefit from the use of adjunctivespinal fusion? 2) Is there evidence that the application ofspinal instrumentation, in addition to PLF, improves out-come compared with fusion without instrumentation inthis patient population?

Search Criteria

A computerized search of the database of the National


686

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 10: fusion following decompression in patients with stenosis without spondylolisthesis



KEY WORDS • lumbar spine • stenosis • spondylolisthesis • fusion • practice guidelines


Abbreviations used in this paper: PLF = posterolateral fusion;RCT = randomized controlled trial.


Library of Medicine from 1966 to March 2003 was con-ducted using the search terms “lumbar stenosis and fusionand spinal surgery” or “lumbar stenosis and arthrodesis.”The search was restricted to the English language andyielded 204 references. The titles and abstracts of eachreference were reviewed, and papers not concerned withdecompression and fusion for lumbar stenosis were dis-carded. Thirty-two references were identified that provid-ed either direct or supporting evidence relevant to the useof spinal fusion in the treatment of lumbar stenosis with-out spondylolisthesis. These papers were reviewed, andrelevant references from their bibliographies were identi-fied. All papers providing Class III medical evidence orbetter regarding the use of fusion and decompression fornondeformity-based lumbar stenosis are summarized inTable 1. Additional supportive data are provided by refer-ences listed in the bibliography.


In the treatment of symptomatic lumbar stenosis lumbardecompression has been demonstrated to have a clinicalefficacy of between 60 and 75% in RCTs.29 In their meta-analysis of the available literature, Turner, et al.,29 identifiedan overall 64% good outcome rate for all surgically treatedpatients with lumbar stenosis. The surgery-derived benefitdecreased over time, however, with increased patient dis-satisfaction with a longer follow-up duration.24,29 In thelumbar stenosis population, several subgroups have beenconsistently identified and subcategorized. These includepatients with preexisting spondylolisthesis, scoliosis, priordestabilizing laminectomies, and the presence of segmentalvertebral motion on flexion–extension radiographs.1,3,4,15–21

Longitudinal studies have identified a discreet proportionof patients with lumbar stenosis without deformity whohave undergone decompression and in whom progressivelumbar spinal instability, deformity, and/or spondylolisthe-sis have developed. The reported incidence of this progres-sive slippage has ranged from as low as 9% in patients withno evidence of preoperative lumbar spinal instability15 to ashigh as 73% in those with preoperative evidence of spon-dylolisthesis.8

In an analysis of 100 laminectomy-treated patients, Ca-puty and Luessenhop4 found that the main risk factor for 5-year clinical and radiographic failure was preoperativespondylolisthesis. In topic reviews and metaanalyses of theliterature, several authors have stressed the importance ofidentifying spondylolisthesis and scoliosis as significantrisk factors for delayed clinical and radiographic failureafter lumbar decompressive procedures.8,25,27,29 Multilevellaminectomies and extensive (wide) decompression havealso been shown to have a positive correlation with an in-creased incidence of progressive spondylolisthesis.13,26

Whereas spondylolisthesis and scoliosis are easily identi-fied on radiography as potential risks for postdecompres-sion instability, several techniques have been advocated asa means by which to identify more subtle forms of preop-erative spinal hypermobility or deformity. The most popu-lar of these methods are criteria based on dynamic lateralflexion–extension images.10,11,31,32 In general, PLF at thetime of lumbar decompression has been reserved for pa-tients with lumbar stenosis and preoperative radiographicevidence of hypermobility or deformity in an attempt to

minimize the chance of delayed symptomatic spondylolis-thesis or deformity.3,7,28,30 Jolles, et al.,15 reported that only in9% of 155 decompression-treated patients without preoper-ative evidence of instability did delayed slippage eventual-ly develop. Hopp and Tsou13 as well as others3,8 have report-ed that aggressive wide decompression and facetectomyperformed at the time of the decompression result in iatro-genic destabilization in certain patients and may accountfor delayed deformity in those with stenosis and normalpreoperative alignment.

At the time of lumbar decompression in patients with-out instability PLF has been performed to prevent late-onset instability and potentially to improve outcome. Cor-nefjord, et al.,6 retrospectively reviewed 124 patients ofwhom 96 were available for follow up. Three treatmentgroups were described: 59 patients underwent lumbarspinal fusion with or without placement of supplementalinstrumentation, in addition to decompression. In all sur-gical patients statistically significant benefits were dem-onstrated for walking tolerance, leg pain, and back pain (p , 0.001), with an overall 65% satisfaction rate at 7years. No significant differences in outcome were identi-fied between those who underwent fusion and those whodid not. This paper provides Class III medical evidence sug-gesting that the addition of fusion does not improve long-term outcomes in patients with stenosis and no evidence ofpreoperative spinal instability. Grob, et al.,10 randomized 45patients with stenosis but no evidence of preoperativespondylolisthesis or instability into three treatment groups:decompression alone (Group 1), decompression and single-segment fusion (Group 2), and decompression and multi-segment fusion (Group 3). In all three treatment groupspatients reported significant improvement in ambulatorystatus and pain (p , 0.001). Blood loss and operativeduration were, however, higher in the lumbar fusiongroups. No differences among the three groups were notedon a patient satisfaction survey administered at the lastfollow-up evaluation. This paper provides Class III med-ical evidence (due to the small sample size and nonvali-dated outcome measure) suggesting that the addition ofPLF does not improve outcome following decompressionin patients with lumbar stenosis and no preoperativedeformity or instability.

Herron and Mangelsdorf12 reported their experiencewith a retrospective cohort study of 140 patients treatedfor symptomatic lumbar stenosis. Nine patients underwentfusion in addition to decompression. Because of this ex-tremely small sample, the authors were unable to demon-strate a significant benefit to fusion. Rompe, et al.,24 re-viewed their results with 117 consecutive patients inwhom they performed surgery for lumbar stenosis: 90patients underwent decompression only and 27 underwentdecompression and fusion. Both groups of patients report-ed improved walking endurance and pain scores (p ,0.01). There were no statistical differences in outcomebetween the fusion and nonfusion groups. Nasca21 firstreported a retrospective series of 80 patients, and 2 yearslater he described 114 patients treated with lumbar steno-sis.22 He classified patients based on the anatomical loca-tion of stenosis, either in the lateral recess, central canal,or both, and noted the presence of associated scoliosis ordeformity. In addition to decompression, fusion was per-formed in 51 patients and decompression alone was per-


Spinal stenosis without spondylolisthesis

687




TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

pos

tdec

ompr

essi

on f

usio

n in

pat

ient

s w

ith

dege

nera

tive

lum

bar

dise

ase*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Tile

, et a

l., 1

976

III

Ret

rosp

ectiv

e se

ries

of

90 p

atie

nts

w/ l

umba

r st

enos

is. A

ll ha

d im

prov

emen

t in

clau

dica

tion

sym

ptom

s. P

atie

nts

No

com

pari

son

of f

usio

n vs

no

fusi

on in

mat

ched

w

/ DS

tend

ed to

hav

e pr

ogre

ssiv

e sl

ip. R

ecom

men

ded

fusi

on in

this

gro

up.

grou

ps. R

ecom

men

ded

fusi

on in

pat

ient

s w

/ lis

thes

is.

Shen

kin

& H

ash,

197

9II

IR

etro

spec

tive

seri

es o

f 59

pat

ient

s w

/ int

ract

able

pai

n m

ultil

evel

lam

inec

tom

ies.

Alth

ough

pro

gres

sive

slip

not

ed

Lam

inec

t alo

ne a

ssoc

iate

d w

/ goo

d ou

tcom

es, p

ar-

in 6

–15%

of

patie

nts,

ove

rall

resu

lts o

f la

min

ect a

lone

wer

e go

od.

ticul

arly

in o

lder

pat

ient

s.H

utte

r, 19

85II

IR

etro

spec

tive

revi

ew o

f 14

2 pa

tient

s tr

eate

d w

/ PL

IF: 7

8% g

ood

resu

lt.

Tech

nica

l rep

ort &

cas

e se

ries

. N

asca

, 198

7II

IR

etro

spec

tive

revi

ew o

f 80

pat

ient

s w

/ lum

bar s

teno

sis.

Pap

er v

ery

sim

ilar t

o th

at p

ublis

hed

in 1

989

(see

Nas

ca, 1

989)

.H

opp

& T

sou,

198

8II

IR

etro

spec

tive

stud

y of

334

pat

ient

s. A

utho

rs n

oted

17%

reo

p ra

te f

or in

stab

ility

. Pre

op p

redi

ctor

s of

inst

abili

ty in

-R

ecom

men

ded

spar

ing

dors

al e

lem

ents

as

muc

h as

cl

ude

trac

tion

spur

s, d

ecre

ased

dis

c he

ight

, lis

thes

is, &

sco

liosi

s.po

ssib

le d

urin

g de

com

pres

sion

. Rec

omm

ende

d fu

-si

on in

cas

es w

/ pre

op r

isk

fact

ors.

Nas

ca, 1

989

III

Ret

rosp

ectiv

e re

view

of

114

patie

nts

w/ l

umba

r st

enos

is. D

ivid

ed in

to 4

gro

ups:

lat r

eces

s st

enos

is (

15)

(uni

lat d

e-O

vera

ll, n

otic

ed tr

end

of b

ette

r ou

tcom

es in

eac

h su

b-co

mpr

essi

on, 2

nee

ded

dela

yed

fusi

on),

cen

tral

mix

ed (

45 [

16 f

used

]), p

rior

lum

bar

surg

ery

(43

[22

fuse

d]),

gr

oup

w/ f

usio

n ex

cept

isol

ated

rec

ess

sten

osis

. sc

olio

sis

(11/

11 f

used

ove

r m

ajor

cur

ve)—

over

all g

ood

resu

lts a

chie

ved

in 7

0%. T

he a

utho

rs id

entif

ied

the

fol-

Subg

roup

s id

entif

ied

who

did

bet

ter

w/ f

usio

n lo

win

g gr

oups

who

did

bet

ter

w/ f

usio

n th

an d

ecom

pres

sion

alo

ne: l

isth

esis

, deg

ener

ated

fac

ets

w/ c

olla

psed

th

an d

ecom

pres

sion

alo

ne.

disc

, rad

iogr

aphi

c in

stab

ility

, or

scol

iosi

s.

Con

ley,

et a

l., 1

990

III

Ret

rosp

ectiv

e re

view

of

25 p

atie

nts,

w/ 2

-yr

FU. A

ll ha

d in

trao

p in

stab

ility

not

ed. K

nodt

rod

fac

et f

usio

ns d

one.

B

asic

ally

a te

chni

que

pape

r.H

ypot

hesi

zed

bette

r ax

ial L

BP

outc

omes

. 15/

25 h

ad s

ome

mov

emen

t on

preo

p ra

diog

raph

s. F

law

ed b

y ph

y-si

cian

ass

essm

ents

. 72%

fus

ion

rate

.H

erko

witz

& K

urz,

199

1II

IPr

imar

ily a

list

hesi

s pa

per.

Incl

uded

in p

rior

met

aana

lysi

s &

dis

cuss

ed h

ere

as s

uppo

rtin

g da

ta r

egar

ding

infl

u-Pr

ogre

ssiv

e de

form

ity f

ollo

win

g de

com

pres

sion

is

ence

of

listh

esis

pro

gres

sion

on

outc

ome.

asso

ciat

ed w

/ poo

r ou

tcom

es. F

usio

n pr

even

ts p

ro-

gres

sive

def

orm

ity.

Her

ron

& M

ange

lsdo

rf, 1

991

III

Ret

rosp

ectiv

e re

view

of

140

patie

nts.

Onl

y 9

unde

rwen

t fus

ion.

No

corr

elat

ion

of o

utco

me

w/ f

usio

n us

ing

x2

No

conc

lusi

ons

draw

n fr

om th

is r

epor

t. m

ultiv

aria

te a

naly

sis.

C

aput

y &

Lue

ssen

hop,

199

2II

IR

etro

spec

tive

seri

es o

f 10

0 pa

tient

s tr

eate

d w

/ lam

inec

t alo

ne; 5

-yr

FU in

88

patie

nts.

27%

fai

lure

rat

e at

5 y

rs,

Cas

e se

ries

. No

sign

ific

ant c

oncl

usio

ns c

an b

e 29

% li

sthe

sis.

Mai

n ri

sk f

acto

r fo

r po

or o

utco

me

was

list

hesi

s.

reac

hed.

Lou

is &

Naz

aria

n, 1

992

III

Ret

rosp

ectiv

e re

view

of

350

patie

nts,

280

w/ 2

-yr

FU. I

nitia

lly u

sed

fusi

on w

/ scr

ews/

plat

es in

43%

cas

es, t

hen

The

use

of

fusi

on w

as a

ssoc

iate

d w

/ bet

ter

relie

f of

drop

ped

to 1

7% o

f ca

ses.

Aga

in o

nly

listh

esis

, ins

tabi

lity,

wid

e un

derc

uttin

g. O

bser

ved

tren

d of

bet

ter

ob-

LB

Pin

som

e pa

tient

gro

ups.

ject

ive

patie

nt c

rite

ria

85 v

s 65

% in

pat

ient

s w

/ fus

ion.

Mix

ed b

ag, s

pars

e da

ta.

Tur

ner,

et a

l., 1

992

III

Met

aana

lysi

s on

op

for

sten

osis

incl

uded

74

artic

les.

Com

men

ted

on p

oor

qual

ity o

f av

aila

ble

stud

ies—

very

dif

-A

utho

rs r

epor

ted

an o

vera

ll 64

% g

ood

outc

ome

rate

fe

rent

out

com

e va

riab

les,

op

tech

niqu

es. O

f 74

artic

les

37 h

ad fu

sion

as

op o

ptio

n. C

ritic

al o

f out

com

e m

easu

res

used

.fo

r op

of

lum

bar

sten

osis

. Cou

ld n

ot m

ake

stat

e-m

ents

reg

ardi

ng r

ole

of f

usio

n.B

ridw

ell,

et a

l., 1

993

III

Pseu

dora

ndom

ized

stu

dy o

f pa

tient

s w

/ ste

nosi

s du

e to

DS.

Pat

ient

s w

ho u

nder

wen

t ins

trum

ente

d PL

F di

d be

tter.

Inst

rum

ente

d PL

F va

luab

le f

or p

atie

nts

w/ s

teno

sis

due

to D

S.G

rob,

et a

l., 1

995

III

RC

Tof

45

patie

nts

w/ s

teno

sis

& n

o in

stab

ility

or

defo

rmity

. Pat

ient

s di

vide

d in

to 3

gro

ups:

lam

inec

t, la

min

ect 1

No

adva

ntag

e to

fus

ion

in th

is p

opul

atio

n.fu

sion

, lam

inec

t 1fu

sion

w/ i

nstr

umen

tatio

n. A

ll pa

tient

s ha

d be

tter

wal

king

& p

ain

(p ,

0.00

1). N

onva

lidat

ed

outc

ome

mea

sure

use

d.Fo

x, e

t al.,

199

6II

IR

etro

spec

tive

seri

es o

f 12

4 pa

tient

s tr

eate

d w

/ lam

inec

t; 32

trea

ted

w/ f

usio

n as

wel

l. M

ean

FU w

as 5

.8 y

rs.

Lis

thes

is w

as m

ajor

pre

dict

or o

f de

laye

d sl

ip/f

ailu

re/

31%

of

patie

nts

w/ n

orm

al a

lignm

ents

trea

ted

w/ l

amin

ect a

lone

suf

fere

d a

dela

yed

slip

; 73%

had

del

ayed

re

op. M

otio

n w

as a

lso

impo

rtan

t. 91

% tr

eate

d w

/ sl

ip if

list

hesi

s id

entif

ied

preo

p.

fusi

on h

ad g

ood

outc

omes

vs

75%

w/ l

amin

ect

alon

e.

Yon

e, e

t al.,

199

6II

IR

etro

spec

tive

seri

es o

f 34

pat

ient

s, m

ean

age

68 y

rs. U

sed

flex

–ext

film

s to

ass

ess

inst

abili

ty. P

atie

nts

divi

ded

into

C

oncl

uded

cri

teri

a us

eful

in id

entif

ying

thos

e at

ris

k3

grou

ps: 1

sta

ble

w/ l

amin

ect o

nly,

7 u

nsta

ble

w/ l

amin

ect o

nly,

10

unst

able

w/ l

amin

ect/i

nstr

umen

ted

fusi

on;

for

inst

abili

ty. I

n th

ese

patie

nts,

fus

ion

impr

oved

JOA

scor

es, S

teff

ee/K

nodt

rod

s; g

roup

s si

mila

r pr

eop

in p

ain/

func

tiona

l sco

res.

outc

ome

(p ,

0.05

). P

atie

nts

w/ u

nsta

ble

fusi

ondi

d ju

st a

s w

ell w

hen

fuse

d as

thos

e w

/o in

stab

ili-

ty &

w/o

fus

ion.

Kat

z, e

t al.,

199

7II

I27

2 pa

tient

s m

aile

d qu

estio

nnai

res

6 yr

s po

stop

. 71%

had

bee

n tr

eate

d w

/ lam

inec

t alo

ne, 1

4% in

situ

fus

ion,

Fu

sion

app

eare

d to

hav

e a

bene

fit.

No

clea

r be

nefi

t15

% in

stru

men

ted

fusi

on. F

usio

n pa

tient

s w

ere

youn

ger

w/ g

reat

er p

reop

pai

n. S

imila

r le

g pa

in, w

alki

ng c

a-of

sup

plem

enta

l ins

trum

enta

tion

was

dem

onst

rate

d.pa

city

, sic

knes

s pr

ofile

s, &

com

orbi

dity

. N

igge

mey

er, e

t al.,

199

7II

IM

etaa

naly

sis

of 3

0 ar

ticle

s th

at d

escr

ibed

166

8 ca

ses:

dec

ompr

essi

on, d

ecom

pres

sion

& in

situ

fus

ion;

dec

ompr

es-

Rec

omm

ende

d le

ast i

nvas

ive

op f

or p

atie

nts

w/

sion

1in

stru

men

ted

fusi

on).

Use

d un

ique

out

com

e m

easu

re. 2

sub

grou

ps c

ompa

red

base

d on

leng

th o

f pr

eop

shor

ter

dura

tion

of s

ympt

oms.

Lon

ger

wou

ld s

ug-

sym

ptom

s. P

atie

nts

w/ s

hort

er d

urat

ion

of s

ympt

oms

did

bette

r w

/ dec

ompr

essi

on a

lone

; tho

se w

/ lon

ger

du-

gest

inst

rum

enta

tion

(p ,

0.05

).ra

tion

of s

ympt

oms

did

bette

r w

/ dec

ompr

essi

on 1

inst

rum

ente

d fu

sion

. At #

8 y

rs: d

ecom

pres

sion

alo

ne b

est.

Con

tinue

d




689

TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Gib

son,

et a

l., 1

999

III

Coc

hran

e re

view

of

14 R

CTs

for

DS/

sten

osis

. 3 o

f th

e id

entif

ied

RC

Ts a

naly

zed

adju

nct f

usio

n w

/ 139

pat

ient

s N

o di

ffer

ence

btw

n la

min

ect 1

fus

ion

vs la

min

ect

tota

l w/ 9

9% F

U a

t 2–3

yrs

. on

ly w

hen

com

bine

d an

alys

is o

f 3

RC

Ts p

er-

form

ed (

OR

0.8

, 95%

CI

0.31

–2.1

0).

Kat

z, e

t al.,

199

9II

I19

9 of

305

pat

ient

s se

nt q

uest

ionn

aire

2 y

rs p

osto

p re

plie

d. P

atie

nts

wer

e tr

eate

d w

/ lam

inec

t onl

y (1

38),

lam

in-

Rec

omm

enda

tion

for

surg

ery

was

str

ong.

ect 1

in s

itu f

usio

n (3

1), o

r la

min

ect 1

inst

rum

ente

d fu

sion

(30

). F

usio

n ca

ses

had

bette

r ou

tcom

e; n

o di

f-fe

renc

e bt

wn

inst

rum

ente

d &

non

inst

rum

ente

d gr

oups

.R

ompe

, 199

9II

IR

etro

spec

tive

revi

ew o

f 11

7 co

nsec

utiv

e pa

tient

s di

vide

d in

to 3

gro

ups:

lam

inec

t & u

nder

cutti

ng d

ecom

pres

-M

ain

conc

lusi

on w

as th

at g

ood

resu

lts a

re a

chie

ved

si

on (

39),

lam

inec

t w/ f

oram

inot

omie

s (n

= 5

1), &

dec

ompr

essi

on &

fus

ion

(27)

. 8-y

r FU

. Wal

king

cap

acity

w/ v

ario

us o

ps b

ut th

at th

e be

nefi

ts o

f op

dec

line

(p ,

0.00

1) &

pai

n (p

,0.

01)

impr

oved

in a

ll gr

oups

; how

ever

, im

prov

emen

ts d

eclin

ed o

ver

time.

ov

er ti

me.

Vita

z, e

t al.,

199

9II

IR

etro

spec

tive

revi

ew o

f 65

pat

ient

s: 2

1% la

min

ect,

79%

lam

inec

t/fus

ion.

Spe

cifi

cally

look

ed a

t old

er (

age

.75

N

o co

mpa

riso

n bt

wn

grou

ps.

yrs)

. Fus

ed p

atie

nts

w/ .

50%

fac

et r

emov

al, s

ever

e L

BP,

inst

abili

ty, o

r lis

thes

is. G

ood

clin

ical

out

com

es

achi

eved

in 8

9–90

%; 9

6% f

usio

n ra

te.

Yon

e &

Sak

ou, 1

999

III

Ret

rosp

ectiv

e. E

xten

sion

of

prio

r pa

per.

60 p

atie

nts:

33

unst

able

(19

fus

ed, 1

4 on

ly la

min

ect)

& 2

2 st

able

by

Patie

nts

w/ u

nsta

ble

fusi

on h

ad s

imila

r ou

tcom

es

flex

–ext

cri

teri

a. A

gain

, goo

d re

sults

for

lam

inec

t onl

y in

sta

ble

grou

p &

lam

inec

t 1fu

sion

in p

atie

nts

w/ d

e-to

thos

e w

/o in

stab

ility

trea

ted

w/ d

ecom

pres

sion

fo

rmity

or

inst

abili

ty.

alon

e.C

orne

fjor

d, e

t al.,

200

0II

IR

etro

spec

tive

revi

ew o

f 12

4 pa

tient

s. 9

6 av

aila

ble

for

FU. I

ndep

ende

nt q

uest

ionn

aire

s. P

aram

eter

s: s

atis

fact

ion,

N

o st

atis

tical

dif

fere

nces

btw

n th

ose

fuse

d or

not

L

BP,

leg

pain

, & w

alki

ng c

apac

ity. O

p be

nefi

ted

all p

aram

eter

s es

peci

ally

wal

king

by

$50

% (

p ,

0.00

1) &

fu

sed.

No

diff

eren

ce b

twn

trea

tmen

t w/ i

nstr

u-le

g 1

back

pai

n (p

,0.

001)

. Lon

g FU

, 7.1

yrs

. Ove

rall

65%

rep

orte

d sa

tisfa

ctio

n. 4

2 ha

d in

stru

men

ted

PLF.

m

ente

d vs

non

inst

rum

ente

d fu

sion

.17

had

onl

y in

situ

fus

ions

. Fus

ion

base

d on

pre

op d

efor

mity

or

iatr

ogen

ic in

stab

ility

.Jo

lles,

et a

l., 2

001

III

Ret

rosp

ectiv

e re

view

of

155

patie

nts,

77

had

6.5-

yr F

U. R

olan

d &

Pro

lo s

cale

s us

ed. 7

9% h

ad d

ecre

ase

in p

ain

9%of

pat

ient

s w

/o p

reop

evi

denc

e of

inst

abili

ty w

ent

w/ e

xcel

lent

out

com

es. 9

% h

ad d

elay

ed s

lip. 1

2 w

/ fus

ion

who

had

pre

op e

vide

nce

of in

stab

ility

had

no

dela

yed

slip

s.on

to h

ave

dela

yed

slip

aft

er la

min

ect w

/o f

usio

n.Sh

eeha

n, e

t al.,

200

1II

ITo

pic

revi

ew—

auth

ors

conc

lude

d th

at th

ere

is n

o be

nefi

t der

ived

fro

m th

e ad

ditio

n of

fus

ion

to d

ecom

pres

sion

To

pic

revi

ew.

alon

e in

the

abse

nce

of s

egm

enta

l ins

tabi

lity.

Ins

trum

enta

tion

was

rec

omm

ende

d on

ly f

or s

colio

sis

or s

ever

e sl

ips

w/ h

igh

risk

of p

rogr

essi

on. O

ther

indi

catio

ns fo

r fus

ion

incl

uded

sco

liosi

s, .

50%

face

t res

ectio

n, o

r lis

thes

is.

*C

I =

con

fide

nce

inte

rval

; DS

= d

egen

erat

ive

spon

dylo

listh

esis

; fle

x–ex

t = f

lexi

on–e

xten

sion

; FU

= f

ollo

w u

p; J

OA

= J

apan

ese

Ort

hopa

edic

Ass

ocia

tion;

lam

inec

t = la

min

ecto

my;

LB

P=

low

-bac

k pa

in;

listh

esis

= s

pond

ylol

isth

esis

; OR

= o

dds

ratio

; PL

IF =

pos

teri

or lu

mba

r in

terb

ody

fusi

on.

formed in 58 patients. He reported an overall good out-come rate of 70%. There was a trend toward better resultswith fusion in patients with spondylolisthesis, scoliosis,severely degenerated facet joints, collapsed disc spaces,and preoperative movement noted on dynamic spinal ra-diographs. These trends, however, did not achieve statis-tical significance, even in the subgroup in which theseidentified risk factors were present. All of the aforemen-tioned papers provide Class III medical evidence suggest-ing that PLF is not necessary or beneficial in the majorityof patients undergoing decompression for symptomaticlumbar stenosis without deformity or instability. Becauseof the retrospective nature of all of these studies, the factthat patient selection criteria for fusion compared with nofusion were not standardized, and the relatively smallnumber of patients involved in each study, definitive con-clusions regarding the role of fusion in this patient popu-lation cannot be made.

Fox, et al.,8 performed a retrospective analysis of 124patients they had treated for symptomatic lumbar stenosis.They observed that 91% of those who had undergone lam-inectomy and fusion reported good outcomes compared to75% of those treated with laminectomy alone. The majori-ty of patients selected for fusion were reported to have pre-operative instability or spondylolisthesis. Only twopatients without instability or spondylolisthesis underwentfusion. Their mean follow-up duration was 5.8 years; theywere administered a patient satisfaction survey. This paperprovides Class III medical evidence supporting the use oflumbar fusion in patients with lumbar stenosis, particular-ly those with evidence of preoperative instability. Her-kowitz and Kurz11 alternately assigned a group of 50patients with stenosis and varying degrees of spondylolis-thesis to laminectomy alone or laminectomy combinedwith in situ noninstrumented fusion (25 patients in eachgroup). Fusion-treated patients fared much better thanthose treated with laminectomy alone at the 2.4-year fol-low up in terms of satisfaction and pain control (visualanalog scale, p , 0.0001). Progressive spondylolisthesisand segmental vertebral angulation were noted more fre-quently in the laminectomy-alone group and were associ-ated with poor outcomes. This paper provides Class IImedical evidence that the addition of a noninstrumentedfusion improves outcomes in patients with lumbar steno-sis and spondylolisthesis. In terms of patients without pre-operative deformity or instability, this paper providesClass III medical evidence in favor of the addition of fu-sion at the time of decompression in that poor outcomeswere associated with late-onset deformity.

Katz, et al.,16 reported on 272 patients who had under-gone treatment for lumbar stenosis. In a multicenter retro-spective trial, 71% of patients were treated with laminec-tomy alone and 29% with instrumented/noninstrumentedfusion. At 6 months, the authors observed that fusion-treated patients fared better with regard to back painscores (p , 0.004) and walking tolerance (p , 0.05), butthis benefit deteriorated by the time of the 24-month fol-low up (p , 0.01 low-back pain; p , 0.09 walk toler-ance). For the subset of patients with preoperative spon-dylolisthesis or scoliosis, the benefit of fusion wasstatistically significant and more stable over time (p ,0.0001). In a different prospective nonrandomized obser-vational trial, Katz, et al.,17 reported a similar benefit of


fusion in 199 patients with lumbar stenosis, 61 of whomwere treated with fusion (31 in situ and 30 instrumented).These two studies provide Class III medical evidence insupport of a beneficial effect derived from adding fusionat the time of decompression in patients with stenosis andconcomitant spinal deformity.

Gibson, et al.,9 performed a metaanalysis on the subjectof spinal stenosis for the Cochrane review. They identifiedthree RCTs with a total of 139 patients; 99% of patientswere available for follow up at 2 to 3 years (these studieshave been discussed in previous paragraphs).2,10,11 In theirmetaanalysis Gibson and colleagues concluded that nosignificant benefit existed for lumbar fusion in the treat-ment of patients with lumbar stenosis (odds ratio 0.8, 95%confidence interval 0.31–2.10). The rationale used to justify the combined analysis of these studies with dispar-ate patient populations (for example, Grob, et al.,10 andBridwell, et al.,2 treated completely different patient pop-ulations) is not well described. Turner, et al.,29 attempted aseparate metaanalysis regarding this issue and concludedthat the medical evidence available from the 74 papersthey reviewed could not be combined to provide definitiveconclusions regarding the use of fusion as an adjunct inthe treatment of patients with lumbar stenosis. Theyreported that the primary problem was the heterogeneousoutcome measures utilized among the different patientseries. A third metaanalysis performed on this subjectrevealed the potential influence of symptom duration.Niggemeyer, et al.,22 demonstrated that among patientswith symptoms of less than 8 years’ duration, decompres-sion alone yielded the best outcomes; however, in patientswith symptoms of stenosis for longer than 15 years’ dura-tion, instrumentation-augmented fusion and decompres-sion yielded the best results (p , 0.05). Because of thepreviously cited difficulty in combining data from differ-ent studies involving different outcome measures, theseobservations must be considered with caution.

Several retrospective series were identified that de-scribe the results of various procedures for instrumentedspinal fusion in addition to decompression for spinal ste-nosis in patients without deformity.5,23 For example, Con-ley, et al.,5 retrospectively reviewed the use of Knodt rodsas a routine adjunct to decompression in patients withstenosis and reported a 66% excellent outcome rate and a72% radiographic fusion rate. In general, these series offerlittle insight into the advantage of adding modern seg-mental instrumentation to a PLF during the treatment oflumbar stenosis.

A few series have compared the use of instrumentedfusion and decompression with lumbar decompressionalone. Louis and Nazarian18 summarized their experiencewith 350 patients of whom nearly 34% underwent fusionsupplemented with pedicle screw fixation in addition todecompression in patients with preoperative spondylolis-thesis or instability, and in cases in which a wide decom-pression was accomplished. In terms of lumbar pain, theywere able to demonstrate a trend toward better outcomes in patients treated with decompression and pediclescrew–augmented fusion (85%) compared with those whounderwent decompression alone (65%). No patient under-went noninstrumented fusion in this cohort. Yone, et al.,31,32

performed a retrospective analysis of 60 patients with lum-bar stenosis. Thirty-three patients were identified as having

preoperative spinal instability based on lateral flexion–ex-tension radiography. Three groups were assessed: patientswith stenosis and instability treated with laminectomy andinstrumented fusion (some pedicle screws and some Knodtrods), those with stenosis and instability treated withlaminectomy alone, and those with stenosis and no insta-bility treated with laminectomy alone. In patients withstenosis and instability, improvements in low-back painoutcomes were significantly greater (p , 0.05) in thefusion group. The authors observed similar rates of goodoutcomes (80%) among fusion-treated patients with pre-operative instability or deformity and those without insta-bility or deformity who underwent decompression alone.Cornefjord, et al.,6 performed a retrospective review of 124patients who underwent surgery for lumbar stenosis. Fifty-nine patients underwent fusion, of whom 42 also receivedsupplemental pedicle screw instrumention. No differenceswere observed between patients in the fusion and nonfu-sion groups or between patients in the instrumentation-aug-mented fusion group and those in the noninstrumented fu-sion group. Because of differences in selection criteria forinstrumentation, the use of nonvalidated outcome mea-sures, and the retrospective nature of these reports, all oftheir medical evidence regarding the use of supplementalinstrumentation for PLF following decompression in pa-tients with lumbar stenosis and without deformity is con-sidered Class III.

Summary

Based on the medical evidence derived from the scien-tific literature on this topic, there does not appear to beevidence to support the hypothesis that fusion (with orwithout instrumentation) provides any benefit over de-compression alone in the treatment of lumbar stenosis inpatients in whom there is no evidence of preoperativedeformity or instability. A single report provides Class IImedical evidence and several papers provide Class IIImedical evidence suggesting that the addition of fusion todecompression in patients with lumbar stenosis and insta-bility evidenced by movement on preoperative flexion–extension radiographs does improve outcome. There arealso reports (Class III medical evidence) indicating thatpatients with lumbar stenosis, without deformity or insta-bility, treated with wide decompression or facetectomymay suffer iatrogenic lumbar instability. Fusion in thesepatients may improve outcome. There is conflicting ClassIII medical evidence regarding the application of instru-mentation in addition to PLF in patients treated for lumbarstenosis without deformity or preoperative instability.

Future Research Directions

Clinical outcome in patients without deformity or insta-bility undergoing decompression for lumbar stenosisshould be radiographically confirmed in patients with andwithout evidence of instability on preoperative x-ray films.Such patients could be randomized into fusion or nonfusiongroups, thus allowing accrual of Class I evidence to supportor refute the hypothesis that the added expense, operativetime, and increased morbidity associated with fusion is jus-tified by a clinical benefit. The study should use validated




clinical outcome measures for both decompression- (suchas walking tolerance) and fusion- (such as back pain anddisability scales) related outcomes. If the value of fusion isconvincingly demonstrated, then the issue of supplementalinternal fixation can be addressed.

References

1. Benz R, Garfin S: Current techniques of decompression of thelumbar spine. Clin Orthop Relat Res 384:75–81, 2001

2. Bridwell K, Sedgewick TA, O’Brien MF, et al: The role of fu-sion and instrumentation in the treatment of degenerative spon-dylolisthesis with spinal stenosis. J Spinal Disord 6:461–472,1993

3. Brunon J, Chazal J, Chirossel J, et al: When is spinal fusionwarranted in degenerative lumbar spinal stenosis? Rev RhumEngl Ed 63:44–50, 1996

4. Caputy A, Luessenhop A: Long-term evaluation of decompres-sive surgery for degenerative lumbar stenosis. J Neurosurg77:669–676, 1992

5. Conley F, Cady C, Lieberson R: Decompression of lumbarspinal stenosis and stabilization with Knodt rods in the elderlypatient. Neurosurgery 26:758–763, 1990

6. Cornefjord M, Byrod G, Brisby H, et al: A long-term (4- to 12-year) follow-up study of surgical treatment of lumbar spinalstenosis. Eur Spine J 9:563–570, 2000

7. DiPierro C, Helm G, Shaffrey C, et al: Treatment of lumbarspinal stenosis by extensive unilateral decompression and con-tralateral autologous bone fusion: operative technique and re-sults. J Neurosurg 84:166–173, 1996

8. Fox MW, Onofrio BM, Onofrio BM, et al: Clinical outcomesand radiological instability following decompressive lumbar laminectomy for degenerative spinal stenosis: a comparison ofpatients undergoing concomitant arthrodesis versus decompres-sion alone. J Neurosurg 85:793–802, 1996

9. Gibson J, Grant I, Waddell G: The Cochrane review of surgeryfor lumbar disc prolapse and degenerative lumbar spondylosis.Spine 24:1820–1832, 1999

10. Grob D, Humke T, Dvorak J: Degenerative lumbar spinal steno-sis: decompression with and without arthrodesis. J Bone JointSurg Am 77:1036–1041, 1995

11. Herkowitz H, Kurz L: Degenerative lumbar spondylolisthesiswith spinal stenosis. J Bone Joint Surg 73:802–808, 1991

12. Herron L, Mangelsdorf C: Lumbar spinal stenosis: results ofsurgical treatment. J Spinal Disord 4:23–33, 1991

13. Hopp E, Tsou P: Postdecompression lumbar instability. ClinOrthop Relat Res 227:143–151, 1988

14. Hutter C: Spinal stenosis and posterior lumbar interbody fusion.Clin Orthop Relat Res 193:103–114, 1985

15. Jolles B, Porchet F, Theumann N: Surgical treatment of lumbarspinal stenosis. Five-year follow-up. J Bone Joint Surg Br 87:949–953, 2001

16. Katz J, Lipson S, Lew R, et al: Lumbar laminectomy alone orwith instrumented or noninstrumented arthrodesis in degenera-

tive lumbar spinal stenosis. Patient selection, costs, and surgicaloutcomes. Spine 22:1123–1131, 1997

17. Katz J, Stucki G, Lipson S, et al: Predictors of surgical outcomein degenerative lumbar spinal stenosis. Spine 24:2229–2233,1999

18. Louis R, Nazarian S: Lumbar stenosis surgery: the experienceof the orthopaedic surgeon. Chir Organi Mov 77:23–29, 1992

19. Nasca R: Lumbar spinal stenosis: surgical considerations. JSouth Orthop Assoc 11:127–134, 2002

20. Nasca R: Rationale for spinal fusion in lumbar spinal stenosis.Spine 14:451–454, 1989

21. Nasca R: Surgical management of lumbar spinal stenosis. Spine12:809–816, 1987

22. Niggemeyer O, Strauss J, Schulitz K: Comparison of surgicalprocedures for degenerative lumbar spinal stenosis: a meta-analysis of the literature from 1975 to 1995. Eur Spine J 6:423–429, 1997

23. Ray C: Transfacet decompression with dowel fixation: a newtechnique for lumbar lateral spinal stenosis. Acta NeurochirSuppl 43:48–54, 1988

24. Rompe J, Eysel P, Zollner J, et al: Degenerative lumbar spinalstenosis. Long-term results after undercutting decompressioncompared with decompressive laminectomy alone or with in-strumented fusion. Neurosurg Rev 22:102–106, 1999

25. Sheehan J, Shaffrey C, Jane JA Sr: Degenerative lumbar steno-sis: the neurosurgical perspective. Clin Orthop Relat Res 384:61–74, 2001

26. Shenkin H, Hash C: Spondylolisthesis after multiple bilaterallaminectomies and facetectomies for lumbar spondylosis. JNeurosurg 50:45–47, 1979

27. Sonntag V, Marciano F: Is fusion indicated for lumbar spinaldisorders? Spine 20 (24 Suppl):S138–S142, 1995

28. Tile M, McNeil S, Zarins R, et al: Spinal stenosis. Results oftreatment. Clin Orthop Relat Res 115:104–108, 1976

29. Turner J, Ersek M, Herron L, et al: Surgery for lumbar spinalstenosis. Attempted meta-analysis of the literature. Spine 17:1–8, 1992

30. Vitaz T, Raque G, Shields C, et al: Surgical treatment of lumbarspinal stenosis in patients older than 75 years of age. J Neu-rosurg (Spine 2) 91:181–185, 1999

31. Yone K, Sakou T: Usefulness of Posner’s definition of spinalinstability for selection of surgical treatment for lumbar spinalstenosis. J Spinal Disord 12:40–44, 1999

32. Yone K, Sakou T, Kawauchi Y, et al: Indication of fusion forlumbar spinal stenosis in elderly patients and its significance.Spine 21:242–248, 1996





691


Recommendations


Guidelines. In the context of a single-level stand-aloneALIF or ALIF with posterior instrumentation, the additionof a PLF is not recommended as it increases operatingroom time and blood loss without influencing the likeli-hood of fusion or the functional outcome.

Options. 1) It is recommended that both PLF and inter-body fusion (PLIF, TLIF, or ALIF) techniques be consid-ered as treatment options for patients with low-back paindue to DDD at one or two levels. 2) Placement of an inter-body graft is recommended as a treatment option toimprove fusion rates and functional outcome in patientsundergoing surgery for low-back pain due to DDD at oneor two levels. The surgeon is cautioned that the marginalimprovement in fusion rates and functional outcome withthese techniques is associated with increased complicationrates, particularly when combined approaches (that is,360˚) are used. 3) The use of multiple approaches (anteri-

or and posterior) to accomplish lumbar fusion is not rec-ommended as a routine option for the treatment of patientswith low-back pain without deformity.

Rationale

The surgical treatment of low-back pain has evolvedover the last several decades, and interbody techniqueshave been proposed as surgical alternatives to posterolater-al lumbar fusion. Placement of the graft within the load-bearing column of the spine has biomechanical advantagesand has been reported to result in higher fusion rates withimproved patient outcomes compared with PLF techni-ques. A variety of techniques are available for the applica-tion of interbody grafts, and each technique has its particu-lar advantages, disadvantages, and champions. The purposeof this review is to examine the literature reporting experi-ence with interbody fusion techniques and their relativesafety and efficacy compared with posterolateral fusiontechniques for the treatment of patients with low-back pain.

Literature Search

A computerized search of the National Library of Me-dicine database of the literature published from 1966 toJune 2003 was performed. A search using the subjectheading “spinal fusion, lumbar, treatment outcome, low-back pain” yielded 1030 citations. Clinical series reportedin English-language journals dealing with adult patients


692

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 11: interbody techniques for lumbar fusion



KEY WORDS • fusion • lumbar spine • bone graft • spondylosis •practice guidelines


Abbreviations used in this paper: ALIF = anterior lumbar inter-body fusion; CT = computerized tomography; DDD = degenerativedisc disease; DPQ = Dallas Pain Questionnaire; LBO = low-backoutcome; LOS = length of stay; ODI = Oswestry Disability Index;PLF = posterolateral fusion; PLIF = posterior LIF; TCBD = thread-ed cortical bone dowel; TLIF = transformal LIF; VAS = visual ana-log scale.


who had undergone fusion with instrumentation fordegenerative lumbar disease were selected (333 refer-ences). Relevant articles pertaining to the comparison ofinterbody fusion techniques with other surgical techniquesor nonsurgically treated controls were selected and aresummarized in the evidentiary table. A number of caseseries provide supporting data and are referenced in thebibliography.


Recent trends in spinal surgery include the use of inter-body fusion techniques including ALIF, PLIF, or TLIF asa means to enhance the rate of successful arthrodesis.Authors of several studies have compared the results ofthese techniques with respect to each other as well as withrespect to PLF.

Christensen, et al.,10 studied a series of 148 patients withsevere low-back pain who were prospectively randomizedto treatment with PLF with pedicle screws or ALIF withBrantigan cages in addition to posterior instrumentationand PLF. Outcomes were measured using the DPQ and theLow Back Pain Rating Scale. There was a trend towardbetter overall functional outcome for patients treated withthe circumferential procedure but this was not statisticallysignificant (p , 0.08). This patient group did have statis-tically significantly less leg pain at the 1-year follow-upevaluation (p , 0.03), and less peak back pain at 2 years(p , 0.04). The patients who underwent circumferentialfusion were found to have a higher PLF rate (92%) thanthe patients treated with PLF with pedicle screws (80%) (p , 0.04) when fusion status was evaluated based on sta-tic plain x-ray films. The circumferential fusion group hadan 82% interbody fusion rate. The plain x-ray film find-ings for the remaining 18% of the circumferential groupwere described as “ambiguous.” The repeated operationrate was significantly lower in the circumferential group(7%) than in the PLF group (22%) (p , 0.009). This paperprovides Class III medical evidence supporting the role ofinterbody grafts in improving arthrodesis rates because ofthe lack of flexion–extension views or CT scans to supple-ment the static radiographs. The medical evidence support-ing the role of interbody grafts in improving outcome withrespect to back and leg pain is considered Class II despitethe randomized design of the study because the differencein the improvement between the surgical groups in the pri-mary outcome measure did not reach statistical signifi-cance. This information must be considered in light of thefact that similar measures of back and leg pain were notsignificantly different between treatment groups at othertime points.

Fritzell and colleagues13,14 performed a randomized,prospective, multicenter trial involving 294 patients withchronic low-back pain due to DDD at one or two levels.Patients were assigned to one of four treatment groups.Patients in Group 1 (73 patients) underwent a noninstru-mented PLF. Those in Group 2 (74 patients) were treatedwith PLF with pedicle screw fixation; patients in Group 3were treated with interbody arthrodesis supplemented withpedicle screw fixation (56 of these patients underwentALIF with pedicle screws, 19 of these patients underwentPLIF with PLF and pedicle screws). Group 4 was treatednonsurgically. Ninety-one percent of patients were avail-

able for follow up by an independent observer. Although allsurgical groups did substantially better than the nonsurgicalgroup, there were no statistically significant differences in ODI, Low Back Pain Questionaire, Million VAS, andGeneral Function Score between the surgical groups. Theearly complication rate was 6% in Group 1, 16% in Group2, and 31% in Group 3. The fusion rate was evaluated byplain radiographs (without flexion–extension views) andwas 72% in Group 1, 87% in Group 2, and 91% in Group3. They concluded that all surgical groups had similar func-tional outcomes, but they noted that their study did lackpower to detect a difference in functional outcome betweenthe surgical groups.13,14 There was an increase in the fusionrate in the instrumented group and in the interbody groupcompared with the noninstrumented group (p = 0.004).This paper provides Class III medical evidence supportingthe beneficial effects of instrumentation and interbodygrafts on fusion rates because of its reliance on static radio-graphs. Because of the sample size, the medical evidenceagainst a beneficial effect on functional outcome is alsoconsidered Class III.

The same authors analyzed their data with respect tocomplication rates and found that overall complicationrates were higher in the instrumented PLF and interbodygroups compared with the noninstrumented PLF group.12

The early complication rate was 6% in the PLF group,18% in the PLF with screw group, and 31% in the 360˚fusion group (p = 0.001). There was no significant differ-ence in the reoperation rate between the interbody groupand the PLF with pedicle screw group. Twenty of the 27reoperations were performed because of “hardware dis-comfort” or the patient’s desire for device removal. Thesereoperations would appear to be unrelated to the use of aninterbody implant. Seventeen of the 29 complications re-ported in the 360˚ fusion group did not necessarily resultfrom the interbody procedure itself. These complicationsincluded donor site pain, pressure sores, and screw mal-position. Four complications were specifically related tothe anterior approach: two iliac vein lacerations and twosympathetic nerve injuries. There were seven instances ofnew nerve root pain, two of which required reoperationwithin 2 years. The 2-year follow-up complication ratewas 12% in the PLF group, 22% in the PLF with screwsgroup, and 40% in the 360˚ group (p = 0.0003). This com-plication rate includes reoperations for instrumentationremoval, whether or not the removal was performed be-cause of any problems associated with the instrumenta-tion. The only delayed complication reported in the inter-body group was continued donor site pain in the patientswho underwent ALIF. The lack of beneficial effect onfunctional outcome, along with the higher complicationrate associated with the circumferential procedures maybe interpreted as evidence against the use of circumferen-tial procedures as a means to improve patient outcomes.

Greenough, et al.,17 performed a nonrandomized studyin which a cohort of 135 patients treated with PLFs withpedicle screws were compared with a historical controlcohort of 151 patients treated with ALIF by the same sur-geon. Fusion assessment was performed based on plainradiographs, occasionally supplemented with flexion–ex-tension films. Outcome was measured using a patient sat-isfaction score as well as the LBO score. The authors di-chotomized the LBO data by considering a score of


Interbody techniques

693




TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

lum

bar

inte

rbod

y fu

sion

in

pati

ents

wit

h de

gene

rati

ve l

umba

r di

seas

e*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Lin

son

& W

illia

ms,

199

1II

IR

etro

spec

tive

stud

y of

51

patie

nts:

Gro

up I

(17

) st

and-

alon

e A

LIF

, Gro

up I

I (1

8) s

tand

-alo

ne A

LIF

red

o,

Sugg

estio

n of

incr

ease

d fu

sion

rat

e w

/ pos

teri

or

Gro

up I

II (

16) A

LIF

w/ H

arri

ngto

n or

Kno

dt r

od &

PL

F. M

ean

FU w

as 2

5, 2

1, a

nd 2

6 m

os, r

espe

ctiv

ely.

in

stru

men

tatio

n 1

AL

IF.

Ass

essm

ent w

as p

erfo

rmed

w/ O

DI

ques

tionn

aire

. Fus

ion

asse

ssed

w/ d

ynam

ic x

-ray

s. P

seud

arth

rosi

sra

te w

as G

roup

I 2

4%, G

roup

II

17%

, Gro

up I

II 1

3% (

p va

lue

not g

iven

).Y

ashi

ro, e

t al.,

199

1II

IN

onra

ndom

ized

ret

rosp

ectiv

e co

hort

stu

dy. G

roup

I h

ad P

LF

1PS

(28

pat

ient

s) &

Gro

up I

I un

derw

ent

PLIF

w/ P

S pr

ovid

es in

crea

sed

fusi

on r

ates

w/ s

i-PL

IF 1

PS (

30 p

atie

nts)

. Mea

n FU

25

mos

. Rad

iogr

aphi

c FU

w/ p

lain

film

s un

til f

usio

n w

as e

stab

lishe

d m

ilar

com

plic

atio

n ra

tes

to P

LF

1PS

.(m

ean

11 m

os P

LF,

6 m

os P

LIF

). T

he o

p tim

e w

as .

in th

e PL

IF g

roup

(p

,0.

05).

Fu

sion

occ

urre

d in

67%

of

Gro

up I

& 9

2% o

f G

roup

II

(p ,

0.01

) fo

r 1

leve

l & 5

6% a

nd 1

00%

for

2le

vel c

ases

(p

,0.

05).

Dif

fere

nce

in c

ompl

icat

ion

rate

s w

as n

ot s

tatis

tical

ly s

igni

fica

nt. P

LIF

gra

ft

mat

eria

l not

spe

cifi

ed.

Rom

pe, e

t al.,

199

5II

I85

pat

ient

s w

/ deg

ener

ativ

e lu

mba

r sp

ine

dise

ase

& r

adio

logi

cal e

vide

nce

of in

stab

ility

. 55

had

PLF

w/

Out

com

e of

PL

F 1

PS w

as th

e sa

me

as P

LIF

1pe

dicl

e sc

rew

s; 3

0 ha

d PL

IF &

PL

F w

/ scr

ews.

The

re w

ere

no d

iffe

renc

es in

out

com

e (n

onva

lidat

ed

PLF

1PS

.ou

tcom

e m

easu

re)

btw

n th

e gr

oups

. The

re w

ere

few

er c

ases

of

hard

war

e fa

ilure

in P

LIF

gro

up.

Hac

ker,

1997

III

Ret

rosp

ectiv

e 75

pat

ient

s. 5

4 PL

IF s

tand

-alo

ne B

AK

, 21

AL

IF w

/ allo

graf

t 1PL

F w

/ aut

ogra

ft (

no

Stan

d-al

one

PLIF

mor

e co

st e

ffec

tive

than

360

˚ sc

rew

s). M

in 2

-yr

FU. O

p tim

e (p

,0.

0001

), E

BL

(p ,

0.00

01),

& s

tay

(p ,

0.00

03)

wer

e si

g-fu

sion

. Out

com

es &

fus

ion

rate

are

sim

ilar.

nifi

cant

ly d

ecre

ased

in P

LIF

gro

up. O

utco

me

mea

sure

s w

ere

inco

mpl

etel

y re

port

ed. M

any

patie

nts

lost

to F

U.

Gre

enou

gh, e

t al.,

199

8II

IPr

ospe

ctiv

e no

nran

dom

ized

coh

ort o

f 13

5 pa

tient

s PL

F 1

PS. F

usio

n as

sess

men

t w/ x

-ray

s w

/ dyn

amic

A

LIF

ass

ocia

ted

w/ h

ighe

r fu

sion

rat

es &

pat

ient

film

s on

som

e pa

tient

s. O

utco

me

mea

sure

is L

BO

sco

re, &

mod

ifie

d so

mat

ic p

erce

ptio

n qu

estio

nnai

re

satis

fact

ion

than

PL

F 1

PS.

& Z

ung

depr

essi

on s

cale

(2

yrs

min

FU

). R

esul

ts in

feri

or to

thos

e ob

tain

ed in

his

tori

cal c

ohor

t of

pa-

tient

s tr

eate

d w

/ AL

IF.

Vam

vani

j, et

al.,

199

8In

stru

men

ta-

56 c

onse

cutiv

e pa

tient

s, o

p w

/ 1 o

f 4

lum

bar

fusi

on p

roce

dure

s. O

utco

mes

det

erm

ined

by

post

op p

ain

AL

IF r

equi

res

post

fixa

tion

to e

qual

res

ults

of

tio

n II

ques

tionn

aire

s, in

depe

nden

t clin

ical

ass

essm

ent,

& r

adio

grap

hic

eval

uatio

n. S

imul

tane

ous

AL

IF w

/PL

F 1

PS.

Inte

rbod

y II

IB

AK

cag

e &

pos

teri

or f

acet

fus

ion

prov

ided

the

high

est r

ate

of f

usio

n (8

8%)

& c

linic

al s

atis

fact

ion

(63%

). P

ain

scor

es s

igni

fica

ntly

low

er th

an f

acet

scr

ew–a

ugm

ente

d PL

F &

AL

IF w

/ fib

ula

allo

graf

t, bu

tno

t sig

nifi

cant

ly d

iffe

rent

fro

m P

Sin

stru

men

ted

post

erol

ater

al f

usio

n. P

atie

nts

w/l

umba

r fu

sion

had

be

tter

clin

ical

out

com

es &

a b

ette

r ch

ance

of

wor

k re

sum

ptio

n.B

arne

s, e

t al.,

200

1II

IR

etro

spec

tivel

y re

view

ed a

ser

ies

of 3

5 pa

tient

s w

/ mec

hani

cal l

ow-b

ack

or 1

- to

2-l

evel

dis

coge

nic

pain

; PL

IF w

/ PS

do b

ette

r th

an s

tand

-alo

ne A

LIF

w/ r

e-23

trea

ted

w/ P

LIF

& P

Ss, 1

2 w

/ non

inst

rum

ente

d A

LIF

. All

had

inte

rbod

y fu

sion

w/ T

CB

Ds.

FU

was

sp

ect t

o fu

sion

& o

utco

me

whe

n T

CB

Ds

are

used

.87

% f

or th

e PL

IF g

roup

& 6

7% f

or A

LIF

w/ a

mea

n of

1 y

r. 70

% s

atis

fact

ory

outc

ome

was

not

ed

in P

LIF

pat

ient

s &

38%

in A

LIF

pat

ient

s. O

sseo

us f

usio

n w

as p

rese

nt in

95%

of

patie

nts

in th

e PL

IFgr

oup

& in

13%

of

the

AL

IF g

roup

. Sm

all s

ampl

e si

ze, d

iffe

ring

pat

ient

sel

ectio

n cr

iteri

a, &

larg

e“l

ost t

o FU

” gr

oup

limit

valu

e of

stu

dy.

Chi

tnav

is, e

t al.,

200

1II

I50

PL

IF p

atie

nts

w/ c

arbo

n fi

ber

cage

s fo

r D

DD

. In

40 p

atie

nts

(80%

) w

/ sta

nd-a

lone

PL

IF, 1

0 w

/ PSs

. PS

aug

men

tatio

n of

car

bon

fibe

r ca

ge P

LIF

may

/may

Prol

o sc

ale.

Fus

ion

outc

ome

base

d on

est

ablis

hed

clas

sifi

catio

n w

/ dyn

amic

x-r

ays.

2/3

of

patie

nts

no

t im

prov

e ou

tcom

e or

fus

ion

rate

.ex

peri

ence

d go

od o

r ex

celle

nt o

utco

mes

(Pr

olo

scor

e $

8) a

t ear

ly &

late

FU

. No

diff

eren

ce in

clin

ical

outc

ome

btw

n th

ose

in w

hom

PSs

wer

e an

d w

ere

not i

mpl

ante

d (p

= 0

.83,

Man

n–W

hitn

ey U

-tes

t).

Una

ble

to d

raw

sig

nifi

cant

con

clus

ion

rega

rdin

g us

e of

PS

due

to s

mal

l sam

ple

size

.H

ee, e

t al.,

200

1II

IR

etro

spec

tive

164

patie

nts.

53

had

sam

e-da

y an

teri

or–p

oste

rior

fus

ion

(Gro

up 1

), a

nd 1

11 h

ad T

LIF

T

LIF

was

bet

ter

than

360

˚ fu

sion

w/ r

egar

d to

fus

ion

(G

roup

2).

Mea

n op

tim

e (p

,0.

0001

) an

d L

OS

(p ,

0.00

01)

wer

e si

gnif

ican

tly lo

nger

for

ant

erio

r–ra

te &

cos

t, bu

t the

360

˚ fu

sion

rat

e w

as lo

wer

po

ster

ior

patie

nts.

Ave

rage

blo

od lo

ss w

as g

reat

er f

or a

nter

ior–

post

erio

r pa

tient

s (p

,0.

01).

Hig

her

than

exp

ecte

d by

the

auth

ors.

com

plic

atio

n ra

tes

in a

nter

ior–

post

erio

r (p

,0.

004)

. FU

was

$2

yrs.

PS

1PL

F w

as p

er-

form

ed f

or a

ll A

LIF

s. T

he T

LIF

s (1

08)

wer

e pe

rfor

med

w/ c

age,

aut

ogra

ft, P

S, &

PLF.

Rad

iogr

aphs

w

ere

asse

ssed

by

Ger

tzbe

in m

etho

d. P

seud

arth

rosi

s ra

te w

as 1

5% fo

r AL

IF &

6%

for T

LIF

gro

up (p

= 0

.07)

.H

umph

reys

, et a

l., 2

001

IIN

onra

ndom

ized

ret

rosp

ectiv

e co

mpa

riso

n of

40

TL

IFs

w/ 1

cag

e, a

utog

raft

, PS,

& P

LF

& 3

4 PL

IFs

w/

PLIF

& T

LIF

com

plic

atio

n ra

tes

equi

vale

nt f

or 1

- 1

cage

& P

S/PL

F. N

o ou

tcom

e m

easu

re o

r fu

sion

ass

essm

ent p

erfo

rmed

. le

vel p

roce

dure

s. T

LIF

res

ulte

d in

few

er c

ompl

i-ca

tions

& a

low

er E

BL

for

2-le

vel c

ases

com

pare

d w

/ PL

IF.

Con

tinu

ed




695

TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Scho

ffer

man

, et a

l., 2

001

IPr

ospe

ctiv

e ra

ndom

ized

com

pari

son

of A

LIF

1PS

1PL

F (2

6) v

s A

LIF

1PS

(22

). FU

ave

rage

d 35

mos

. PL

F is

unn

eces

sary

in th

e co

ntex

t of A

LIF

1PS

&

Out

com

e m

easu

res

incl

ude

NR

S &

OD

I. Fu

sion

was

eva

luat

ed w

/ sta

tic p

lain

rad

iogr

aphs

. 68%

pse

ud-

is a

ssoc

iate

d w

/ hig

her

bloo

d lo

ss, O

R ti

me,

&

arth

rosi

s ra

te n

oted

in P

LF.

No

sign

ific

ant d

iffe

renc

e in

fus

ion

rate

of

the

inte

rbod

y gr

aft b

twn

grou

ps,

cost

.bu

t AL

IF 1

PS g

roup

tend

ed to

war

d hi

gher

fus

ion

rate

. AL

IF 1

PS h

ad s

hort

er o

p tim

e, E

BL

, LO

S,

& c

ost.

All

diff

eren

ces

wer

e st

atis

tical

ly s

igni

fica

nt.

Chr

iste

nsen

, et a

l., 2

002

inte

rbod

yPr

ospe

ctiv

ely

rand

omiz

ed s

tudy

of

148

patie

nts

w/ L

BP.

Com

pare

d PL

F w

/ scr

ews

vs A

LIF

(B

rant

igan

36

0˚ f

usio

n ha

d a

sign

ific

antly

hig

her

fusi

on r

ate,

II

for

out

com

eca

ge)

1PS

1PL

F. O

utco

mes

: DPQ

, the

LB

PR. T

ende

ncy

tow

ard

bette

r ov

eral

l fun

ctio

nal o

utco

me

bette

r sa

gitta

l lor

dosi

s, lo

wer

reo

p ra

te, &

a te

n-II

I fo

r fu

sion

for

patie

nts

w/ c

ircu

mfe

rent

ial p

roce

dure

(p

,0.

08),

& p

atie

nt g

roup

had

sig

nifi

cant

ly le

ss le

g pa

inde

ncy

for

bette

r cl

inic

al o

utco

me

than

PL

F 1

PS

at th

e 1

yr F

U (

p ,

0.03

) &

less

pea

k ba

ck p

ain

at 2

yrs

(p

,0.

04).

The

cir

cum

fere

ntia

l fus

ion

pa-

at 2

-yr

FU.

tient

s sh

owed

a h

ighe

r PL

F ra

te (

92%

) th

an th

e po

ster

olat

grou

p (8

0%)

(p ,

0.04

). T

he in

terb

ody

fu-

sion

rat

e w

as 8

2% s

olid

w/ 1

8% a

mbi

guou

s on

pla

in x

-ray

(no

ndyn

amic

). R

epea

ted

op r

ate

was

sig

-ni

fica

ntly

low

er in

cir

cum

fere

ntia

l gro

up (

7%)

(,0.

009)

than

in p

oste

rola

ter

grou

p (2

2%).

DeB

erar

d, e

t al.,

200

2II

IR

etro

spec

tive

coho

rt s

tudy

of

185

patie

nts

w/ P

LF

& 1

85 p

atie

nts

w/ B

AK

(65

% A

LIF

& 3

5% P

LIF

).

Inte

rbod

y gr

oup

did

bette

r th

an P

LF

grou

p w

/ FU

by

char

t rev

iew

at 2

yrs

. re

spec

t to

fusi

on r

ate

(94

vs 7

4%)

& f

unct

iona

l ou

tcom

e (R

olan

d &

Mor

ris

8.8

vs 1

1.4)

.Pr

adha

n, e

t al.,

200

2II

Ret

rosp

ectiv

e st

udy

w/ 1

22 p

atie

nts

exam

inin

g A

LIF

vs

PLF.

FU

at 2

2 &

26

mos

. Fus

ion

asse

ssed

w/

AL

IF is

ass

ocia

ted

w/ l

ess

mor

bidi

ty th

an P

LF.

dyna

mic

x-r

ay &

CT

was

use

d fo

r am

bigu

ous

case

s. N

o st

anda

rd c

linic

al o

utco

me.

AL

IF le

ss b

lood

loss

(p

,0.

01)

& s

tay.

No

diff

eren

ce in

com

plic

atio

n ra

te, f

usio

n ra

te, o

r cl

inic

al o

utco

me.

Fritz

ell,

et a

l., 2

002

outc

ome

Pros

pect

ran

dom

ized

mul

ticen

ter

stud

y of

294

pat

ient

s. G

roup

1 (

PLF,

73)

, Gro

up 2

(PL

F w

/ scr

ews,

74)

, C

ompa

riso

n of

non

inst

rum

ente

d fu

sion

vs

PLF

1PS

C

lass

Ian

d G

roup

3 (

PLF

w/ s

crew

s &

inte

rbod

y fu

sion

, 75;

AL

IF [

56]

or P

LIF

[19

]). G

roup

4 n

onop

(72

) vs

360

˚. N

o di

ffer

ence

s in

out

com

e. 3

60˚

fusi

on

fusi

on

FU 2

01 (

91%

) of

222

pat

ient

s af

ter

2 yr

s. A

ll su

rgic

al te

chni

ques

red

uced

pai

n &

dec

reas

ed d

isab

ility

led

to th

e hi

ghes

t fus

ion

rate

s &

com

plic

atio

n C

lass

III

subs

tant

ially

but

no

sign

ific

ant d

iffe

renc

es w

ere

foun

d am

ong

the

surg

ical

gro

ups.

Gro

ups

2 &

3 c

on-

rate

s fo

llow

ed b

y PL

F 1

PS.

sum

ed s

igni

fica

ntly

mor

e re

sour

ces.

Ear

ly c

ompl

icat

ion

rate

was

6%

in G

roup

1, 1

6% in

Gro

up 2

, &31

% in

Gro

up 3

. The

fus

ion

rate

(pl

ain

radi

ogra

phy)

, was

72%

in G

roup

1, 8

7% in

Gro

up 2

, & 9

1%

in G

roup

3. T

he h

igh

com

plic

atio

n ra

te o

f in

stru

men

tatio

n is

rel

ated

to 9

pat

ient

s w

/ scr

ew r

emov

al f

or

new

ons

et r

adic

ulop

athy

. At t

he ti

me

of r

evis

ion

only

3 w

ere

foun

d to

hav

e a

cort

ical

wal

l bre

ach.

Fr

itzel

l, et

al.,

200

3 ou

tcom

e (c

om-

Mul

ticen

ter

pros

pect

ive

rand

omiz

ed s

tudy

of

211

patie

nts.

PL

F (7

1), P

LF

1V

SP(6

8), s

truc

tura

l aut

o-36

0˚ f

usio

n w

as a

ssoc

iate

d w

/ hig

hest

com

plic

atio

n

plic

atio

n gr

aft i

nter

body

1 P

LF

1PS

(72

). N

o si

gnif

ican

t dif

fere

nce

in O

DI,

VA

S, G

ener

al F

unct

ion

Scor

e, o

r ra

te.

rate

) I

Zun

g de

pres

sion

sca

le a

t 2 y

rs (

p .

0.05

). P

ower

to d

etec

t suc

h a

diff

eren

ce w

as lo

w. E

arly

com

plic

a-tio

n ra

tes

wer

e 6,

18,

and

31%

, res

pect

ivel

y (p

= 0

.001

). T

otal

com

plic

atio

n ra

te a

fter

2 y

rs in

the

PLF

grou

p w

as 1

2 vs

22%

in th

e PS

gro

up, &

40%

in th

e 36

0˚ g

roup

(p

= 0.

0003

). R

eop

rate

s w

ere

6, 2

2, &

17%

, res

pect

ivel

y (p

= 0

.02)

. No

fusi

on te

chni

que

prod

uced

a s

uper

ior

clin

ical

out

com

e.

Zha

o, e

t al.,

200

2II

IPr

ospe

ctiv

e ra

ndom

ized

of

1 (1

3) v

s 2

(12)

sta

nd-a

lone

PL

IF B

AK

cag

es in

L4–

5 de

gene

rativ

e sp

ondy

lo-

1 ca

ge m

ay o

r m

ay n

ot b

e as

eff

ectiv

e as

2 in

sta

nd-

listh

esis

. FU

2 y

rs. B

lood

loss

& o

p tim

e si

gnif

ican

tly le

ss in

1 c

age.

LO

S no

t dif

fere

nt. C

osts

less

in 1

al

one

PLIF

w/ r

espe

ct to

fus

ion

rate

s &

out

com

e.ca

ge. F

usio

n ba

sed

on d

ynam

ic x

-ray

. Mod

ifie

d Z

debl

ick

outc

ome

scor

e. N

o di

ffer

ence

btw

n gr

oups

.V

ery

smal

l sam

ple

size

pre

clud

es d

efin

itive

con

clus

ion.

*E

BL

= e

stim

ated

blo

od lo

ss; F

U =

fol

low

up;

NR

S =

Num

eric

al R

atin

g Sc

ale;

PS

= p

edic

le s

crew

.


greater than 50 as good or excellent and a score of 49 orless as fair or poor. Using this dichotomy, the authors re-ported that 40% of the ALIF group had good-to-excellentoutcomes compared with 19% of the PLF group (p ,0.001, chi square). When comparing the subgroup of pa-tients who were not involved in Workers’ Compensationclaims, the good-to-excellent outcome in the ALIF groupwas 68% compared with 27% in the PLF group (p , 0.01,chi square). This paper provides Class III medical evi-dence supporting the beneficial effect of an interbodygraft on patient outcome.

Yashiro, et al.,34 performed a nonrandomized retrospec-tive cohort study of 58 patients treated with lumbar spinalfusion at one or two levels for the treatment of low-backpain due to DDD, spondylolisthesis, or fracture (eightpatients). Patients in Group I underwent PLF with pediclescrews (28 patients), and those in Group II underwentPLIF and pedicle screws (30 patients). There was no men-tion of whether PLF was added to the PLIF group. Themean follow-up period was 25 months. Radiographic fol-low up was performed using static radiographs; no flex-ion–extension views were used to measure fusion. Theoperating time was longer in the PLIF group (p , 0.05).Fusion occurred in 67% of Group I and 92% of Group IIpatients for single-level cases (p , 0.01). Fusion occurredin 56% of Group I and 100% of Group II patients for two-level cases (p , 0.05). The difference in complicationrates between the groups was not statistically significant.This paper provides Class III medical evidence supportingthe use of interbody grafts to increase lumbar fusion rates.

Pradhan and colleagues27 performed a retrospective re-view of 122 patients who were divided into two treatmentgroups. Group I consisted of 58 patients who were treatedwith lumbar ALIF with BAK cages, and Group II consist-ed of 64 patients who were treated with PLF with pediclescrew fixation. The follow-up period was 22 months forGroup I and 26 months for Group II. Fusion was assessedbased on flexion–extension x-ray films, and CT scanningwas used for cases without clear-cut radiographic fusion.There was evidence of radiographic fusion in 95% of theGroup I patients and in 92% of the Group II patients; how-ever, this difference was not found to be significant. TheALIF cohort had a lower operative blood loss, shorteroperative time, and shorter duration LOS (p , 0.01).There were no significant differences in complicationrates or clinical outcomes between the groups. Althoughthis paper provides Class III medical evidence indicatingthat placement of an interbody graft through a stand-aloneALIF technique does not improve fusion rates comparedwith PLF, the small size of the treatment groups in thisstudy makes any statement regarding functional outcomessuspect. The ALIF group was reported to have a shorterLOS, less blood loss, and less exposure to anestheticagents.

DeBerard, et al.,11 performed a retrospective cohort stu-dy in which 185 patients receiving Workers’ Compen-sation were treated with PLF were compared with 185patients treated with LIF for intractable low-back pain. Anoutcome survey was conducted a mean of 5 years aftersurgery. The authors were able to contact approximately70% of the surgical cohort, which encompassed approxi-mately 55% of Workers’ Compensation patients. Ar-throdesis rates, patient satisfaction, patient function, and

overall health were better for the LIF cohort. The resultssuggest improved arthrodesis rates and outcomes for pa-tients who received Workers’ Compensation and under-went interbody fusion compared with similar patientsundergoing PLF. Because of the retrospective chart re-view nature of the study, the fact that patients in eachgroup were from different states (Minnesota comparedwith Utah), the large number of patients lost in the follow-up process, and the variable means used to determinefusion, this medical evidence supporting the role of inter-body grafts for improving outcomes and fusion rates isconsidered Class III.

Rompe and colleagues29 retrospectively reviewed 85patients with degenerative lumbar spine disease and radio-graphic evidence of instability. Fifty-five of the patientswere treated with PLF with pedicle screws; the remaining30 patients were treated with PLIF and PLF with pedi-cle screws. Patients were followed up for a mean of 32months. Of these patients, 86% reported improvements intheir pain symptoms but only 46% reported good-to-excellent overall results. Patients with fair and poor out-comes had undergone significantly more operations on thelumbar spine (p , 0.001) and had a greater extent of pre-operative lumbar kyphosis (p , 0.05) than patients withgood-to-excellent outcomes. In these patients with radio-graphic instability, no improvement in functional outcomewas realized through the addition of an interbody graft.Because of significant differences between the treatmentgroups, the small size of the interbody group, and a non-validated outcome measure, this paper provides Class IIImedical evidence that fails to support the placement of aninterbody graft in addition to PLF.

Additional Class III medical evidence supporting theuse of interbody fusion techniques is provided by numer-ous case series in which excellent results are describedwith ALIF and PLIF procedures performed using a vari-ety of techniques.2–5,7,8,15,16,21–24,26,31,32

Comparison of Interbody Techniques

Several authors have compared different LIF tech-niques with regard to fusion rates and patient outcomes.The most common comparison has been between anteriorand posterior interbody techniques. Schofferman and col-leagues30 performed a prospective, randomized compari-son of 26 patients treated with ALIF with pedicle screwswith PLF (360˚ group) to 22 patients treated with ALIFand pedicle screws without PLF (270˚ group). All ALIFswere performed with a femoral ring allograft filled withcancellous allograft chips. The follow-up period averaged35 months. Outcomes were measured by the NumericalRating Scale and the ODI. Functional outcomes as mea-sured by both scales improved significantly in both groupswithout significant differences between the two groupswhen using either outcome measure. Fusion was evaluat-ed based on flexion–extension plain radiographs. In the360˚ group, the PLF part of the procedure failed to heal68% of the time. There was no significant difference (p =0.6, chi square) in the fusion rate of the interbody graftbetween the groups, although there was a trend favoringthe 270˚ group (77% fusion rate in the 360˚ group com-pared with 89% fusion rate in the 270˚ group). The 270˚group had shorter operating time, less intraoperative blood




loss, and shorter LOS (all p , 0.05). This study providesClass I medical evidence that the addition of PLF to anALIF with pedicle screw construct increases blood loss,LOS, and operating time without any resultant benefit.Although the increased interbody fusion rate noted in the270˚ group was not statistically significant, the magnitudeof the difference is notable and may have reached signifi-cance in an adequately powered study.

Barnes, et al.,1 retrospectively reviewed a series of 35patients with mechanical low-back pain due to DDD atone or two levels. Twenty-three patients were treated withPLIF with TCBDs, PLF, and pedicle screw fixation.Twelve patients were treated with stand-alone ALIF withTCBDs. The follow-up rates were 87% for the PLIF groupand 67% for the ALIF group (mean follow up 1 year).Fusion was assessed based on flexion–extension radio-graphs, and CT scanning was performed to assess fusionin questionable cases. Outcomes were assessed with themodified Prolo scale. Satisfactory outcomes were noted in70% of the PLIF patients and in 38% of the ALIF patients.Osseous fusion was present in 95% of the patients in thePLIF group and in 13% of those in the ALIF group (prob-ability values not provided). Although this paper providesClass III medical evidence supporting the use of PLIFcompared with stand-alone ALIF with TCBDs for betterpatient outcomes and fusion rates, differences in patientselection criteria, small sample size, and the use of a non-validated outcome measure limit the conclusions that canbe drawn.

Similarly, Hacker and colleagues18 performed a retro-spective review of 75 patients. Fifty-four of these patientswere treated with stand-alone PLIF with BAK cages, and21 patients were treated with ALIF with allograft spacersand PLF without screws. A minimum 2-year follow upwas performed. The North American Spine Society out-come study form was used but full data on the outcomewere not provided. Operative time (p , 0.0001), bloodloss (p , 0.0001), and LOS (p , 0.0003) were signifi-cantly decreased for the PLIF group. Fusion assessmentwas performed based on flexion–extension x-ray films.Only 24 of the 54 PLIF patients underwent radiographicfollow up. The authors reported a 4% pseudarthrosis ratein the PLIF group. The pseudarthrosis rate in the ALIFgroup was “approximately 20%” (five patients). Therewas a quicker return to work and closure of Workers’Compensation claims in the PLIF group. This study pro-vides Class III medical evidence suggesting that outcomesand fusion rates were improved using a PLIF as opposedto an ALIF plus PLF without screws; however, the large“lost to follow-up” rate and the incomplete reporting ofthe functional outcome measure decrease the value of thisevidence substantially.

Hee, et al.,19 performed a retrospective review of 164patients. Fifty-three of these patients were treated withsame-day ALIF with PLF and screws (Group 1), and theremaining 111 patients were treated with TLIF (Group 2).In Group 1 patients, ALIF was performed with autograftinside either a carbon fiber or titanium cage (21 patients),structural autograft iliac crest (19 patients), or allograftstrut graft (13 patients). The TLIFs were performed withtitanium cages filled with autograft and supplementedwith pedicle screws and PLF. The mean operating time (p , 0.0001) and LOS (p , 0.0001) were statistically sig-

nificantly longer for Group 1 patients. The average bloodloss was greater in Group 1 patients (p , 0.01). Highercomplication rates were found in Group 1 patients (p ,0.004). The follow-up period in both groups was at least 2years. Fusion assessment was performed based on plainradiographs. The pseudarthrosis rate was 15% for Group1 patients and 6% for Group 2 patients (p = 0.07). Thispaper provides Class III medical evidence supporting anadvantage for the use of a single posterior approach asopposed to a combined anterior–posterior approach for a360˚ fusion.

Class III medical evidence in favor of interbody tech-niques is also provided by Vamvanij, et al.33 These authorsreviewed the data in 56 patients who were treated with oneof four different lumbar fusion procedures. Group I pa-tients underwent ALIF with BAK cage supplemented withfacet screws without PLF. Group II patients underwentfacet screw fixation with PLF; Group III patients under-went pedicle screw fixation with PLF. Group IV patientsunderwent stand-alone ALIF with fibula strut grafts. Si-multaneous anterior interbody fusion by using BAK cageand posterior facet fixation provided the highest rate offusion (88%) and clinical satisfaction (63%). Patients inwhom successful lumbar fusion was achieved had betterclinical outcomes and a better chance of work resumption.

Other authors have reported comparisons of differenttechniques of achieving interbody fusion with posteriorapproaches. For example, Zhao, et al.,37 performed a pros-pective randomized study in which they compared the useof one stand-alone PLIF BAK cage (13 patients) with twostand-alone PLIF BAK cages (12 patients) to treat L4–5degenerative Grade I spondylolisthesis. There was a min-imum follow up of 2 years. Intraoperative blood loss andoperative time were significantly less in the one-cagegroup (p , 0.01); LOS was not different between thegroups (p , 0.01). Fusion was assessed based on flex-ion–extension x-ray films, and fusion rates were 92% inboth groups (no significant difference; p . 0.9, x2 =0.0035). Outcomes were assessed with a scale adoptedfrom Zdeblick.35 No significant difference in clinical out-comes was found between the groups. Given the smallsample size, no meaningful conclusions can be drawnfrom this study regarding clinical or radiographic out-come. Humphreys and colleagues20 performed a nonran-domized retrospective comparison of 40 patients treatedwith TLIF with a single cage filled with autograft and sup-plemented by pedicle fixation and PLF compared with 34patients undergoing PLIF with a single cage filled withautograft and supplemented with pedicle screws with PLF.For single-level fusion, blood loss, operative time, andLOS were no different between the groups. For two-levelfusion, however, the TLIF patients had less blood loss (p , 0.01) than the PLIF patients. No outcome measuresor fusion assessment were provided.

The requirement for supplemental instrumentation inaddition to interbody grafting has been addressed in a fewstudies that provide Class III medical evidence. Barnes, etal.,1 recommend pedicle screw fixation as an adjunct toALIF or PLIF with TCBDs. Similarly, in an older study,Linson and Williams25 found that posterior instrumenta-tion improved fusion rates following interbody grafting.These authors performed a retrospective study in 51 pa-tients who were divided into three groups. Patients in



697


Group I (17) were treated with stand-alone ALIF, patientsin Group II (18) were treated with stand-alone ALIF fol-lowing previous lumbar surgery, and Group III patients(16) underwent ALIF with Harrington or Knodt rod fixa-tion and PLF. The mean follow up was 25, 21, and 26months, respectively. The ODI questionnaire was used toassess outcome. Fusion was assessed based on flexion–ex-tension radiographs. The pseudarthrosis rate was 24 % inGroup I, 17% in Group II, and 13% in Group III. Con-versely, other authors of retrospective series did not detectimproved fusion rates with the addition of posterior instru-mentation. Chitnavis and colleagues9 retrospectively stud-ied a cohort of 50 patients with multiple recurrent discherniations (49 one level, one two level). These patientswere treated with PLIF by using carbon fiber cages. Tenof these patients underwent pedicle screw fixation be-cause of suspected instability or because they were treat-ed early in the series. Overall results in both groups wereexcellent; however, as the authors have stated, a meaning-ful comparison between the pedicle screw and nonpediclescrew groups is not possible.

Authors of several large case series of stand-aloneALIF with threaded titanium cages report excellent resultswith regard to patient outcomes and fusion rates.5–8,28,36 Forexample, Burkus, et al.,7 published a large multicenterseries on patients who underwent stand-alone ALIF witha threaded titanium cage with autograft or rrecombinanthuman bone morphogenetic protein–2. The results in bothgroups with respect to fusion rates (89–95%) and clinicaloutcome (ODI, back pain, leg pain, and patient satisfac-tion) were excellent.

Summary

The majority of reviewed medical evidence suggests thatinterbody techniques are associated with higher fusion ratescompared with PLF when applied to patients with low-back pain due to DDD limited to one or two levels. The evi-dence is generally of poor quality and retrospective innature. Conflicting evidence exists supporting the role ofinterbody graft placement for improvement of functionaloutcomes; however, there is no Class I or II evidence to sug-gest that the use of an interbody graft is associated withworse outcomes, and Class II evidence exists to suggest thatoutcomes are improved. Complication rates of interbodygraft placement, particularly of circumferential procedures,are higher in most series. Many complications, however, areassociated with pedicle screw fixation and not with inter-body graft placement per se. In the context of a single-levelstand-alone ALIF or ALIF with posterior instrumentation,there does not appear to be a substantial benefit to the addi-tion of a PLF. The addition of a PLF to a construct thatalready includes an interbody graft is, however, associ-ated with increased costs and complications. Therefore, al-though the addition of supplemental fixation (a 270˚ fu-sion) may be necessary for biomechanical reasons, it maynot be appropriate to subject the patient to the morbidity ofa full posterior exposure for placement of graft material.

Significant differences in clinical outcomes between thevarious interbody techniques have not been convincinglydemonstrated. No general recommendation can thereforebe made regarding the technique that should be used toachieve interbody fusion.


Future studies focusing on patient outcomes are re-quired to establish whether the increased fusion rates seenwith interbody techniques are truly associated with im-proved functional outcomes. Application of reliable, valid,and responsive outcome measures in a multicenter random-ized trial would serve to answer this question. In terms ofthe techniques used to achieve an interbody arthrodesis, itis likely that certain techniques will be more applicable to different patient populations. Future studies should be focused on evaluating the individual techniques within specific patient populations. Well-designed cohort studieswould provide needed Class II medical evidence. Ran-domized studies would need to include adequate numbersof patients to ensure sufficient power to be able to assesswhether the incremental improvement achieved with in-terbody techniques is clinically significant.

References

1. Barnes B, Rodts GE, McLaughlin MR, et al: Threaded corticalbone dowels for lumbar interbody fusion: over 1-year mean fol-low up in 28 patients. J Neurosurg 95:1–4, 2001

2. Branch CL, Branch CL Jr: Posterior lumbar interbody fusionwith the keystone graft: technique and results. Surg Neurol 27:449–454, 1987

3. Brantigan JW, Steffee AD: A carbon fiber implant to aid inter-body lumbar fusion. Two-year clinical results in the first 26 pa-tients. Spine 18:2106–2107, 1993

4. Brantigan JW, Steffee AD, Lewis ML, et al: Lumbar interbodyfusion using the Brantigan I/F cage for posterior lumbar inter-body fusion and the variable pedicle screw placement system:two-year results from a Food and Drug Administration investi-gational device exemption clinical trial. Spine 25:1437–1446,2000

5. Burkus JK: Intervertebral fixation: clinical results with anteriorcages. Orthop Clin North Am 33:349–357, 2002

6. Burkus JK, Dorchak JD, Sanders DL: Radiographic assessmentof interbody fusion using recombinant human bone morpho-genetic protein type 2. Spine 28:372–377, 2003

7. Burkus JK, Gornet MF, Dickman CA, et al: Anterior lumbarinterbody fusion using rhBMP-2 with tapered interbody cages.J Spinal Disord Tech 15:337–349, 2002

8. Burkus JK, Transfeldt EE, Kitchel SH, et al: Clinical and radio-graphic outcomes of anterior lumbar interbody fusion usingrecombinant human bone morphogenetic protein-2. Spine 27:2396–2408, 2002

9. Chitnavis B, Barbagallo G, Selway R, et al: Posterior lumbarinterbody fusion for revision disc surgery: review of 50 cases inwhich carbon fiber cages were implanted. J Neurosurg (Spine2) 95:190–195, 2001


11. DeBerard MS, Colledge AL, Masters KS, et al: Outcomes ofposterolateral versus BAK titanium cage interbody lumbar fu-sion in injured workers: a retrospective cohort study. J SouthOrthop Assoc 11:157–166, 2002

12. Fritzell P, Hagg O, Nordwall A: Complications in lumbar fu-sion surgery for chronic low back pain: comparison of threesurgical techniques used in a prospective randomized study. Areport from the Swedish Lumbar Spine Study Group. EurSpine J 12:178–189, 2003

13. Fritzell P, Hagg O, Wessberg P, et al: 2001 Volvo Award Win-ner in Clinical Studies: Lumbar fusion versus nonsurgical treat-




ment for chronic low back pain: a multicenter randomized con-trolled trial from the Swedish Lumbar Spine Study Group. Spine26:2521–2534, 2001

14. Fritzell P, Hagg O, Wessberg P, et al: Chronic low back painand fusion: a comparison of three surgical techniques: a pros-pective multicenter randomized study from the Swedish lumbarspine study group. Spine 27:1131–1141, 2002

15. Geerdes BP, Geukers CW, van Erp WF: Laparoscopic spinalfusion of L4-l5 and L5–S1. Surg Endosc 15:1308–1312, 2001

16. Gill K, Blumenthal SL: Posterior lumbar interbody fusion. A 2-year follow-up of 238 patients. Acta Orthop Scand Suppl251:108–110, 1993


18. Hacker RJ: Comparison of interbody fusion approaches for dis-abling low back pain. Spine 22:660–666, 1997

19. Hee HT, Castro FP Jr, Majd ME, et al: Anterior/posterior lum-bar fusion versus transforaminal lumbar interbody fusion: anal-ysis of complications and predictive factors. J Spinal Disord14:533–540, 2001

20. Humphreys SC, Hodges SD, Patwardhan AG, et al: Compa-rison of posterior and transforaminal approaches to lumbar in-terbody fusion. Spine 26:567–571, 2001

21. Hutter CG: Spinal stenosis and posterior lumbar interbody fu-sion. Clin Orthop Relat Res 193:103–114, 1985

22. Lin PM: Radiographic evidence of posterior lumbar interbodyfusion with an emphasis on computed tomographic scanning.Clin Orthop Relat Res 242:158–163, 1989

23. Lin PM: A technical modification of Cloward’s posterior lum-bar interbody fusion. Neurosurgery 1:118–124, 1977

24. Lin PM, Cautilli RA, Joyce MF: Posterior lumbar interbody fu-sion. Clin Orthop Relat Res 180:154–168, 1983

25. Linson MA, Williams H: Anterior and combined anteroposteri-or fusion for lumbar disc pain. A preliminary study. Spine 16:143–145, 1991

26. Ma GW: Posterior lumbar interbody fusion with specialized in-struments. Clin Orthop Relat Res 193:57–63, 1985

27. Pradhan BB, Nassar JA, Delamarter RB, et al: Single-level lum-bar spine fusion: a comparison of anterior and posterior ap-proaches. J Spinal Disord Tech 15:355–361, 2002

28. Regan JJ, Yuan H, McAfee PC: Laparoscopic fusion of thelumbar spine: minimally invasive spine surgery. A prospective

multicenter study evaluating open and laparoscopic lumbar fu-sion. Spine 24:402–411, 1999

29. Rompe JD, Eysel P, Hopf C: Clinical efficacy of pedicle instru-mentation and posterolateral fusion in the symptomatic degen-erative lumbar spine. Eur Spine J 4:231–237, 1995

30. Schofferman J, Slosar P, Reynolds J, et al: A prospective ran-domized comparison of 270 degrees fusions to 360 degrees fu-sions (circumferential fusions). Spine 26:E207–E212, 2001

31. Simmons JW: Posterior lumbar interbody fusion with posteriorelements as chip grafts. Clin Orthop Relat Res 193:85–89,1985

32. Slosar PJ, Reynolds JB, Schofferman J, et al: Patient satisfac-tion after circumferential lumbar fusion. Spine 25:722–726,2000

33. Vamvanij V, Fredrickson BE, Thorpe JM, et al: Surgical treat-ment of internal disc disruption: an outcome study of four fu-sion techniques. J Spinal Disord 11:375–382, 1998

34. Yashiro K, Homma T, Hokari Y, et al: The Steffee variable screwplacement system using different methods of bone grafting.Spine 16:1329–1334, 1991


36. Zdeblick TA, David SM: A prospective comparison of surgicalapproach for anterior L4-l5 fusion: laparoscopic versus minianterior lumbar interbody fusion. Spine 25:2682–2687, 2000

37. Zhao J, Wang X, Hou T, et al: One versus two BAK fusioncages in posterior lumbar interbody fusion to L4-l5 degenera-tive spondylolisthesis: a randomized, controlled prospectivestudy in 25 patients with minimum two-year follow-up. Spine27:2753–2757, 2002

Manuscript received December 7, 2004.Accepted in final form march 2, 2005.Address reprint requests to: Daniel K. Resnick, M.D., Depart-




699


Recommendations

Standard. There is insufficient evidence to recommenda treatment standard.


Options. 1) Pedicle screw fixation is recommended as atreatment option for patients with low-back pain treatedwith PLF who are at high risk for fusion failure becausethe use of pedicle screw fixation improves fusion successrates. 2) Pedicle screw fixation as a routine adjunct to PLFin the treatment of patients with chronic low-back paindue to DDD is not recommended because there is con-flicting evidence regarding a beneficial effect of pediclescrew fixation on functional outcome, and there is consis-tent evidence that the use of pedicle screw fixation is asso-ciated with higher costs and complications.

Rationale

The use of instrumentation as an adjunct to lumbar fusionprocedures has increased over the past two decades.5,12 Mul-tiple techniques have been described for the surgical treat-ment of patients with chronic low-back pain. Posterolateralfusion is one of the more widespread techniques and may beperformed with or without the use of pedicle screw fixationto provide internal fixation as a surgical adjunct to the fusionprocedure. The addition of instrumentation is associated withhigher costs and higher complication rates. The purpose ofthis review is to establish whether the medical evidence inthe scientific literature demonstrates a clinical benefit ofinternal pedicle screw fixation as an adjunct to PLF in thetreatment of patients with low-back pain due to degenera-tive lumbar disc disease or low-grade degenerative spondy-lolisthesis.

Search Criteria

A computerized search of the National Library of Medi-cine database of the literature published from 1966 to June2003 was performed. A search using the subject heading


700

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 12: pedicle screw fixation as an adjunct to posterolateralfusion for low-back pain



KEY WORDS • bone screw • pedicle screw fixation • fusion • bone graft •practice guidelines


Abbreviations used in this paper: ALIF = anterior lumbar inter-body fusion; DDD = degenerative disc disease; DPQ = Dallas PainQuestionnaire; ODI = Oswestry Disability Index; PLF = postero-lateral fusion; PLIF = posterior LIF; VAS = visual analog scale.


“spinal fusion, lumbar, treatment outcome, low-backpain” yielded 1030 citations. Clinical series reported inEnglish-language journals dealing with adult patients whohad fusion with instrumentation for degenerative lumbardisease were selected (333 references). Among the articlesreviewed, references were included that dealt with thecomparison of fusion techniques with or without in-strumentation. These references are summarized in Table1. All of these articles reported at least 1 year of clinicaland radiographic follow up.


Authors of several well-designed studies have evaluat-ed the effect of pedicle screw fixation on radiographic and functional outcomes in patients treated with PLF forsymptomatic lumbar spinal degenerative disease. BjarkeChristensen, et al.,3 performed a randomized study with a5-year follow up of 129 patients who underwent surgeryfor chronic low-back pain. Sixty-four patients underwentinstrumented PLF and 66 underwent noninstrumentedPLF. Five-year follow up was obtained in 93% of patients.The reoperation rate for the instrumented group was 25%compared with 14% for the noninstrumented group. Thisdifference in reoperation rate was the result of nine pa-tients having to undergo implant removal due to eitherback or leg pain. Two of these nine patients had malposi-tioned screws. Both groups exhibited significant improve-ments in all outcome measures. There were no differencesin work capacity, DPQ scores, or Low Back Pain RatingScale scores between the groups. A subgroup analysis re-vealed that patients with primary DDD (as opposed topatients with spondylolisthesis or patients undergoing reop-eration following a previous decompression) had signifi-cantly improved long-term outcomes when treated withpedicle screw instrumentation compared with those pa-tients treated with fusion without internal fixation (p ,0.02). Only 20 patients with primary degenerative diseaseof the lumbar spine were included in each treatment arm.There was no difference in radiographic fusion rates be-tween the groups; however, only static plain films wereused to determine successful fusion. This paper is thoughtto provide Class II medical evidence supporting the use ofpedicle screw fixation as a means to improve functionaloutcomes in patients with degenerative lumbar spinal dis-ease that requires fusion because of the improvementnoted in the subgroup analysis. The paper provides ClassIII medical evidence on the benefit (or lack thereof) ofpedicle screw fixation on fusion success rates.

Fritzell and colleagues8,9 performed a randomized mul-ticenter clinical trial in which 294 patients with low-backpain due to DDD at L4–5, L5–S1, or both levels were com-pared. Patients were divided into four treatment groups.Group 1 consisted of 73 patients treated with PLF withoutinternal fixation; Group 2 consisted of 74 patients treatedwith PLF with pedicle screw fixation; and Group 3 con-sisted of 75 patients who had interbody fixation and pos-terolateral onlay fusion with pedicle fixation (56 of thesepatients underwent ALIF and 19 underwent PLIF). Pa-tients in Group 4 were treated by nonsurgical means (72patients). Follow up was performed by an independent ob-server and included 219 of the 222 surgically treatedpatients and 70 of the 72 patients treated with nonsurgical

means (overall follow-up rate of 98%). All three surgicalgroups had superior clinical outcomes compared with thenonoperative group. Although the group of patients treat-ed with pedicle screw fixation had slightly greater im-provements on the ODI, the General Function Score, andthe Million VAS than the group treated without pediclescrew fixation, these differences were not statistically sig-nificant. Approximately 60% of patients in the PLF-alonegroup considered themselves to have achieved an excel-lent or good outcome compared with approximately 70%of patients in the PLF plus pedicle screw group. This dif-ference was not statistically significant. Nine patients whowere treated with pedicle screw fixation (with or withoutan interbody graft) developed a new radiculopathy follow-ing surgery. Three of these patients were found to havemalpositioned screws. No patient in the noninstrumentedfusion group had new radicular complaints. The overallcomplication rates were significantly higher among pa-tients in the instrumented groups compared with the nonin-strumented groups. Fusion success was evaluated by plainradiographs (without flexion–extension views) and wasfound to occur in 72% of patients in Group 1, 87% ofpatients in Group 2, and in 91% of patients in Group 3. Theauthors concluded that all surgical groups had similar clin-ical outcomes; however, there was an increase in both thefusion rate and the complication rate in both instrumentedgroups (p = 0.004). This paper provides Class III medicalevidence suggesting that functional outcomes are notinfluenced by the addition of pedicle screw fixation toPLF. It also provides Class III evidence suggesting that theaddition of pedicle screw fixation improves fusion successrates, albeit at the cost of a higher complication rate.

Fischgrund, et al.,6 performed a randomized trial in 76patients with degenerative spondylolisthesis. Thirty-fivepatients underwent PLF with pedicle screw fixation and33 patients underwent PLF without internal fixation. Two-year follow up was achieved in 67 (88%) of the 76 pa-tients. Clinical outcome (based on a patient satisfactionsurvey) was excellent or good in 78% of the patients inwhom instrumentation was placed and was excellent orgood in 85% among those patients treated without instru-mentation (p = 0.45). Fusion success was assessed withflexion–extension radiographs. Successful fusion occur-red in 83% of the instrumented cases compared with 45%of the noninstrumented cases (p = 0.0015). In this study,successful fusion did not statistically significantly influ-ence patient outcome (p = 0.435). The authors concludedthat the use of pedicle screw fixation may lead to a higherfusion success rate but that the use of pedicle screw fixa-tion did not improve clinical outcome. This paper pro-vides Class I medical evidence suggesting that the addi-tion of pedicle screw fixation improves the fusion successrate. It also provides Class III medical evidence (due to theuse of a nonvalidated outcome measure) suggesting thatthe addition of pedicle screw fixation to PLF does notinfluence functional outcome.

Thomsen and colleagues17 performed a prospective ran-domized trial of 130 patients with Grade I or II degenerativeor isthmic spondylolisthesis. Sixty-six patients underwentPLF without internal fixation and 64 patients underwentPLF with pedicle screws. Two-year follow up was availablefor 127 (98%) of the patients. Fusion success was evaluatedby plain radiographs (without flexion–extension views) and


Pedicle screws

701




TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

ped

icle

scr

ew f

ixat

ion

in p

atie

nts

wit

h de

gene

rati

ve l

umba

r di

seas

e*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Lor

enz,

et a

l., 1

991

II f

usio

nPr

ospe

ctiv

e ra

ndom

ized

stu

dy o

f 68

pat

ient

s w

/ $ 6

-mo

hist

ory

of d

isab

ling

back

pai

n. G

roup

I (

29)

PS in

stru

men

tatio

n im

prov

es f

usio

n ra

tes

& r

etur

n to

II o

utco

me

PLF

w/o

inst

rum

enta

tion.

Gro

up I

I (3

9) P

LF

1PS

. All

case

s w

ere

1 le

vel.

Mea

n FU

26

mos

.Fu-

w

ork

for

PLF.

(ret

urn

to w

ork)

sion

bas

ed o

n fl

exio

n–ex

tens

ion

x-ra

y w

as 5

9% G

roup

I &

100

% G

roup

II.

Ret

urn

to s

imila

r w

ork

was

31%

in G

roup

I &

72%

Gro

up I

I.B

ernh

ardt

, et a

l., 1

992

III

Ret

rosp

ectiv

e st

udy

of 4

7 pa

tient

s w

/ PL

F, 2

1 w

/ PSs

& 2

7 w

/o. 1

8 of

21

PS h

ad F

U. I

ndep

ende

nt

PLF

w/o

PS

is a

s ef

fect

ive

& s

afer

than

fus

ion

w/P

S.re

tros

pect

ive

revi

ew. A

ssig

nmen

t of

patie

nts

into

the

2 gr

oups

is p

oorl

y de

fine

d. P

seud

arth

rosi

s di

d no

t sig

nifi

cant

ly d

iffe

r bt

wn

the

2 gr

oups

(PS

22%

; non

scre

w g

roup

, 26%

. No

p va

lue

give

n).

12 (

67%

) of

18

PS &

19

(70%

) of

27

unin

stru

men

ted

PS p

atie

nts

had

good

or

exce

llent

res

ults

. 2PS

pat

ient

s ha

d po

stop

leg

dyse

sthe

sias

, 1 r

equi

ring

har

dwar

e re

mov

al (

5%).

No

unin

stru

men

ted

case

s ex

peri

ence

d dy

sest

hesi

as.

Bri

dwel

l, et

al.,

199

3II

I fu

sion

Pr

ospe

ctiv

e ra

ndom

ized

44

patie

nts

w/ D

S. P

atie

nts

w/ .

10˚

or 3

mm

mov

emen

t wer

e in

stru

men

ted.

PL

F 1

PS d

ecre

ases

cha

nces

of

slip

pro

gres

sion

, in-

III

outc

ome

Mea

n FU

3 y

rs &

2 m

os, m

in 2

4 m

os. G

roup

I n

o fu

sion

(9)

, Gro

up I

I PL

F no

PS

(10)

, Gro

up I

II

crea

ses

fusi

on r

ate,

& in

crea

sed

patie

nt s

atis

fact

ion.

PLF

1PS

(24

). F

usio

n ba

sed

on L

enke

–Bri

dwel

l sys

tem

(st

atic

pla

in x

-ray

). Fu

nctio

nal o

utco

me

not

valid

ated

, em

phas

ized

cla

udic

atio

n. R

esul

ts: s

pond

ylol

isth

esis

pro

gres

sion

wor

se in

Gro

up I

1II

th

an I

II, p

= 0

.001

. Fus

ion

rate

hig

her

in I

II th

an I

I, p

= 0

.002

. Z

debl

ick,

199

3I

fusi

onA

pros

pect

ive

rand

omiz

ed s

tudy

of

124

patie

nts

divi

ded

into

3 g

roup

s. G

roup

1 w

/ PL

F on

lay

fusi

on

Clin

ical

out

com

e di

rect

ly r

elat

ed to

fus

ion

rate

. In-

III

outc

ome

(51)

, Gro

up 2

w/ P

LF

w/ s

emir

igid

PS

(35)

, Gro

up 3

w/ P

LF

w/ r

igid

PS

syst

em (

37).

FU

was

9–2

8 cr

easi

ng s

tiffn

ess

of in

stru

men

tatio

n le

ads

to

mos

(m

ean

16 m

os).

Fus

ion

stat

us d

eter

min

ed b

y pl

ain

radi

ogra

phs:

fle

xion

–ext

ensi

on v

iew

s at

1 y

r. hi

gher

fus

ion

rate

s.Fu

sion

rat

es 6

5% G

roup

1, 7

7% G

roup

2, &

95%

Gro

up 3

. Dif

fere

nces

btw

n 1

& 3

(p

= 0

.002

) &

1

& 2

(p

= 0

.034

) w

ere

stat

istic

ally

sig

nifi

cant

. Clin

ical

out

com

es a

sses

sed

w/ a

non

stan

dard

mea

sure

gr

adin

g pa

tient

s as

exc

elle

nt, g

ood,

fai

r, or

poo

r. A

t the

late

st F

U, m

ean

16 m

os, G

roup

1 h

ad 4

9% e

x-ce

llent

, 22%

goo

d, &

29%

fai

r or

poo

r. G

roup

II

was

60,

29,

& 1

1%. G

roup

III

70,

24,

& 5

%.

Schw

ab, e

t al.,

199

5II

I fu

sion

Ret

rosp

ectiv

e co

hort

stu

dy w

/ FU

of

2 yr

s. F

usio

n as

sess

ed w

/ x-r

ay. G

roup

I 1

26 p

atie

nts

w/ P

LF

alon

e.

PS f

ixat

ion

impr

oves

fus

ion

rate

vs

unin

stru

men

ted

II

I ou

tcom

eG

roup

II

89 p

atie

nts

augm

ente

d w

/ PS.

Fus

ion

rate

ass

esse

d by

dyn

amic

x-r

ay: 6

5% &

91%

(p

,fu

sion

. No

incr

ease

in c

ompl

icat

ion

rate

: out

com

e 0.

001)

. Com

plic

atio

ns: p

seud

arth

rosi

s lo

wer

in th

e in

stru

men

tatio

n gr

oup.

Out

com

es a

sses

sed

w/

scal

e w

as n

onst

anda

rd b

ut in

stru

men

ted

fusi

ons

nons

tand

ardi

zed

mea

sure

w/ i

nstr

umen

ted

grou

p fa

ring

bet

ter.

did

bette

r.Fi

schg

rund

, et a

l., 1

997

I fu

sion

Pros

pect

ive

rand

omiz

ed 7

6 pa

tient

s sp

ondy

lolis

thes

is &

spi

nal s

teno

sis.

All

patie

nts

had

post

erol

atPL

F w

/ PS

is a

ssoc

iate

d w

/ hig

her

fusi

on r

ate,

but

no

II o

utco

me

inte

rtra

nsve

rse

proc

ess

arth

rode

sis.

Pat

ient

s w

ere

rand

omiz

ed to

a s

egm

enta

l tra

nspe

dicu

lar

inst

ru-

diff

eren

ce in

clin

ical

out

com

e.m

ente

d or

non

inst

rum

ente

d gr

oup.

Res

ults

: 67

patie

nts

w/ 2

-yr

FU. C

linic

al o

utco

me

was

exc

elle

nt

or g

ood

in 7

6% o

f pa

tient

s w

/ ins

trum

enta

tion

& in

85%

w/o

inst

rum

enta

tion

(p =

0.4

5). F

usio

n in

82

% o

f th

e in

stru

men

ted

vs 4

5% o

f no

nins

trum

ente

d ca

ses

(p =

0.0

015)

. Ove

rall,

suc

cess

ful f

usio

n di

d no

t inf

luen

ce o

utco

me

(p =

0.4

35).

K

atz,

et a

l., 1

997

III

Pros

pect

ive

nonr

ando

miz

ed s

tudy

of

272

patie

nts

w/ d

ecom

pres

sion

for

lum

bar

sten

osis

. Lam

inec

tom

y N

onin

stru

men

ted

arth

rode

sis

resu

lted

in s

uper

ior

vs

lam

inec

tom

y w

/ PL

F (3

7) v

s la

min

ecto

my

1PS

PL

F (4

1). N

onin

stru

men

ted

arth

rode

sis

asso

ci-

relie

f of

bac

k pa

in a

fter

6 &

24

mos

.at

ed w

/ sup

erio

r re

lief

of L

BP

at 6

(p

= 0.

004)

and

24

mos

(p

= 0.

01).

Fin

ding

s lim

ited

by

smal

l no.

of

part

icip

atin

g su

rgeo

ns; m

odes

t sam

ple

size

pro

duce

d p

valu

es o

f bo

rder

line

sign

ific

ance

& th

e po

tent

ial c

onfo

undi

ng b

ias

of c

hoic

e of

trea

tmen

t w/ f

usio

n be

ing

surg

eon

depe

nden

t.T

hom

sen,

et a

l., 1

997

II f

usio

nPr

ospe

ctiv

e ra

ndom

ized

tria

l of

130

patie

nts

w/ G

rade

1 o

r 2

dege

nera

tive

or is

thm

ic s

pond

ylol

isth

esis

:In

stru

men

tatio

n do

es n

ot im

prov

e re

sults

of

PLF.

I

outc

ome

66 p

atie

nts

w/ P

LF

& 6

4 pa

tient

s w

/ PL

F 1

PS. 9

8% o

f pa

tient

s ha

d 2-

yr F

U. F

usio

n ra

tes

by d

y-na

mic

rad

iogr

aphs

wer

e no

t sig

nifi

cant

ly d

iffe

rent

btw

n gr

oups

. DPQ

impr

oved

sig

nifi

cant

ly in

bot

h gr

oups

but

no

sign

ific

ant d

iffe

renc

e bt

wn

grou

psex

cept

for

the

daily

act

iviti

es s

ubse

ctio

n of

the

DPQ

whi

ch w

as im

prov

ed in

the

inst

rum

ente

d gr

oup

(p,

0.05

). I

nstr

umen

tatio

n in

crea

sed

op ti

me,

blo

od

loss

, & e

arly

reo

p ra

te s

igni

fica

ntly

. Fun

ctio

nal o

utco

me

was

not

sig

nifi

cant

ly im

prov

ed w

/ use

of

PSex

cept

whe

n pa

tient

s w

/ PS

had

conc

omita

nt d

ecom

pres

sion

s (p

,0.

05).

Fr

ance

, et a

l., 1

999

III

fusi

on &

Pr

ospe

ctiv

e ra

ndom

ized

71

patie

nts

PLF

6PS

. No

stat

istic

al d

iffe

renc

e in

pat

ient

-rep

orte

d ou

tcom

e N

o di

ffer

ence

in o

utco

me

or a

rthr

odes

is r

ate

of P

LF

outc

ome

btw

n th

e 2

grou

ps. S

light

non

sign

ific

ant t

rend

tow

ard

incr

ease

dra

diog

raph

ic f

usio

n ra

te in

the

grou

p vs

PL

F 1

PS.

w/ i

nstr

umen

tatio

n th

at d

id n

ot c

orre

late

w/ a

n in

crea

sed

patie

nt-r

epor

ted

impr

ovem

ent r

ate.

Cla

ss

III

due

to n

onst

anda

rd o

utco

me

mea

sure

s, la

rge

drop

out r

ate,

& p

oor

x-ra

y cr

iteri

a fo

r fu

sion

.Fr

itzel

l, et

al.,

200

1I

func

tiona

lPr

ospe

ctiv

e ra

ndom

ized

mul

ticen

ter

stud

y of

294

pat

ient

s. G

roup

1 (

PLF,

73)

, Gro

up 2

(PL

F 1

PS, 7

4), &

C

ompa

riso

n of

non

inst

rum

ente

d fu

sion

vs

PLF

vs

III

radi

ogra

phic

Gro

up 3

(PL

F 1

PS

1in

terb

ody

fusi

on, 7

5:A

LIF

[56

] or

PL

IF [

19])

. Gro

up 4

non

op (

72).

Res

ults

: PL

F 1

PS v

s in

terb

ody

fusi

on. N

o di

ffer

ence

in

FU 2

01 (

91%

) of

222

pat

ient

s af

ter

2 yr

s. A

ll su

rgic

al te

chni

ques

red

uced

pai

n &

dec

reas

ed d

isab

ility

outc

ome.

Int

erbo

dy f

usio

n ha

d hi

ghes

t fus

ion

rate

s su

bsta

ntia

lly, b

ut n

o si

gnif

ican

t dif

fere

nces

fou

nd a

mon

g gr

oups

. Gro

ups

2 &

3 c

onsu

med

sig

nifi

-&

com

plic

atio

n ra

tes

follo

wed

by

PLF

1PS

.ca

ntly

mor

e re

sour

ces:

op ti

me,

blo

od tr

ansf

usio

ns, &

sta

y. E

arly

com

plic

atio

n ra

te 6

% in

Gro

up 1

, 16

% in

Gro

up 2

, & 3

1% in

Gro

up 3

. Fus

ion

rate

by

plai

n ra

diog

raph

s: 7

2% in

Gro

up 1

, 87%

in

Gro

up 2

, & 9

1% in

Gro

up 3

. Hig

h co

mpl

icat

ion

rate

of

inst

rum

enta

tion

rela

ted

to 9

pat

ient

s w

/ PS

rem

oval

for

new

ons

et r

adic

ulop

athy

. At t

he ti

me

of r

eop

only

3 a

ctua

lly h

ad c

ortic

al w

all b

reac

h.

Con

tinu

ed


was not significantly different between the groups. Therewere no reported complications associated with the place-ment of pedicle screws in this study. Outcome wasassessed with the DPQ. The DPQ improved significantlyfor patients in both treatment groups. There was no statis-tically significant difference between the groups except forthe daily activities subsection of the DPQ, which was sta-tistically significantly better among patients in the instru-mented group (p , 0.05). The use of instrumentationincreased operative time, blood loss, and the early reoper-ation rate compared with patients treated with fusion with-out instrumentation. As cited in previous studies, theincreased reoperation rate was due to procedures to removeinstrumentation following successful fusion. These authorsconcluded that functional outcome was not significantlyimproved with the use of pedicle screw fixation with theexception of an improved score in activities of daily livingas measured by the DPQ (p , 0.05). Improvements,although not statistically significant, in all functional out-come measures were noted in the patients with primary de-generative disease treated with pedicle screw fixation com-pared with similar patients treated without instrumentation.Because the study only included 20 patients with primarydegenerative disease of the lumbar spine in each treatmentgroup, these results never reached significance. This studytherefore provides Class III medical evidence suggestingthat the use of pedicle screw fixation is not associated withincreased fusion rates. The study also provides Class IIImedical evidence indicating that pedicle screw fixationmay improve functional outcome in patients with DDD ofthe lumbar spine.

Zdeblick18 performed a prospective study of 124 patientswith degenerative spondylolisthesis and DDD who wererandomized into three groups. Patients in Group 1 weretreated with PLF without internal fixation (51 patients),those in Group 2 were treated with PLF with semirigidpedicle screw fixation (35 patients), and patients in Group3 were treated with PLF with rigid pedicle screw fixation(37 patients). Nine patients initially assigned to Groups 2and 3 were transferred to Group 1 because of osteoporosis,which prevented adequate pedicle screw fixation. Follow-up periods ranged from 9 to 28 months (mean 16 months).Fusion success was determined using plain radiographsincluding flexion–extension views. The fusion successrates were 65% in Group 1, 77% in Group 2, and 95% inGroup 3. The differences in fusion success rates betweenGroups 1 and 3 (p = 0.002) and between Groups 1 and 2 (p = 0.034) were both statistically significant. Clinical out-comes were assessed with nonvalidated outcome mea-sures, grading patients as excellent, good, fair, or poor. Atlast reported follow up (mean 16 months follow up), Group1 had 49% excellent, 22% good, and 29% fair or poor out-comes. Group 2 had 60% excellent, 29% good, and 11%fair or poor outcomes. Group 3 had 70% excellent, 24%good, and 5% fair or poor outcomes (no tests of signifi-cance were done). This study provides Class I medical evi-dence that the addition of pedicle screw instrumentationimproves fusion rates. Because of the nonvalidated func-tional outcome measure used, it provides Class III medicalevidence supporting a beneficial effect of pedicle screwfixation on functional outcome.

France and colleagues7 performed a randomized pros-pective study of 83 patients treated with lumbar fusion for


Pedicle screws

703

TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Bja

rke

Chr

iste

nsen

, I

func

tiona

lPr

ospe

ctiv

e ra

ndom

ized

w/ 5

-yr

FU. 1

29 p

atie

nts

w/ L

BP

had

PLF

w/ o

r w

/o P

Ss. T

he D

PQ, L

BPR

S,

In p

atie

nts

w/ D

DD

, the

PF

w/ P

S gr

oup

had

bette

r et

al.,

200

2II

I ra

diog

raph

ic&

que

stio

nnai

re c

once

rnin

g w

ork

stat

us a

sses

sed

the

outc

ome.

Res

ults

:5-y

r FU

of

93%

: reo

pra

te in

ou

tcom

e (p

,0.

02).

Fus

ion

rate

s no

t sig

nifi

cant

ly

inst

rum

ente

d is

25%

vs

14%

in n

onin

stru

men

ted

grou

p (p

,0.

03).

No

diff

eren

ce in

wor

k ca

paci

ty

diff

eren

t btw

n in

stru

men

ted

& n

onin

stru

men

ted

or f

unct

iona

l out

com

e as

mea

sure

d by

bot

h th

e D

PQ &

LB

PRS.

Sub

grou

p of

pri

mar

y de

gene

rativ

e gr

oups

.in

stab

ility

bet

ter

outc

ome

w/ i

nstr

umen

ted

grou

p (p

,0.

02).

Fus

ion

rate

s no

t sig

nifi

cant

ly d

iffe

rent

.Fr

itzel

l, et

al.,

200

2I

outc

ome

Mul

ticen

ter

pros

pect

ive

rand

omiz

ed s

tudy

of

211

patie

nts.

PL

F (7

1), P

LF

1PS

(68

), s

truc

tura

l aut

o-C

ompl

icat

ions

hig

her

w/ u

se o

f in

stru

men

tatio

n.gr

aft i

nter

body

1 P

LF

1PS

(72

). N

o si

gnif

ican

t dif

fere

nce

in O

DI,

VA

S, G

ener

al F

unct

ion

Scor

e,or

Zun

g D

epre

ssio

n Sc

ale

at 2

yrs

(p

.0.

05).

Pow

er to

det

ect s

uch

a di

ffer

ence

was

low

. Ear

ly c

om-

plic

atio

n ra

te 6

, 18,

& 3

1% (

p =

0.0

01).

Tot

al c

ompl

icat

ion

rate

aft

er 2

yrs

in P

LF

grou

p 12

%,

com

pare

d w

/ 22%

in P

S gr

oup,

& 4

0% in

360

˚ gr

oup

(p =

0.0

003)

. Rei

nter

vent

ion

rate

was

6,

22, &

17%

(p

= 0

.02)

No

fusi

on te

chni

que

prod

uced

sup

erio

r cl

inic

al o

utco

me.

The

fus

ion

asse

ssm

ent m

etho

dolo

gy w

as n

ot p

rese

nted

.A

nder

sen,

et a

l., 2

003

III

5-yr

FU

by

ques

tionn

aire

of

pros

pect

ive

rand

omiz

ed tr

ial o

f PL

F w

/ or

w/o

PS

in d

egen

erat

ive

or is

th-

No

sign

ific

ant d

iffe

renc

e in

pai

n bt

wn

PLF

& P

LF

1m

ic s

pond

ylol

isth

esis

. 53

patie

nts

wer

e tr

eate

d w

/ PS,

56

w/o

PS.

No

diff

eren

ce in

pai

n bt

wn

grou

psPS

gro

ups.

(p =

0.0

07).

No

fusi

on r

ates

pre

sent

ed.

Jage

r, et

al.,

200

3II

IN

onra

ndom

ized

, ret

rosp

ectiv

e st

udy:

17

patie

nts

unde

rwen

t PL

F +

PS &

wer

e co

mpa

red

w/ 1

6 pa

tient

s N

o ad

vant

age

to P

S fo

r lu

mba

r fu

sion

.w

ho u

nder

wen

t PL

F. T

here

wer

e no

sta

tistic

al d

iffe

renc

es in

pat

ient

-rep

orte

d, c

linic

al, &

radi

ogra

phic

outc

ome.

Ins

trum

ente

d gr

oup

had

mor

e pa

in p

osto

p.

*FU

= f

ollo

w u

p; L

BP

= lo

w-b

ack

pain

; LB

PRS

= L

ow B

ack

Pain

Rat

ing

Scal

e; P

S =

ped

icle

scr

ew.


low-back pain. The treatment groups included patients withisthmic spondylolisthesis, failed–back surgery syndrome,herniated nucleus pulposus, and degenerative spondy-lolisthesis. The mean follow-up interval was 29 months.Forty-one of the patients underwent PLF with pediclescrews, and the remaining 42 patients underwent PLF with-out internal fixation. Final clinical follow up (VAS andsome elements of the DPQ) was available in 71 of the 83patients. Only 57 of the 83 patients were available for radi-ographic follow up. There was an increase in the fusionsuccess rate (based on static radiographs alone) in the in-strumented group (76% instrumented group compared with64% noninstrumented group). There were no statisticallysignificant differences in functional outcome between thegroups. The small sample size, high dropout rate, inclusionof patients with multiple diagnoses, and reliance on staticplain films to assess fusion success makes it impossible todraw meaningful conclusions from this study regarding theutility of pedicle screw fixation as an adjunct to fusion in the management of patients with lumbar degenerativedisease.

Lorenz and colleagues14 prospectively studied a groupof 68 patients they treated with PLF. Twenty-nine patientswere treated with pedicle screw fixation and 39 patientswere treated without pedicle screw fixation. The patientstreated with pedicle screw fixation had improved fusionsuccess rates (100%) compared with patients treated with-out instrumentation (58%); fusion success was based onflexion–extension radiographs. Patients treated with in-strumentation had improved relief of back pain (nonvali-dated outcome measure) and higher return-to-work rates(72% compared with 31%) than patients treated withoutinstrumentation. This report is considered to provide ClassII medical evidence in favor of pedicle screw fixation as ameans to improve fusion success rates and Class II med-ical evidence in favor of pedicle screw fixation as a meansto improve return-to-work rates.

In addition to these prospective studies, retrospectivecohort studies have also been performed. Some of thesestudies provide relevant Class III medical evidence.Schwab, et al.,16 performed a retrospective review of 215patients treated with lumbar fusion for low-back pain. Onehundred twenty-six patients were treated with PLF and 89patients were treated with PLF with pedicle screw fixa-tion. Fusion success rates were assessed with dynamic ra-diographs. The pedicle screw fixation group had a fusionsuccess rate of 91% compared with 65% in the noninstru-mented group (p , 0.001). Outcomes were assessed usinga nonstandardized scale. The outcome in the instrumentedgroup was superior to that in the noninstrumented group(no tests of significance were reported). This study pro-vides Class III medical evidence suggesting that pediclescrew fixation results in higher fusion rates and betterfunctional outcomes.

A number of retrospective cohort studies have beenreported that provide less useful data. Several of thesestudies are widely cited and warrant discussion. Katz, etal.,13 performed a cohort study of 272 patients with lumbarstenosis. Of this group, 37 were treated with PLF withoutinternal fixation and 41 were treated with PLF with pedi-cle screw fixation. The remaining patients were treatedwith decompression without fusion. Overall, the nonin-strumented PLF group experienced the greatest relief of

low-back pain at 6 (p = 0.004) and 24 months (p = 0.01)following surgery. A subgroup analysis of patients withlumbar stenosis and spondylolisthesis treated with PLFrevealed no significant difference in outcomes betweenpatients treated with or without pedicle screw instrumen-tation. Fusion success rates were not reported. Because theselection criteria used to determine whether a patientreceived a decompression, a decompression and fusion, ora decompression and fusion with pedicle screw fixationwas entirely surgeon dependent, no meaningful conclu-sions can be drawn from this study.

Bernhardt and colleagues2 retrospectively reviewed aseries of 47 patients of whom 27 were treated with nonin-strumented PLF and 21 were treated with PLF with pedi-cle screw fixation (three of these 21 were lost to followup). Pseudarthrosis rates did not significantly differ be-tween the two groups (pedicle screw group 22%; onlayPLF group 26%). Twelve (67%) of the 18 patients treatedwith pedicle screw fixation and 19 (70%) of the 27patients treated without fixation had good or excellentresults. Two patients treated with pedicle screw fixationexperienced postoperative leg dysesthesias, whereas noneof the noninstrumented PLF patients experienced dyses-thesias. The authors concluded that lumbar PLF withoutpedicle screw fixation is as effective as and safer thanfusion with pedicle screw fixation. Jager, et al.,11 performeda nonrandomized cohort trial of 16 patients treated withPLF and 17 patients treated with PLF with pedicle screwfixation. Outcomes were measured using the ODI. Fusionsuccess was assessed based on plain radiographs withoutdynamic views. There was no statistically significant dif-ference between the groups with respect to outcome orfusion success rate. Andersen, et al.,1 performed a retro-spective analysis by using a questionnaire mailed topatients 5 years following lumbar fusion: 53 with pediclescrew instrumentation and 56 without. There was no dif-ference in pain reporting between the groups. Fusion rateswere not assessed. All of these studies suffer from severemethodological shortcomings that prevent the developmentof meaningful conclusions. The most common shortcom-ings in these reports relate to small sample size, retrospec-tive design, and inadequate outcome measures.

Mardjetko, et al.,15 performed a metaanalysis of the med-ical literature on the topic of pedicle screw fixation andconcluded that pedicle screw fixation increased fusionsuccess rates. They summarized 25 articles on the treat-ment of degenerative lumbar spondylolisthesis and foundthat the addition of pedicle screw instrumentation to a PLFenhances fusion rates (93% compared with 86%) and alsoimproves functional outcomes (90% compared with86%). Further supporting evidence for the role of pediclescrew instrumentation resulting in increased fusion rates isderived from studies in which PLF with or without instru-mentation is analyzed using different techniques or differ-ent patient selection criteria. For example, Grubb andLipscomb10 performed a retrospective cohort study of pa-tients treated with either noninstrumented PLF (49 pa-tients) or PLF with a compression U-rod (52 patients),with a mean follow up of 30 months. Fusion was assessedbased on plain radiographs including dynamic films. Thepseudarthrosis rate in the noninstrumented group was35% compared with 6% in the instrumented group. Inboth groups of patients with chronic low-back pain sec-




ondary to DDD, solid lumbosacral fusion was associatedwith decreased pain and higher return-to-work rates (noprobability values reported).

Summary

This review focused on an examination of the literatureon the surgical treatment of low-back pain in patients withDDD or low-grade degenerative spondylolisthesis treatedwith PLF, with or without the use of pedicle screw fixa-tion. All Class I and the majority of Class II and Class IIImedical evidence on this topic indicates that the additionof pedicle screw fixation to PLF increases fusion successrates when assessed based on plain x-ray films with dy-namic imaging. Although there does appear to be a posi-tive relationship between radiographic fusion and clinicaloutcome, no convincing correlation has been demonstrat-ed. Although several reports suggest that clinical out-comes are improved with the addition of pedicle screwfixation, there are conflicting findings from similarly clas-sified evidence sources (primarily Class II and III). Fur-thermore, the largest contemporary randomized controlledstudy on this topic failed to demonstrate a significant ben-eficial effect for the use of pedicle screw fixation in pa-tients treated with PLF for chronic low-back pain.8

This absence of proof should not, however, be interpret-ed as a proof of absence. For example, in this same study,8patient satisfaction scores improved from approximately60% to approximately 70% with the addition of pediclescrew fixation. This difference in outcome may be clinical-ly relevant. Similarly, the improvement in ODI scores was40% greater in the group of patients treated with pediclescrew fixation compared with those treated with PLF alone.If an analysis to determine the sample size necessary toensure a power of 0.8 (or an 80% chance of detecting a sig-nificant effect) in a study in which the good outcome rate is60% in the control group and 70% in the treatment group isperformed, approximately 355 patients would be needed ineach treatment group (http://department.obg.cuhk.edu.hk).Alternatively, if a similar analysis is performed using thedifferential scores obtained in the ODI measurements re-ported in the paper by Fritzell, et al.,9 approximately 225patients would be needed per treatment group (http://calcu-lators.stat.ucla.edu/powercalc). Although Fritzell, et al., didnot detect a significant benefit associated with the use ofpedicle screw fixation as an adjunct to PLF, their samplesize severely limited the power of their study to detect sucha benefit. All studies reviewed suffer from similar lack ofpower. Therefore, no definitive statement regarding the ef-ficacy of pedicle screw fixation as a means to improvefunctional outcomes in patients undergoing PLF for chron-ic low-back pain can be made. There appears to be consis-tent evidence suggesting that pedicle screw fixation in-creases the costs and complication rate of PLF. It isrecommended, therefore, that the use of pedicle screw fix-ation as a supplement to PLF be reserved for those patientsin whom there is an increased risk of nonunion when treat-ed with PLF. High-risk patients include, but are not limitedto patients who smoke, who are undergoing revision sur-gery, or who suffer systemic diseases known to be associ-ated with poor bone healing.


The most important issue confronting the surgeon whendeciding whether to use a particular surgical adjunct is theexistence of any evidence that the inclusion of that adjunctimproves functional outcome. Whereas it is clear that theaddition of pedicle screw fixation improves radiographi-cally demonstrated fusion rates, the evidence supportingits role in improving patient outcome is less clear. Most ofthe medical evidence encountered in this review is full ofpotential sources of error, including inadequate samplesize, bias in accruing and assigning patients to treatments,and lack of statements about reliability and validity of out-come measures. This led to the downgrading of medicalevidence in this review, particularly for the use of vague,nonvalidated functional outcome measures. Future inves-tigators are encouraged to use any or several of the manyvalidated measures now available for the investigation offunctional gains following lumbar fusion. In addition, ifthe question of the usefulness of pedicle screw fixation asan adjunct to PLF is ever to be resolved, it will be in thecontext of randomized clinical trials that are appropriatelydesigned and implemented.

References

1. Andersen T, Christensen FB, Hansen ES, et al: Pain 5 yearsafter instrumented and non-instrumented posterolateral lumbarspinal fusion. Eur Spine J 12:393–399, 2003

2. Bernhardt M, Swartz DE, Clothiaux PL, et al: Posterolaterallumbar and lumbosacral fusion with and without pedicle screwinternal fixation. Clin Orthop 284:109–115, 1992

3. Bjarke Christensen F, Stender Hansen E, Laursen M, et al: Long-term functional outcome of pedicle screw instrumentation as asupport for posterolateral spinal fusion: randomized clinicalstudy with a 5-year follow-up. Spine 27:1269–1277, 2002

4. Bridwell KH, Sedgewick TA, O’Brien MF, et al: The role offusion and instrumentation in the treatment of degenerativespondylolisthesis with spinal stenosis. J Spinal Disord 6:461–472, 1993

5. Deyo R, Cherkin D, Conrad D, et al: Cost, controversy, crisis:low back pain and the health of the public. Ann Rev PubHealth 12:141–156, 1991

6. Fischgrund JS, Mackay M, Herkowitz HN, et al: 1997 VolvoAward winner in clinical studies. Degenerative lumbar spondy-lolisthesis with spinal stenosis: a prospective, randomized studycomparing decompressive laminectomy and arthrodesis withand without spinal instrumentation. Spine 22:2807–2812, 1997

7. France JC, Yaszemski MJ, Lauerman WC, et al: A randomizedprospective study of posterolateral lumbar fusion. Outcomeswith and without pedicle screw instrumentation. Spine 24:553–560, 1999

8. Fritzell P, Hagg O, Wessberg P, et al: 2001 Volvo AwardWinner in Clinical Studies: Lumbar fusion versus nonsurgicaltreatment for chronic low back pain: a multicenter randomizedcontrolled trial from the Swedish Lumbar Spine Study Group.Spine 26:2521–2534, 2001

9. Fritzell P, Hagg O, Wessberg P, et al: Chronic low back painand fusion: a comparison of three surgical techniques: a pros-pective multicenter randomized study from the Swedish Lum-bar Spine Study Group. Spine 27:1131–1141, 2002

10. Grubb SA, Lipscomb HJ: Results of lumbosacral fusion for de-generative disc disease with and without instrumentation. Two-to five-year follow-up. Spine 17:349–355, 1992

11. Jager M, Seller K, Raab P, et al: Clinical outcome in monoseg-mental fusion of degenerative lumbar instabilities: instrument-


Pedicle screws

705


ed versus non-instrumented. Med Sci Monit 9:CR324–CR327,2003

12. Katz JN: Lumbar spinal fusion: surgical rates, costs, and com-plications. Spine 20 (24 Suppl):S78–S83, 1995

13. Katz JN, Lipson SJ, Lew RA, et al: Lumbar laminectomy aloneor with instrumented or noninstrumented arthrodesis in degen-erative lumbar spinal stenosis. Patient selection, costs, and sur-gical outcomes. Spine 22:1123–1131, 1997

14. Lorenz M, Zindrick MR, Schwaegler P, et al: A comparison of single level fusions with and without hardware. Spine 16:S455–S458, 1991

15. Mardjetko SM, Connolly PJ, Shott S: Degenerative lumbarspondylolisthesis. A meta-analysis of the literature. Spine 19(20 Suppl):S2256–S2265, 1994

16. Schwab FJ, Nazarian DG, Mahmud F, et al: Effects of spinalinstrumentation on fusion of the lumbosacral spine. Spine 20:2023–2028, 1995

17. Thomsen K, Christensen FB, Eiskjaer SP, et al: 1997 VolvoAward winner in clinical studies. The effect of pedicle screwinstrumentation on functional outcome and fusion rates in pos-terolateral lumbar spinal fusion: a prospective, randomizedclinical study. Spine 22:2813–2822, 1997







Therapeutic Recommendations

Standards. Facet injections are not recommended as long-term treatment for chronic low-back pain.


Options. The use of lumbar epidural injections or TPIsis not recommended as a treatment option for long-termrelief of chronic low-back pain. The use of lumbar epidu-ral injections, facet injections, or TPIs is recommended asa treatment option to provide temporary, symptomatic re-lief in selected patients with chronic low-back pain.

Diagnostic Recommendations

Standards. There is insufficient evidence to recommenda diagnostic standard.

Guidelines. There is insufficient evidence to recommenda diagnostic guideline.

Options. The use of lumbar facet injections is recom-mended as a diagnostic tool for predicting the response to

lumbar facet RF ablation. The use of lumbar facet injec-tions is not recommended as a diagnostic tool to predictthe response to lumbar fusion surgery.

Rationale

Several injection techniques, using an anesthetic agent,typically in combination with another agent such as asteroid, are used in the treatment of patients with chroniclow-back pain due to degenerative disease of the lumbarspine. An analysis of the literature regarding these treat-ments was performed in an attempt to answer three ques-tions. 1) Are lumbar ESIs effective for improving the out-comes of patients with chronic low-back pain resultingfrom degenerative disease of the lumbar spine? 2) Arelumbar facet injections effective for improving the out-comes of patients with chronic low-back pain resultingfrom degenerative disease of the lumbar spine? 3) Arelumbar TPIs effective for improving the outcomes ofpatients with chronic low-back pain resulting from degen-erative disease of the lumbar spine?

Search Criteria

A computerized search of the database of the NationalLibrary of Medicine was conducted for the period of 1996to July 2003 using the terms “epidural steroid injections orblocks,” “caudal injections or blocks,” “selective nerve root



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 13: injection therapies, low-back pain, and lumbar fusion



KEY WORDS • spinal injection • epidural injection • treatment outcome •practice guidelines

707

Abbreviations used in this paper: ESI = epidural steroid injec-tion; MPQ = McGill Pain Questionnaire; ODI = Oswestry Disabi-lity Index; RCT = randomized controlled trial; RF = radiofrequen-cy; ROM = range of motion; TP = trigger point; TPI = TP injection;VAS = visual analog scale.


injections or blocks,” “transforaminal injections or blocks,”“facet injections or blocks,” “median nerve blocks,” and“trigger point injections.” The search was limited to publi-cations in the English language. Review papers were also asource of references and any new titles identified wereadded to the search results yielding a total of 1486 papers.Papers not concerned with lumbar ESIs, lumbar facet injec-tions, or lumbar TPIs were discarded. Papers restricted tothe evaluation of the role of injections for the managementof radiculopathy or acute back pain were also rejected.Fifty papers were identified as providing relevant medicalevidence concerning the use of lumbar ESIs, 48 papers forlumbar facet injections, and 17 papers for lumbar TPIs. Allpapers providing Class II or better evidence and selectedpapers providing Class III evidence regarding the use ofinjections in the treatment of chronic low-back pain are re-corded by injection type in Tables 1 to 3. Significant sup-portive data from Class III studies and review articles areprovided by the references listed in the bibliography.


The Use of Lumbar ESIs (Epidural Injections, CaudalInjections, Transforaminal Injections) in the Treatment of Chronic Low-Back Pain due to Degenerative Disease of the Lumbar Spine

Epidural injections have been used for the treatment ofspinal disorders for more than 50 years.8,35,55,69 One en-counters several difficulties when interpreting the litera-ture regarding the use of these techniques for the treatmentof chronic (here defined as . 3 months) low-back pain.Recent studies have indicated that the application of ESIswithout radiographic confirmation is associated with aneedle malposition rate as high as 40%.27,32,51,72 Becauseneedle malposition may be responsible for treatment fail-ure and because the needle malposition rate in studies per-formed without confirmatory fluoroscopic imaging arenot known, the results of such studies must be interpretedwith caution.20,27,31,51,69,71 In most reports published before2000 fluoroscopic imaging and contrast administration toconfirm needle placement were not consistently used.

Another difficulty encountered in the literature is relat-ed to mixed patient populations. Many studies mixed pa-tients with primary radicular complaints with patients whohad primary low-back pain complaints.4,26,28,33,35,46,54,55,58,59,

68,71 There are several randomized studies that report ben-eficial effects of ESI on radicular complaints. 21,54,69 Theuse of ESIs in the treatment of acute low-back pain (ratherthan radiculopathy) has not been subjected to the samescientific scrutiny. The results of several of these studiesdo not support ESIs for the treatment of acute low-backpain alone.13,33,35,39,58,68 The ability to discern the effect ofESI on acute low-back pain is confounded by the inclu-sion of patients with radiculopathy. It is generally not pos-sible to separate out the treatment effect of ESI on the twoconditions because of the manner in which the data arepresented.

Nelemans and colleagues47,48 published a Cochrane Re-view regarding the use of injection therapy for subacuteand chronic benign low-back pain. In their review theauthors cite nine randomized clinical trials in which ESIsin the treatment of chronic low-back pain were investigat-

ed.3,6,7,10,12,13,24,44,57 The authors of this review state that nosignificant evidence could be derived from these trials tosupport the use of ESIs as an effective treatment forchronic low-back pain. A close evaluation of the trialscited reveals that six of the nine trials evaluated the effec-tiveness of epidural injections for the treatment of suba-cute and chronic sciatica, not back pain.3,7,10,12,44,57 A total offour randomized clinical trials in which epidural injectionswere performed in the treatment of chronic low-back painwithout significant sciatica have been identified.

Brievik, et al.,6 reported on 35 patients with incapacitat-ing chronic low-back pain treated with epidural injections.Between one and three injections were given at weeklyintervals. One treatment group of 16 patients receivedinjections containing 20 ml of bupivacaine 0.25% and 80mg of depomethylprednisolone. The other treatment group(19 patients) received injections containing 20 ml bupiva-caine 0.25% followed by 100 ml saline. They reported nodifference in the degree of pain relief or changes in physi-cal examination between the two groups. Unfortunately,saline in this volume, especially if forcefully injected, hasbeen suggested as therapeutic in and of itself.45,53 This studyprovides Class III medical evidence suggesting that ESIsmay be an effective treatment for chronic low-back painbecause there was no true control group.

Dallas and colleagues13 also compared two ESI treat-ments in a small group of patients with chronic postopera-tive low-back pain. They used a randomized, double-blind-ed, cross-over design. In their study, 10 patients received anepidural injection of 8 mg morphine and 10 patientsreceived an epidural injection of 10 ml saline. All patientsthen received an injection of 80 mg of methylprednisolone1 hour later. After 8 weeks, the groups crossed over and the treatment was repeated. Outcomes (a VAS and changesin physical examination) were gathered at 30 minutes, 2hours, 16 hours, weekly for 1 month, biweekly for a 2ndmonth, and again at 6 months. The only significant differ-ence identified between groups occurred at 6 weeks. At thistime point, the morphine plus steroid group had superiorpain relief compared with the saline plus steroid group.Rocco, et al.,56 published a randomized, double-blind studyin which 22 patients with chronic low-back pain resultingfrom postlaminectomy syndrome were examined. In theirstudy, all patients received 50 mg lidocaine epidurally andwere randomly assigned to receive 75 mg triamcinolonediacetate, 8 mg morphine, or both, in addition to the lido-caine. Pain VAS as well as ODI and Waddell disabilityscores were used in follow-up study. The study was termi-nated due to marked respiratory depression in the groupsthat received morphine. Only one patient in the entire studyachieved lasting pain relief. This paper provides Class IIImedical evidence suggesting that ESIs do not provide long-term relief of pain in patients with chronic low-back paindue to postlaminectomy syndrome. This study also suggeststhat respiratory depression is a common and significant sideeffect of morphine injection into the epidural space.

Epidural steroid injections were compared with intra-thecal injections of midazolam in a study by Serrao, et al.61

Theirs was a prospective, double-blind trial involving 28patients. All patients had chronic low-back pain, and allother attempts at conservative care had failed. Outcomeswere assesed by the MPQ, VAS score, and a record of painmedication requirements prior to and at 2 weeks and 2





Injection therapies

709

TAB

LE

1E

pidu

ral

ster

oid

inje

ctio

ns a

nd c

hron

ic l

ower

-bac

k pa

in

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Bre

ivik

, et a

l., 1

976

II35

pat

ient

s w

/ inc

apac

itatin

g ba

ck p

ain

rand

omiz

ed to

20

ml b

upiv

acai

ne w

/ 80

mg

ster

oid

or 1

00 m

l sal

ine.

Up

No

sign

ific

ant s

hort

-ter

m d

iffe

renc

es b

twn

the

2 to

3 in

ject

ions

. Out

com

e m

easu

res

wer

e su

bjec

tive

pain

rel

ief

& c

hang

e in

phy

sica

l eva

luat

ion.

grou

ps.

Dal

las,

et a

l., 1

987

IIR

ando

miz

ed c

ontr

olle

d cr

oss-

over

stu

dy w

/ 20

patie

nts.

All

rece

ived

80

mg

met

hylp

redn

isol

one,

10

also

rec

eive

d

No

sign

ific

ant d

iffe

renc

es in

trea

tmen

t gro

ups

exce

pt

salin

e &

10

mor

phin

e w

/ cro

ss-o

ver

& r

epea

t tre

atm

ent i

n 6

wks

. Out

com

es w

ere

VA

S &

cha

nge

in p

hysi

cal

for

mor

phin

e at

6 w

ks o

nly.

exam

at m

ultip

le le

vels

up

to 6

mos

aft

er la

st in

ject

ion.

R

occo

, et a

l., 1

989

II22

pat

ient

s w

/ pos

tlam

inec

tom

y sy

ndro

me

rece

ived

50

mg

lidoc

aine

epi

dura

lly &

wer

e ra

ndom

ized

to 1

of

3 N

o gr

oup

had

effe

ctiv

e pa

in r

elie

f be

yond

1 m

o af

ter

grou

ps: 7

5 m

g tr

iam

cino

lone

, 8 m

g m

orph

ine,

or

both

. Out

com

e m

easu

res

wer

e su

bjec

tive

pain

eva

luat

ion,

&la

st in

ject

ion.

Stu

dy te

rmin

ated

due

to v

entil

ator

y fu

nctio

nal i

mpa

irm

ent u

p to

6 m

os a

fter

inje

ctio

n.

depr

essi

on in

mor

phin

e gr

oup.

Serr

ao, e

t al.,

199

2II

The

ther

apeu

tic e

ffec

ts o

f ep

idur

al m

ethy

lpre

dnis

olon

e (8

0 m

g) w

ere

com

pare

d w

/ int

rath

ecal

mid

azol

am (

2 m

g).

50–7

5% o

f pa

tient

s in

bot

h gr

oups

impr

oved

at 2

28

pat

ient

s ra

ndom

ized

to 1

of

thes

e 2

grou

ps in

the

doub

le-b

lind

stud

y. O

utco

me

mea

sure

s w

ere

the

MPQ

, m

os i

n 1

of 3

pri

mar

y ou

tcom

e m

easu

res.

The

V

AS,

& a

nalg

esic

adm

inis

trat

ion

befo

re, a

t 2 w

ks, &

at 2

mos

. E

SI g

roup

s ta

king

at l

east

as

muc

h an

alge

sic

med

i-ca

tion

as p

rior

to th

e st

udy.

TAB

LE

2L

umba

r fa

cet

inje

ctio

ns

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Jack

son,

et a

l., 1

988

II39

0 pa

tient

s w

/ low

-bac

k pa

in r

ecei

ved

face

t inj

ectio

ns &

com

plet

ed a

trea

tmen

t pro

toco

l. 12

7 va

riab

les

wer

e Fa

cet i

njec

tions

as

trea

tmen

t in

390

patie

nts

did

not

anal

yzed

. No

face

t syn

drom

e as

suc

h co

uld

be id

entif

ied

& f

acet

join

ts w

ere

not t

houg

ht to

be

a co

mm

on o

r re

veal

evi

denc

e fo

r a

face

t syn

drom

e or

for

the

sing

le s

ourc

e of

bac

k pa

in in

.90

% o

f pa

tient

s.fa

cet j

oint

as

an im

port

ant s

ourc

e of

bac

k pa

in in

the

maj

ority

of

patie

nts.

Lili

us, e

t al.,

199

0II

109

patie

nts

wer

e ra

ndom

ly a

ssig

ned

to 1

of

3 gr

oups

: Gro

up 1

, int

rafa

cet i

njec

tions

of

ster

oid

& a

nest

hetic

; Fa

cet i

njec

tions

are

a n

onsp

ecif

ic f

orm

of

trea

tmen

t G

roup

2, p

eric

apsu

lar

inje

ctio

ns o

f st

eroi

d &

ane

sthe

tic;

& G

roup

3, p

eric

apsu

lar

inje

ctio

n of

sal

ine

as a

con

trol

. fo

r lo

w-b

ack

pain

& g

ood

resu

lts d

epen

d on

aO

utco

mes

mea

sure

s w

ere

subj

ectiv

e pa

in s

cale

, wor

k, &

dis

abili

ty in

com

e. S

igns

of

inap

prop

riat

e be

havi

or

tend

ency

tow

ard

spon

tane

ous

regr

essi

on o

f ba

ckpr

ior

to in

ject

ion

wer

e be

st p

redi

ctor

of

outc

omes

& n

o di

ffer

ence

in o

vera

ll ou

tcom

es a

mon

g th

e 3

grou

ps.

pain

& o

n ps

ycho

soci

al a

spec

ts o

f ba

ck p

ain.

Car

ette

, et a

l., 1

991

II91

pat

ient

s w

/ chr

onic

pai

n w

ho r

espo

nded

to in

ject

ions

in th

e fa

cet j

oint

s w

/ a d

imin

utio

n in

pai

n w

ere

rand

om-

Inje

ctio

n of

met

hylp

redn

isol

one

into

the

face

t joi

nts

iz

ed to

inje

ctio

n w

/ ste

roid

or

inje

ctio

n w

/ sal

ine

cont

rol.

No

diff

eren

ce in

out

com

e m

easu

res

at 1

& 3

mos

btw

n is

of

little

trea

tmen

t val

ue.

the

2 gr

oups

& n

o di

ffer

ence

in s

usta

ined

impr

ovem

ent f

rom

Mo

1 to

Mo

6.

Mar

ks, e

t al.,

199

2II

86 p

atie

nts

w/ c

hron

ic lo

w-b

ack

pain

wer

e ra

ndom

ized

to a

fac

et b

lock

–onl

y w

/ ane

sthe

tic o

r in

traa

rtic

ular

in-

Face

t blo

cks

or in

traa

rtic

ular

inje

ctio

ns w

/ ane

sthe

tic

ject

ion

w/ s

tero

id &

ane

sthe

tic. I

mm

edia

te r

espo

nse

was

sam

e fo

r bo

th g

roup

s; s

tero

id g

roup

was

mar

gina

lly

& s

tero

id w

ere

equa

lly g

ood

diag

nost

ical

ly. B

oth

bette

r in

a p

ain

mea

sure

at 1

mo

(p ,

0.05

), &

by

3 m

os o

nly

2 pa

tient

s re

port

ed a

ny p

ain

relie

f.

wer

e eq

ually

uns

atis

fact

ory

trea

tmen

t for

chr

onic

ba

ck p

ain.

Gal

lagh

er, e

t al.,

199

4II

Pros

pect

ive

RC

Tof

41

patie

nts

by in

volv

ing

VA

S &

MPQ

. Pat

ient

s ha

d ch

roni

c ba

ck p

ain

& w

ere

sele

cted

by

posi

-In

pat

ient

s se

lect

ed b

y re

spon

se to

dia

gnos

tic f

acet

tiv

e re

spon

se to

dia

gnos

tic f

acet

inje

ctio

ns. P

atie

nts

rece

ivin

g R

F ab

latio

n w

ere

stat

istic

ally

sup

erio

r to

a p

la-

inje

ctio

ns, t

hose

w/ R

F ab

latio

n ha

d be

tter

out-

ce

bo c

ontr

ol a

t 1 &

6 m

os.

com

es a

t 1 &

6 m

os v

s pl

aceb

o co

ntro

l.V

an K

leef

, et a

l., 1

999

IPr

ospe

ctiv

e R

CT

doub

le b

lind

of 3

1 pa

tient

s w

/ $1

yr o

f lo

w-b

ack

pain

& p

ositi

ve r

espo

nse

to f

acet

blo

cks

In p

atie

nts

sele

cted

by

resp

onse

to d

iagn

ostic

fac

et

wer

e pl

aced

in a

n R

F gr

oup

or s

ham

con

trol

. Out

com

e m

easu

rem

ents

wer

e V

AS,

per

ceiv

ed im

prov

emen

t, na

r-in

ject

ions

, tho

se w

/ RF

abla

tion

had

supe

rior

co

tics

usag

e, &

OD

I. D

iffe

renc

es in

out

com

es w

ere

stat

istic

ally

impr

oved

in th

e tr

eatm

ent g

roup

at 3

, 6, &

12

mos

.ou

tcom

es a

t 3, 6

, & 1

2 m

os v

s sh

am c

ontr

ol.

Lec

lair

e, e

t al.,

200

1I

Pros

pect

ive

doub

le-b

lind

RC

Tof

70

patie

nts

w/ b

ack

pain

for

.3

mos

. All

resp

onde

d to

fac

et b

lock

s &

ran

dom

-In

pat

ient

s se

lect

ed b

y re

spon

se to

dia

gnos

tic f

acet

iz

ed to

an

RF

abla

tion

grou

p or

a s

ham

con

trol

gro

up. O

utco

me

mea

sure

s w

ere

OD

I &

Rol

and–

Mor

ris

inst

ru-

inje

ctio

ns, n

o di

ffer

ence

in o

utco

me

mea

sure

s at

4

men

ts a

long

w/ V

AS.

At 4

wks

no

diff

eren

ce in

VA

S or

OD

I sc

ores

& a

t 12

wks

no

diff

eren

ce b

twn

the

2 &

12

wks

.gr

oups

on

any

outc

ome

mea

sure

s.


months after treatment. One treatment group received anepidural injection of 80 mg prednisolone in 10 ml salinealong with 3 ml 5% dextrose delivered with an intrathecalinjection. The second treatment group received an epidu-ral injection of 10 ml normal saline epidurally and anintrathecal injection of 2 mg midazolam in 3 ml dextrose.Both groups noted significant pain improvement reflectedon the primary outcome measures at 2 weeks. The mida-zolam group was found to use less oral pain medication at2 months. No significant improvement over baseline wasnoted at 2 months for either group, although approximate-ly half of the patients in each group reported that they wereimproved. This paper provides Class III medical evidencethat ESIs and intrathecal midazolam injections are an ef-fective short-term treatment for low-back pain. It also pro-vides Class III evidence that such injections do not providelasting pain relief (beyond 2 weeks). Evidentiary Table 1summarizes the literature regarding ESIs.

Use of Lumbar Facet Injections for Chronic Low-BackPain From Degenerative Disease of the Lumbar Spine

Injections of the facet joints (zygapophysial joints) inthe lumbar spine are performed for the diagnosis andtreatment of chronic low-back pain. A “facet joint syn-drome” was described by Ghormley23 in 1933. The diag-nostic criteria and anatomical basis for this chronic low-back pain syndrome have not been precisely defined, andthe existence of such a syndrome has been debated.2,5,

15–17,27,29,37,40,42,60 Authors have suggested that older age,relief of back pain with recumbency, exacerbation of painon extension but not flexion, localized tenderness on pal-pation over a facet, radiological changes of degenerationat facet joints, and the lack of leg pain, muscle spasm, orpain with valsalva are all indicators of facet-related low-back pain.29,37,52

Schwarzer, et al.,60 published a prospective study inwhich clinical signs thought to be indicative of a lumbarfacet syndrome were investigated by performing selectivefacet injections. A double-injection technique was usedwith administration of a short- and a longlasting anestheticon separate occasions in the each patient. A patient was notthought to have true zygapophysial joint pain unless therewas a positive response (relief of pain with both injectionsfor a period of time consistent with the anesthetic used).One-hundred seventy-six consecutive patients with low-back pain were studied. A history and physical examina-tion were obtained in each patient prior to injection. TheVAS was used to measure the response to the injections. Athreshold of greater than 50% improvement was consid-ered significant. Of the 176 patients injected with ligno-caine, 83 (47%) responded to the injections but only 26(15%) reported a 50% or greater relief of back pain. Con-firmatory blocks were performed in 71 of these 83 patientswith bupivacaine (the remaining 12 patients were droppedfrom the study for a variety of reasons). Of the patientswho received both injections, only seven (4% of the origi-nal sample) achieved relief of back pain with both blocks.There was no statistically significant association betweenresponse to any or both blocks and any single finding onhistory or physical examination. Finally, combinations ofclinical features were assessed by logistical regressionanalysis and no model could be generated that would dis-



TAB

LE

3Tr

igge

r po

int

inje

ctio

ns

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Ham

erof

f, e

t al.,

198

1II

15 p

atie

nts

w/ l

ow-b

ack

pain

wer

e ra

ndom

ly a

ssig

ned

to 1

of

3 gr

oups

in a

dou

ble-

blin

d cr

oss-

over

stu

dy:

Inje

ctio

n of

an

anes

thet

ic w

as p

refe

rred

to a

pla

cebo

Gro

up 1

, inj

ecte

d w

/ bup

ivac

aine

; Gro

up 2

, inj

ecte

d w

/ etid

ocai

ne; &

Gro

up 3

, inj

ecte

d w

/ sal

ine

as a

con

trol

. in

ject

ion

of s

alin

e.O

utco

mes

wer

e m

easu

red

on s

ubje

ctiv

e re

spon

ses

to 6

pai

n-re

late

d ca

tego

ries

. Inj

ectio

n w

/ ane

sthe

tic w

as s

u-pe

rior

to th

at w

/ sal

ine.

So

nne,

et a

l., 1

985

II30

pat

ient

s w

ere

rand

omly

ass

igne

d to

1 o

f 2

inje

ctio

n gr

oups

. Gro

up 1

, met

hylp

redn

isol

one

w/ l

igno

cain

e;

Inje

ctio

n of

ste

roid

& a

nest

hetic

was

sig

nifi

cant

ly

Gro

up 2

, iso

toni

c sa

line

inje

ctio

n. O

utco

mes

wer

e m

easu

red

by V

AS,

sel

f-as

sess

men

t, &

spi

nal R

OM

sho

rt

bette

r fo

r sh

ort-

term

rel

ief

of lo

w-b

ack

pain

than

te

rm. V

AS

& s

elf-

asse

ssm

ent w

ere

sign

ific

antly

bet

ter

for

the

ster

oid

inje

ctio

n &

no

chan

ge in

spi

nal R

OM

sa

line

inje

ctio

n.fo

r ei

ther

gro

up.

Gar

vey,

et a

l., 1

989

II63

pat

ient

s w

/ low

-bac

k pa

in w

ere

trea

ted

for

$4

wks

& d

id n

ot im

prov

e. R

ando

mly

ass

igne

d to

1 o

f 4

grou

ps:

TPI

s de

crea

se lo

w-b

ack

pain

, but

it d

oes

not a

ppea

rG

roup

1,T

PI o

f lid

ocai

ne; G

roup

2, i

njec

tion

of li

doca

ine

w/ s

tero

id; G

roup

3, d

ry n

eedl

ing

of th

e T

P; &

Gro

up

nece

ssar

y to

inje

ct a

ny d

rug

to a

chie

ve th

is.

4, v

apoc

oole

nt s

pray

w/ a

cupr

essu

re. D

ry n

eedl

e gr

oup

resp

onde

d at

leas

t as

wel

l as

the

grou

ps w

/ med

icat

ion

at in

ject

ion

(63

vs 4

2%; p

,0.

09).

Col

lee,

et a

l., 1

991

II41

pat

ient

s ra

ndom

ized

to a

n in

ject

ion

w/ 0

.5%

lign

ocai

ne o

r an

equ

al a

mou

nt o

f sa

line.

Fol

low

up

was

at 2

wks

T

PI w

/ ane

sthe

tic is

mor

e ef

fect

ive

than

sal

ine

in r

e-w

/ a V

AS

& s

ubje

ctiv

e pa

in s

core

. Sig

nifi

cant

ly m

ore

patie

nts

wer

e im

prov

ed a

fter

lign

ocai

ne (

52%

) th

an a

f-du

cing

bac

k pa

in if

adm

inis

tere

d by

som

eone

te

r sa

line

inje

ctio

ns (

39%

) (p

, 0

.05)

. Sub

grou

p an

alys

is f

ound

the

effe

ct o

nly

appl

ied

to p

atie

nts

trea

ted

in a

fa

mili

ar &

com

pete

nt w

/ tec

hniq

ue.

rheu

mat

olog

y gr

oup

& n

ot in

a f

amily

pra

ctic

e gr

oup.


criminate between patients who responded to the blocksand those who did not. This findings were true even whenapplied to only the 4% of patients who responded to bothblocks. The one definitive finding in the study was that nopatient with midline back pain responded to either facetblock. The authors concluded that the facet joint is animportant contributor to low-back pain in a small minorityof patients.

Despite the controversy regarding the existence of fa-cet-related low-back pain, the clinical use of facet injec-tions in the treatment of chronic low-back pain remainscommon. Facet injections are used for both diagnosticpurposes and as a treatment. Several authors have investi-gated the effectiveness of facet blocks as a diagnostic toolto predict the outcome of subsequent RF facet rhizotomy(RF ablation) and/or lumbar spinal fusion.18,21,36,41,49,70

A prospective, double-blind RCT on this topic was re-ported by Gallagher, et al.21 In their study, 60 patients un-derwent diagnostic facet joint injections. Forty-one patientsreported either a strongly positive response or an equivocalresponse. These 41 patients were randomized to receive aninjection of local anesthetic (placebo) or injection of anes-thetic plus RF ablation. Outcomes were assessed with theMPQ and the VAS at 1 and 6 months following injection orablation. Patients who had a strongly positive response tothe initial injection and were randomized to the ablationgroup did better on both outcome measures at both timepoints compared with similar patients in the placebo group.There was no positive effect of RF ablation in patients withan equivocal response to the initial injection. In fact, pa-tients with an equivocal response to the initial injection didworse with RF ablation than similar patients treated withplacebo. The magnitude of this difference is identical to themagnitude of the beneficial effect enjoyed by the strongresponse/RF ablation group; however, it does not reach sta-tistical significance due to very small group sizes. Thisstudy provides Class I medical evidence supporting thehypothesis that RF ablation is more effective than injectionalone for the treatment of low-back pain in patients with astrongly positive response to an initial facet joint injection.It provides Class III medical evidence suggesting that RFablation is contraindicated in patients with an equivocalresponse to facet joint injection. Finally, it also providesClass III evidence supporting the ability of a strongly pos-itive response to facet injection to predict a positive res-ponse to RF ablation.

Van Kleef, et al.,70 studied 31 patients with at least 1year of chronic back pain. All had responded to diagnos-tic facet blocks and were randomly assigned to one of twogroups. In the treatment group RF probes were insertedand they received RF ablation. The control group wentthrough the same procedure without receiving RF abla-tion. Both the patients and their treating doctors wereblinded as to treatment. Outcome measures were the VAS,the ODI, a quality-of-life questionnaire, patient satisfac-tion, and quantification of the amount of narcotic usage.Measurements were made prior to and at 8 weeks and 3,6, and 12 months after treatment. Success was defined asa two-point reduction in VAS score or 50% reduction inperceived pain. More patients achieved successful out-comes in the RF ablation group than in the control groupat 3, 6, and 12 months (66% treatment compared with 38%control). The RF ablation patients also reported decreased

analgesic use and improved disability scores. The authorsconcluded that facet blocks were useful in selecting pa-tients who would respond to RF ablation. They postulatedthat their results would have been even more convincingif the more stringent selection criteria of Bogduk5 hadbeen used in their study. This paper provides Class IIImedical evidence of the diagnostic utility of blocks forpredicting RF ablation results as patients were selected forinclusion in the study based on the positive results of facetblocks. The paper provides Class I medical evidence thatRF ablation is more effective than placebo for the treat-ment of low-back pain in patients with positive facet in-jections.

A larger RCT examining this issue was published byLeclaire, et al.36 In their randomized, double-blind con-trolled study, 70 patients with more than 3 months of backpain who had responded to diagnostic facet blocks wereassigned to a treatment group of RF ablation or a shamcontrol group. The outcome measures were the functionaldisabilities scale on the ODI, Roland–Morris instruments,and VAS measurements. Although the RF ablation groupdid better when assessed 2 weeks following treatment,there were no statistical differences between the twogroups on the ODI or VAS measures 4 weeks followingtreatment. At 12 weeks posttreatment, there were no dif-ferences between the two groups in any of the outcomesmeasures. At 12 weeks 94.3% of the patients were evalu-ated. This paper provides Class I medical evidence thatRF ablation is not more effective than placebo for thelong-term treatment of low-back pain in patients with pos-itive responses to diagnostic facet injections. The paper al-so provides Class III evidence refuting the ability of diag-nostic facet injections to predict outcome following RFablation.

North, et al.,49 retrospectively studied their patients withlumbar facet blocks and RF denervation. An outside ob-server assessed outcomes in 82 patients with a mean fol-low-up time of 3.2 years. Of the 82 patients in their re-view, 42 had positive diagnostic facet block with greaterthan a 50% reduction in back pain. These patients weretreated with RF ablation. Of these, 45% reported at least50% relief of pain on a VAS at last follow up. Of the 40patients in whom pain relief failed from facet injectionsand thus did not undergo ablation after facet block, 13%improved at least 50% at last follow up. This retrospectivestudy provides Class III medical evidence supporting theuse of RF ablation of facets in patients with chronic low-back pain selected by diagnostic facet injections.

Other authors have assessed the value of facet blocksfor predicting outcome following lumbar fusion. Essesand Moro18 retrospectively studied a cohort of 296 patientswho underwent diagnostic facet blocks between 1980 and1989 to investigate the source of their chronic back pain.Of these 296 patients, 126 could be contacted and res-ponded to a questionnaire. Patients were classified intogroups: those who had undergone lumbar spinal fusionand those who had continued with nonsurgical care. Anextensive statistical analysis failed to demonstrate anycorrelation between a successful facet block and the re-sults of subsequent lumbar spinal fusion. Furthermore, nocorrelations were found between a successful facet blockand the results of any of the conservative care measuresthe nonoperative patients received. This paper provides


Injection therapies

711


Class III medical evidence indicating that facet joint injec-tions should not be used as a diagnostic tool to determinethe need for or potential benefit of lumbar spinal fusion.Jackson29 studied a group of 31 patients who were treatedwith attempted lumbar spinal fusion after facet injections.Twenty-one of the patients responded favorably to thediagnostic injections and 10 did not. As a group, all 31 pa-tients had significantly improved on pain and functionscores after the surgery, but there was no difference in out-come between the 26 patients who responded to the facetinjections and the 10 who did not. Jackson concluded thatfacet injections were not predictive of outcome followinglumbar spinal fusion for back pain. This paper also pro-vides Class III medical evidence indicating that facetinjections are not useful for predicting outcome followingattempted lumbar spinal fusion surgery.

Lovely and Rastogi41 evaluated 91 patients with chronicback pain lasting for more than 6 months. All had beentreated with at least 3 months of aggressive conservativecare and who reportedly were improved with a subsequenttrial of lumbar bracing. All 91 patients underwent facetblocks with marcaine. A positive response was defined asrelief of pain greater than 70% at 6 hours after injection.Blocks were repeated, and to be considered a surgical can-didate, a patient had to respond to a block with greater than70% pain relief on three separate occasions. Twenty-eightpatients with a mean duration of 6.1 years of back painwere ultimately selected for lumbar spinal fusion. Pain re-lief following surgery in these 28 patients was measured bypatient satisfaction score and functional improvement wasdetermined with the use of the Prolo functional outcomescore. A successful fusion was achieved in 23 of 28 pa-tients, of whom 95% rated their results as good or excel-lent. The five patients with pseudarthrosis all had a pooroutcome by their own estimates and Prolo scores. Thisstudy did not include an appropriate control group andtherefore provides Class III medical evidence supportingthe use of facet injections as a diagnostic tool to predict out-come following successful lumbar fusion surgery.

Several authors have investigated facet injections as atreatment for chronic low-back pain.9,30,38,43 Jackson, etal.,30 performed facet injections in 454 patients with low-back pain thought to originate in the facet joint. Threehundred ninety patients completed a protocol in which127 clinical variables were studied. After analysis of all ofthese variables, the authors were unable to identify a clin-ical facet syndrome. They concluded that the facet jointswere not the primary source of back pain in the great ma-jority of patients studied (. 90%). Only 30 (7.7%) of thepatients selected for injection based on clinical criteria hadcomplete relief of their symptoms (assessed within 4hours of injection). This compares with 22 (5%) similarlyselected patients whose symptoms were made worse withinjection. This report provides Class III medical evidencethat facet injections are not an effective treatment for pa-tients with low-back pain, even if the patients fulfill clini-cal criteria for a “facet syndrome.” Lilius, et al.,38 reportedtheir results in 109 patients with unilateral low-back pain(mean duration 13.4 months) who they randomly assignedto three treatment groups. Group 1 received facet injec-tions of cortisone and a local anesthetic in the facet joint,Group 2 received the same injection pericapsularly, andGroup 3 received physiological saline injected in the facet

joint. Pain (subjective), work attendance, and a disabilityscore were the outcome measures used in their study. Allmeasures improved to the same degree across all groups.Seventy patients experienced some initial relief, 36% ofwhom reported persisting relief at 3-month follow up. Theauthors concluded that facet injections were a nonspecificmethod of treatment for low-back pain. The majority ofgood results observed in the study resulted from a tenden-cy for spontaneous regression of low-back pain and frompsychosocial aspects of back pain. Their report providesClass I medical evidence that facet injections are not aneffective treatment for low-back pain in patients withchronic unilateral low-back pain.

A prospective, randomized controlled study of the treat-ment of chronic low-back pain by facet joint injectionswas also performed by Carette, et al.9 Ninety-five patientswho reported immediate relief of their pain when an anes-thetic was injected into the facet joints were randomizedto one of two groups. Group I received a facet injection ofmethylprednisolone into the same joints, and Group II re-ceived an injection of saline into the same joints. Thesepatients were followed for 6 months with pain VAS scores,measures of back mobility and function, and the MPQ.After 1 month and 3 months, none of the outcome mea-sures differed between the two groups. At 6 months, thesteroid group reported more improvement but also had un-dergone more concurrent interventions. There was no sta-tistical difference between the two groups from the 1st tothe 6th month. This paper provides Class I medical evi-dence that injection of steroids into the facet joint is not aneffective treatment for low-back pain, even among pa-tients selected based on a prior favorable response to ananesthetic facet joint injection. Marks, et al.,43 randomized84 patients with refractory back pain into either a facetblock group with anesthetic only or a group treated withanesthetic and a steroid injection into the facet. There wasno difference in the immediate response to the injectionand at 1 month the steroid group had a marginally im-proved response to the injection as measured by a VAS.By 3 months, however, only two patients continued toexperience any relief. The authors concluded that neitherinjection with steroid and anesthetic nor anesthetic alonewas a satisfactory treatment for chronic back pain. Evi-dentiary Table 2 summarizes the reviewed literature onfacet joint injections.

The Use of Local Lumbar Injections (TPIs) in theTreatment of Chronic Low-Back Pain due to DegenerativeDisease of the Lumbar Spine

The existence of myofascial TPs as a source of low-back pain has been studied by Simons and Travell.63–66

According to these authors, an active myofascial TP isdiagnosed by focal tenderness on palpation, by a restrict-ed stretch ROM, and by a local twitch response on needlestimulation. These authors have stated that treatment byTPI can result in long-term relief of low-back pain, but“only if mechanical and systemic perpetuating factors arecorrected.”64 Many reviews of the conservative treatmentof chronic low-back pain reference TPIs as a potentialtreatment modality, with varying degrees of enthusi-asm.1,14,19,34,62,73 Five RCTs in which TPIs were investigatedas treatment for low-back pain were identified.11,22,25,50,67




Hameroff, et al.,25 randomized 15 patients into three TPIgroups in a double-blind fashion. Group 1 received bupi-vacaine, Group 2 received etidocaine, and Group 3 re-ceived a saline control injection. A subjective assessmentof pain in six categories was used as the outcome measureand was obtained 15 minutes, 24 hours, and 7 days afterinjection. Trigger point injections with an anesthetic weremore effective for pain relief than the control injection ofsaline. Sonne, et al.,67 prospectively investigated 30 pa-tients with back pain of at least 1 month’s duration andrandomized them into two treatment groups in a double-blind fashion. Group I received an injection of methyl-prednisolone with lignocaine and Group II received aninjection of isotonic saline. Outcome measures were aVAS, spinal ROM, and patient self-assessment. Signifi-cant decreases in VAS scores and patients’ self-assessedlevel of pain were reported in the steroid and anestheticgroup only. There was no change in spinal ROM in eithergroup. Garvey, et al.,22 performed a prospective, random-ized, double-blind evaluation of 63 patients with low-backpain who had been treated conservatively for at least 4weeks without improvement. Patients were randomizedinto one of four groups. Patients in Group I were treatedwith a TPI with lidocaine, those in Group II were treatedwith a TPI with lidocaine and a steroid, patients in GroupIII were treated with dry needling of the TP, and those inGroup IV were treated with acupressure and vapocoolentspray. Sixty-three percent of the patients reported a de-crease in pain based on a VAS with dry needling com-pared with 42% with a drug injection, a difference thatwas not statistically significant (p , 0.09). The authorsconcluded that TPIs appear to have some value in treatinglow-back pain, but that injection of a drug is not necessar-ily a component of that treatment. Finally, Collee, et al.,11

randomly injected 41 patients with either 0.5% lignocaineor an equivalent amount of saline as a control in a 2-week-long study. Patients and treating doctors were blinded asto treatment. Outcome measures were a pain score and apain-intensity score (VAS) administered 2 weeks follow-ing the procedure. Fifty-two percent of the patients in thelignocaine group were improved at 2 weeks comparedwith 30% of the patients treated with saline (p , 0.05). Asubgroup analysis revealed that these effects were identi-fied in patients treated by a rheumatology group but not inpatients treated by a family practice group. These studieson TPIs were all performed in patients with the relativelyacute onset of low-back pain who were selected based onthe presence of TPs and who were assessed shortly afterinjection. As such, they provide only Class III medicalevidence suggesting that TPIs may be an effective short-term treatment for selected patients with low-back pain.Evidentiary Table 3 summarizes the reviewed literature onTPIs.

Summary

In summary, there is no meaningful evidence in the med-ical literature that the use of epidural injections is of anylong-term value in the treatment of patients with chroniclow-back pain. The literature does indicate that the use oflumbar epidural injections can provide short-term relief inselected patients with chronic low-back pain.

There is evidence that suggests that facet joint injec-tions can be used to predict outcome after RF ablation ofa facet joint. The predictive ability of facet joint injectionsdoes not appear to apply to lumbar fusion surgery. No evi-dence exists to support the effectiveness of facet injectionsin the treatment of patients with chronic low-back pain.

There is conflicting evidence suggesting that the use oflocal TPIs can be effective for the short-term relief of low-back pain. There are no data to suggest that TPIs witheither steroids or anesthetics alone provide lasting benefitfor patients suffering from chronic low-back pain.


The effectiveness of ESIs for the treatment of chronicback pain should be addressed through a randomized clin-ical trial comparing the clinically most common injectiontechnique using steroids combined with anesthetic to anappropriate control group. Given the positive effect notedin the TP studies with needle stimulation, needle place-ment alone may be a viable control group.

The diagnostic utility of facet injections in predictingtherapeutic outcomes for RF ablation appears promising.The lack of a predictive effect for lumbar fusion surgery isproblematic. The relationship between RF and lumbarfusion surgery in terms of mechanism of pain relief repre-sents a fertile field for further clinical and basic scienceresearch.

The fact that needle placement alone provides a benefi-cial effect in some patients treated with TPIs is fascinat-ing. Further research into the mechanism of pain relief inthese patients is warranted.

References

1. Alvarez DJ, Rockwell PG: Trigger points: diagnosis and man-agement. Am Fam Physician 65:653–660, 2002

2. Barnsley L: Steroid injections: effect on pain of spinal origin.Best Pract Res Clin Anaesthesiol 16:579–596, 2002

3. Beliveau P: A comparison between epidural anaesthesia withand without corticosteroid in the treatment of sciatica.Rheumatol Phys Med 11:40–43, 1971

4. Benzon HT: Epidural steroid injections for low back pain andlumbosacral radiculopathy. Pain 24:277–295, 1986

5. Bogduk N: Diagnostic nerve blocks in chronic pain. Best PractRes Clin Anaesthesiol 16:565–578, 2002

6. Breivik H, Helsa P, Molnar I, et al: Treatment of chronic lowback pain and sciatica. Comparison of caudal epidural injec-tions of bupivacaine and methylprednisolone with bupivacainefollowed by saline. Adv Pain Res Ther 1:972–932, 1976

7. Bush K, Hillier S: A controlled study of caudal epidural injec-tions of triamcinolone plus procaine for the management ofintractable sciatica. Spine 16:572–575, 1991

8. Cannon DT, Aprill CN: Lumbosacral epidural steroid injec-tions. Arch Phys Med Rehabil 81 (Suppl 1):S87–S100, 2000

9. Carette S, Leclaire R, Marcoux S, et al: Epidural corticosteroidinjections for sciatica due to herniated nucleus pulposus. NEngl J Med 336:1634–1640, 1997

10. Carette S, Marcoux S, Truchon R, et al: A controlled trial ofcorticosteroid injections into facet joints for chronic low backpain. N Engl J Med 325:1002–1007, 1991

11. Collee G, Dijkmans BA, Vandenbroucke JP, et al: Iliac crestpain syndrome in low back pain. A double blind, randomizedstudy of local injection therapy. J Rheumatol 18:1060–1063,1991


Injection therapies

713


12. Cuckler JM, Bernini PA, Wiesel SW, et al: The use of epiduralsteroids in the treatment of lumbar radicular pain. A prospec-tive, randomized, double-blind study. J Bone Joint Surg Am67:63–66, 1985

13. Dallas TL, Lin RL, Wu WH, et al: Epidural morphine and me-thylprednisolone for low-back pain. Anesthesiology 67:408–411, 1987

14. Deyo RA: Drug therapy for back pain. Which drugs help whichpatients? Spine 21:2840–2850, 1996

15. Dreyer SJ, Dreyfuss PH: Low back pain and the zygapophysial(facet) joints. Arch Phys Med Rehabil 77:290–300, 1996

16. Dreyfuss PH, Dreyer SJ, Herring SA: Lumbar zygapophysial(facet) joint injections. Spine 20:2040–2047, 1995

17. el-Khoury GY, Renfrew DL: Percutaneous procedures for thediagnosis and treatment of lower back pain: diskography, facet-joint injection, and epidural injection. AJR 157:685–691, 1991

18. Esses SI, Moro JK: The value of facet joint blocks in patientselection for lumbar fusion. Spine 18:185–190, 1993

19. Fast A: Low back disorders: conservative management. ArchPhys Med Rehabil 69:880–891, 1988

20. Furman MB, O’Brien EM, Zgleszewski TM: Incidence of in-travascular penetration in transforaminal lumbosacral epiduralsteroid injections. Spine 25:2628–2632, 2000

21. Gallagher J, Petriccione di Vadi P, Wedley J, et al: Radiofre-quency facet joint denervation in the treatment of low backpain. A prospective controlled double-blind study to assess itsefficacy. Pain Clin 7:193–198, 1994

22. Garvey TA, Marks MR, Wiesel SW: A prospective, random-ized, double-blind evaluation of trigger-point injection therapyfor low-back pain. Spine 14:962–964, 1989

23. Ghormley R: Low back pain with special reference to the artic-ular facets, with presentation of an operative procedure. JAMA101:1773–1777, 1933

24. Glynn C, Dawson D, Sanders R: A double-blind comparisonbetween epidural morphine and epidural clonidine in patientswith chronic non-cancer pain. Pain 34:123–128, 1988

25. Hameroff SR, Crago BR, Blitt CD, et al: Comparison of bupi-vacaine, etidocaine, and saline for trigger-point therapy. An-esth Analg 60:752–755, 1981

26. Hesla E, Breivik H: [Epidural analgesia and epidural steroidinjection for treatment of chronic low back pain and sciatica.]Tidsskr Nor Laegeforen 99:936–939, 1979 (Nor)

27. Hildebrandt J: [Relevance of nerve blocks in treating and diag-nosing low back pain—is the quality decisive?] Schmerz 15:474–483, 2001 (Ger)

28. Hopayian K, Mugford M: Conflicting conclusions from twosystematic reviews of epidural steroid injections for sciatica:which evidence should general practitioners heed? Br J GenPract 49:57–61, 1999 (Errratum in Br J Gen Pract 49:840,1999)

29. Jackson RP: The facet syndrome. Myth or reality? Clin OrthopRelat Res 279:110–121, 1992

30. Jackson RP, Jacobs RR, Montesano PX: 1988 Volvo award inclinical sciences. Facet joint injection in low-back pain. Aprospective statistical study. Spine 13:966–971, 1988

31. Johnson BA: Image-guided epidural injections. NeuroimagingClin N Am 10:479–491, 2000

32. Johnson BA, Schellhas KP, Pollei SR: Epidurography and ther-apeutic epidural injections: technical considerations and experi-ence with 5334 cases. AJNR 20:697–705, 1999

33. Kepes ER, Duncalf D: Treatment of backache with spinal injec-tions of local anesthetics, spinal and systemic steroids. A re-view. Pain 22:33–47, 1985

34. Kim PS: Role of injection therapy: review of indications fortrigger point injections, regional blocks, facet joint injections,and intra-articular injections. Curr Opin Rheumatol 14:52–57, 2002

35. Koes BW, Scholten RJ, Mens JM, et al: Efficacy of epiduralsteroid injections for low-back pain and sciatica: a systematic

review of randomized clinical trials. Pain 63:279–288, 199536. Leclaire R, Fortin L, Lambert R, et al: Radiofrequency facet

joint denervation in the treatment of low back pain: a placebo-controlled clinical trial to assess efficacy. Spine 26:1411–1417,2001

37. Lewinnek GE, Warfield CA: Facet joint degeneration as a causeof low back pain. Clin Orthop Relat Res 213:216–222, 1986

38. Lilius G, Harilainen A, Laasonen EM, et al: Chronic unilaterallow-back pain. Predictors of outcome of facet joint injections.Spine 15:780–782, 1990

39. Linck CA: The effectiveness of epidural steroid injections forthe treatment of back pain. J Post Anesth Nurs 2:249–256,1987

40. Lippitt AB: The facet joint and its role in spine pain. Man-agement with facet joint injections. Spine 9:746–750, 1984

41. Lovely T, Rastogi P: The value of provocative facet blocking asa predictor of success in lumbar spine fusion. J Spinal Disord10:512–517, 1997

42. Manchikanti L: Facet joint pain and the role of neural blockadein its management. Curr Rev Pain 3:348–358, 1999

43. Marks RC, Houston T, Thulbourne T: Facet joint injection andfacet nerve block: a randomised comparison in 86 patients withchronic low back pain. Pain 49:325–328, 1992

44. Mathews JA, Mills SB, Jenkins VM, et al: Back pain and sciat-ica: controlled trials of manipulation, traction, sclerosant andepidural injections. Br J Rheumatol 26:416–423, 1987

45. Meadeb J, Rozenberg S, Duquesnoy B, et al: Forceful sacrococ-cygeal injections in the treatment of postdiscectomy sciatica. Acontrolled study versus glucocorticoid injections. Joint BoneSpine 68:43–49, 2001

46. Mulligan KA, Rowlingson JC: Epidural steroids. Curr PainHeadache Rep 5:495–502, 2001

47. Nelemans PJ, de Bie RA, de Vet HC, et al: Injection therapy forsubacute and chronic benign low back pain. Cochrane Data-base Syst Rev 2:CD001824, 2000

48. Nelemans PJ, deBie RA, deVet HC, et al: Injection therapy forsubacute and chronic benign low back pain. Spine 26:501–515,2001

49. North RB, Han M, Zahurak M, et al: Radiofrequency lumbar fa-cet denervation: analysis of prognostic factors. Pain 57:77–83,1994

50. Ongley MJ, Klein RG, Dorman TA, et al: A new approach tothe treatment of chronic low back pain. Lancet 2:143–146,1987

51. Renfrew DL, Moore TE, Kathol MH, et al: Correct placementof epidural steroid injections: fluoroscopic guidance and con-trast administration. AJNR 12:1003–1007, 1991

52. Revel M: Value of percutaneous treatments of the lumbar spinein back and nerve root pain. Semin Musculoskelet Radiol 1:349–354, 1997

53. Revel M, Auleley GR, Alaoui S, et al: Forceful epidural injec-tions for the treatment of lumbosciatic pain with post-operativelumbar spinal fibrosis. Rev Rhum Engl Ed 63:270–277, 1996

54. Riew KD, Yin Y, Gilula L, et al: The effect of nerve-root injec-tions on the need for operative treatment of lumbar radicularpain. A prospective, randomized, controlled, double-blind stu-dy. J Bone Joint Surg Am 82:1589–1593, 2000

55. Ringsdal VS, Nielsen NA, Slot O, et al: [Epidural glucocor-ticoid injection in lumbago sciatica.] Ugeskr Laeger 159:5653–5657, 1997 (Dan)

56. Rocco AG, Frank E, Kaul AF, et al: Epidural steroids, epiduralmorphine and epidural steroids combined with morphine in thetreatment of post-laminectomy syndrome. Pain 36:297–303,1989

57. Rogers P, Nash T, Schiller D, et al: Epidural steroids for sciati-ca. Pain Clin 5:67–72, 1992

58. Rozenberg S, Dubourg G, Khalifa P, et al: Efficacy of epiduralsteroids in low back pain and sciatica. A critical appraisal by a French task force of randomized trials. Critical Analysis




Group of the French Society for Rheumatology. Rev RhumEngl Ed 66:79–85, 1999

59. Schneeberger A, Gerster JC, So A: [General practice of lum-bosacral peridural infiltrations in rheumatology: considerationsbased on a review of the literature.] Schweiz Rundsch MedPrax 87:476–480, 1998 (Fr)

60. Schwarzer AC, Aprill CN, Derby R, et al: Clinical features ofpatients with pain stemming from the lumbar zygapophysialjoints. Is the lumbar facet syndrome a clinical entity? Spine 19:1132–1137, 1994

61. Serrao JM, Marks RL, Morley SJ, et al: Intrathecal midazolamfor the treatment of chronic mechanical low back pain: a con-trolled comparison with epidural steroid in a pilot study. Pain48:5–12, 1992

62. Shah RV, Ericksen JJ, Lacerte M: Interventions in chronic painmanagement. 2. New frontiers: invasive nonsurgical interven-tions. Arch Phys Med Rehabil 84 (Suppl 1):S39–S44, 2003

63. Simons DG, Travell JG: Myofascial origins of low back pain. 1.Principles of diagnosis and treatment. Postgrad Med 73:66,8–70, 3 passim, 1983

64. Simons DG, Travell JG: Myofascial origins of low back pain. 2.Torso muscles. Postgrad Med 73:81–92, 1983

65. Simons DG, Travell JG: Myofascial origins of low back pain. 3. Pelvic and lower extremity muscles. Postgrad Med 73:99–105, 108, 1983

66. Simons DG, Travell J: Myofascial trigger points, a possible ex-planation. Pain 10:106–109, 1981

67. Sonne M, Christensen K, Hansen SE, et al: Injection of steroidsand local anaesthetics as therapy for low-back pain. Scand JRheumatol 14:343–345, 1985

68. Spaccarelli KC: Lumbar and caudal epidural corticosteroid in-jections. Mayo Clin Proc 71:169–178, 1996

69. Vad VB, Bhat AL, Lutz GE, et al: Transforaminal epiduralsteroid injections in lumbosacral radiculopathy: a prospectiverandomized study. Spine 27:11–16, 2002

70. Van Kleef M, Barendse G, Kessels A, et al: Randomized trial ofradiofrequency lumbar facet denervation for chronic low backpain. Spine 24:1937–1942, 1999

71. Weinstein SM, Herring SA, Derby R: Contemporary conceptsin spine care. Epidural steroid injections. Spine 20:1842–1846,1995

72. White AH, Derby R, Wynne G: Epidural injections for the diag-nosis and treatment of low-back pain. Spine 5:78–86, 1980

73. Wittenberg RH, Steffen R, Ludwig J: [Injection treatment ofnon-radicular lumbalgia.] Orthopade 26:544–552, 1997 (Ger)




Injection therapies

715


Recommendations


Guidelines. The short-term use of a rigid lumbar support(1–3 weeks) is recommended as a treatment for low-backpain of relatively short duration (, 6 months). The use ofa lumbar brace for patients with chronic low-back pain isnot recommended because there is no pertinent medicalevidence of any long-term benefit or evidence that bracetherapy is effective in the treatment of patients with chron-ic (. 6 months) low-back pain.

Options. 1) Lumbar braces are recommended as ameans of decreasing the number of sick days lost due tolow-back pain among workers with a previous lumbarinjury. They are not recommended as a means of decreas-ing low-back pain in the general working population. 2)The use of lumbar brace therapy as a preoperative diag-nostic tool to predict outcome following lumbar fusion

surgery is not recommended. 3) The use of transpedicularexternal fixation as a tool to predict outcome followinglumbar fusion surgery is not recommended.

Rationale

Lumbosacral supports or back braces have been usedfor the prevention and treatment of a wide variety ofdegenerative disorders of the lumbar spine.11 The potentialmechanisms of action for brace therapy include limitingspinal ROM, correcting posture and deformity, preventinggross trunk motion, increasing the intraabdominal pres-sure, reducing force exerted by trunk muscles, and pro-viding soft-tissue massage and heat.23,28,42 Critics of lumbarsupports have argued that braces may provide workerswith a false sense of support or allow muscles to waste,thereby increasing the incidence of injury.26 Although themechanism of action of lumbar supports remains open todebate, their clinical utility in the prevention and treatmentof low-back pain must be determined to justify their use.22

Braces have been used for the preoperative evaluationof patients in an attempt to predict outcome followingfusion surgery. They are also applied following lumbarsurgery to promote fusion.9,19 Because lumbar orthoses donot eliminate motion in the lumbar spine, their utility hasbeen questioned.2,3 The purpose of this review is to exam-


716

Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 14: brace therapy as an adjunct to or substitute for lumbar fusion



KEY WORDS • lumbar spine • fusion • orthotic device • brace therapy •practice guidelines


Abbreviations used in this paper: NPV = negative predictivevalue; ODI = Oswestry Disability Index; PLF = posterolateral fu-sion; PPV = positive PV; RSA = roentgen stereophotogrammetricanalysis; ROM = range of motion; TEPF = temporary external pedi-cle fixation; TMS = transcutaneous muscle stimulation; VAS =visual analog scale.


ine the medical evidence for the utility of brace therapy asa treatment of low-back pain, as a predictor of outcomefollowing lumbar fusion surgery, and as an adjunct to lum-bar fusion procedures.

Search Criteria

A computerized search of the National Library of Med-icine from 1966 to 2003 was conducted using the follow-ing search terms: “lumbar support and low-back pain,”“brace and low-back pain,” and “orthosis and low-backpain.” After discarding duplicates, 760 papers were identi-fied. After reviewing the abstracts of each paper, 17 rele-vant studies were identified. We then searched the databaseusing the following phrases: “orthosis and fusion,” “braceand fusion,” “external lumbar fixation and fusion,” and“lumbar support and fusion.” After discarding duplicates,1951 papers were identified. After reviewing the abstractsof each paper, 12 relevant studies were identified. Severalreview papers, metaanalyses, biomechanical studies, tech-nical notes, and small case series serve as supporting data.The bibliography of each paper was reviewed and other rel-evant studies were identified. All clinical studies providingClass III medical evidence or better regarding the use oflumbar brace therapy for the prevention and treatment oflow-back pain, for the prediction of outcome followinglumbar fusion surgery, and as an adjunct to fusion surgeryare summarized in Tables 1 through 4.


Lumbar braces have been used as a means of prevent-ing low-back pain in industrial workers.18,42 Van Poppel, etal.,43 randomized 282 individuals empoyed as baggagehandlers into one of four groups: 1) education and lumbarbrace; 2) education; 3) lumbar brace; and 4) no interven-tion. Soft lumbar braces were worn during working hoursfor a 6-month period by workers randomized to Groups 1and 3. There was no decrease in the incidence of reportedback pain (36% for braced individuals and 34% for non-braced) or in the number of sick days lost when compar-ing braced with nonbraced workers; however, a subgroupanalysis of workers with prior back pain revealed that theuse of a soft lumbar brace reduced the number of days lostdue to back pain from 6.5 to 1.2 days per month (p =0.03). It should be noted that only 43% of the workerscomplied with the bracing protocol. There was no differ-ence in the incidence of low-back pain or number of sickdays lost between compliers and noncompliers in thebrace-treated group. The authors concluded that bracetherapy does not diminish the incidence of low-back painor time lost from work when used as a preventive strate-gy. The use of a lumbar support by workers with a previ-ous low-back injury may result in fewer days lost due tolow-back pain. Because of the high number of noncom-pliant workers, this study is considered to provide ClassIII medical evidence.

Reddell and colleagues33 randomized 642 individualsemployed as baggage handlers into one of four groups: 1)education; 2) weight belt–type brace; 3) education andbrace; and 4) no intervention. During an 8-month period,the authors examined the total incidence of reported low-back injury, lost or restricted workdays due to low-back

pain, and Workers’ Compensation claims related to low-back pain. They found no differences among the groupswith respect to these outcome measures. Similar to the vanPoppel study, only 42% of the brace-treated group wascompliant with the use of the brace. The noncompliantgroup (158 individuals) was followed and found to have ahigher incidence of lost work days, but the differencebetween the compliant and noncompliant groups was notsignicant. This study also provides Class III medical evi-dence suggesting no benefit for the use of a lumbar ortho-sis to prevent back injury. Alexander, et al.,1 reported theresults of a prospective study of 60 healthcare workersdivided into two groups. One group was assigned to weara beltlike corset for a 3-month period. No differences inwork-related back injuries or perception of back pain werenoted. Because a nonvalidated outcome measure wasused, this paper also provides Class III evidence suggest-ing that brace therapy involving a corset-type orthosis isnot an effective measure to prevent low-back pain.

Walsh and Schwartz45 randomly selected 90 individualsfrom a pool of 800 warehouse workers and assigned theminto one of three groups: 1) no intervention; 2) 1-hour edu-cation; or 3) 6-month lumbosacral molded semi-rigidorthosis therapy and education. Outcomes were assessedusing various measures including work injury incidence,work productivity, and utilization of healthcare resources.Brace-treated workers missed 2.5 days less work (p =0.03) than those not wearing braces (both controls andeducation-only groups), but there were no differences inproductivity or utilization of healthcare resources betweenthe groups. A subgroup analysis revealed that the benefitin terms of number of lost work days was greatest in pa-tients with a previous back injury. The authors concludedthat the combination of brace therapy and education waseffective in reducing time lost to injury, especially amongpatients with a history of back injury. Because there wasno information given regarding worker compliance withthe bracing routine, the medical evidence offered in thisreport in support of brace therapy as a measure to preventlow-back pain is considered Class III.

Several historical cohort studies have examined the in-cidence of back pain and days lost to work in groups ofworkers before and after being issued a brace or lumbarsupport belt by their employer. Analysis of these studiesrevealed mixed results. One study identified no change inthe incidence of back pain and sick days after applicationof the brace and two studies reported a reduction in theseparameters following the issue of a lumbar support toemployees.21,27,38 Overall, the medical evidence supportingthe use of braces as a preventive measure for low-backpain is inconsistent. The authors of several systematic lit-erature reviews have concluded that lumbar support de-vices are not useful for the prevention of low-back pain inthe general working population.18,44 It does appear, howev-er, that braces may be useful as a measure to decrease thenumber of sick days lost due to low-back pain in workerswith a history of low-back injury.

In several randomized control trials the role of bracing asa treatment for low-back pain has been examined. Pope, etal.,32 studied 164 patients with low-back pain drawn from achiropractic clinic. Patients were randomized to one of fourtreatments: 1) chiropractic manipulation; 2) TMS; 3) mas-sage; and 4) lumbar corset. Patients were assessed for pain


Brace therapy in lumbar degenerative disease

717




TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

the

use

of

prev

enta

tive

bra

ce t

hera

py*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Wal

sh &

Sch

war

tz, 1

990

II90

indi

vidu

als

rand

omly

sel

ecte

d fr

om 8

00w

areh

ouse

wor

kers

. 3 g

roup

s w

/o in

terv

entio

n, 1

hr

of e

duca

tion,

or

Wor

kers

in G

roup

3 m

isse

d 2.

5 da

ys le

ss w

ork

in th

e

6 m

os o

f L

SO 1

educ

atio

n. O

utco

me

asse

ssed

as

1) a

bdom

inal

str

engt

h; 2

) ed

ucat

iona

l ass

essm

ent;

3) w

ork

6-m

o tr

ial.

No

diff

eren

ces

in p

rodu

ctiv

ity. H

igh-

inju

ry in

cide

nce;

4)

prod

uctiv

ity; &

5)

use

of h

ealth

care

res

ourc

es. .

90%

FU

. No

info

rmat

ion

on b

race

com

-ri

sk in

divi

dual

s (

thos

e w

/ pre

viou

s ba

ck c

om-

plia

nce

or s

ubje

ct s

elec

tion.

pl

aint

s) h

ad a

gre

ater

eff

ect,

w/ 5

.9 f

ewer

lost

w

ork

days

in G

roup

3 &

2 in

Gro

up 2

.R

edde

ll, e

t al.,

199

2II

642

bagg

age

hand

lers

ran

dom

ized

to 1

of

4 gr

oups

: 1)

educ

atio

n; 2

) br

acin

g; 3

) ed

ucat

ion

1br

acin

g; &

4)

noth

ing.

A

utho

rs r

ecom

men

ded

that

bac

k br

aces

not

be

rou-

58

% o

f gr

oup

had

prev

ious

LB

I &

26%

had

a s

peci

fic

LB

I. L

ooke

d at

tota

l inc

iden

ce o

f in

jury

, los

t or

rest

rict

-tin

ely

used

for

the

prev

entio

n of

LB

I in

this

pop

u-

ed w

orkd

ays,

& W

orke

rs’C

ompe

nsat

ion

rate

s ov

er 8

-mo

peri

od. F

ound

no

diff

eren

ces

amon

g gr

oups

w/ r

es-

latio

n. D

ropo

ut g

roup

had

hig

her

inci

denc

e of

in-

pect

to th

ese

outc

omes

. Com

plia

nce

rate

was

onl

y 42

%.

jury

than

con

trol

or

educ

atio

n gr

oups

(da

ta n

ot in

-cl

uded

). E

ven

thou

gh n

o be

nefi

t dem

onst

rate

d, a

l-m

ost 7

0% o

f th

e pa

rtic

ipan

ts f

ound

a b

race

hel

pful

.T

hom

pson

, et a

l., 1

994

IIT

he p

rim

ary

aim

of

the

stud

y w

as a

pro

spec

tive

stud

y of

60

heal

thca

re tr

ansp

ort w

orke

rs w

ho w

ere

divi

ded

into

A

2-pa

rt s

tudy

w/ a

n in

itial

sur

vey

of a

ttitu

des

in 1

45

2 gr

oups

w/ 4

1 br

aced

(1

back

sch

ool i

nter

vent

ion)

& 1

9 no

nbra

ced

for

3 m

os. A

ttitu

des

impr

oved

w/ b

elt

wor

kers

who

wer

e gi

ven

a ba

ck b

elt.

The

se w

ork-

avai

labi

lity

& f

requ

ency

of

back

pai

n de

crea

sed

in th

e be

lted

grou

p.er

s ha

d be

tter

attit

udes

& a

red

uctio

n in

bac

k in

-ju

ry to

0%

. No

data

are

pro

vide

d fo

r th

e ra

tes

of

back

pai

n in

the

pros

pect

ive

grou

p &

no

desc

rip-

tion

is p

rovi

ded

on h

ow th

e au

thor

s m

easu

red

atti-

tude

.A

lexa

nder

, et a

l., 1

995

II60

hea

lthca

re w

orke

rs w

ere

divi

ded

into

2 g

roup

s, 3

0 w

/ cor

set w

ho w

ore

brac

e du

ring

wor

k fo

r 3

mos

. No

dif-

No

men

tion

of c

ompl

ianc

e w

/ the

bra

cing

. Doe

s th

e fe

renc

es in

wor

k-re

late

d ba

ck in

juri

es o

r su

bjec

tive

perc

eptio

n of

bac

k pa

in; h

owev

er, 7

0% o

f th

e co

rset

gro

up

brac

e m

ake

the

empl

oyee

fee

l “ov

erco

nfid

ent?

” fe

lt th

at th

e be

lt ai

ded

in a

void

ing

inju

ry &

29

of 3

0 sa

id th

e be

lt m

ade

them

“fe

el g

ood.

”N

o be

nefi

t of

brac

ing.

Kra

us, e

t al.,

199

6II

His

tori

cal c

ohor

t stu

dy e

xam

inin

g in

cide

nce

of L

BP

in w

orke

rs a

t 1 c

ompa

ny p

rior

to in

stitu

tion

of a

bac

k-su

p-A

utho

rs c

oncl

uded

that

lum

bar

supp

orts

wer

e ef

fec-

po

rt-u

se p

olic

y &

aft

er th

at p

olic

y w

as in

stitu

ted.

36,

000

empl

oyee

s pa

rtic

ipat

ed. A

fter

the

polic

y w

as in

stitu

ted

tive

in r

educ

ing

the

inci

denc

e of

bac

k in

jury

in

ther

e w

as a

34%

red

uctio

n in

rat

e of

LB

P/m

illio

n hr

s w

orke

d. T

his

effe

ct w

as n

oted

in b

oth

sexe

s, a

cros

s th

is p

opul

atio

n of

wor

kers

.al

l age

gro

ups

& th

roug

hout

all

type

s of

jobs

.va

n Po

ppel

, et a

l., 1

998

II28

2 (o

f 31

2) D

utch

bag

gage

han

dler

sra

ndom

ized

into

1 o

f 4

grou

ps: 1

) ed

ucat

ion

& lu

mba

r su

ppor

t; 2)

edu

ca-

Lum

bar

supp

orts

do

not r

educ

e L

BP

inci

denc

e or

tio

n; 3

) lu

mba

r su

ppor

t; &

4)

no in

terv

entio

n. L

umba

r su

ppor

ts u

sed

duri

ng w

ork

hour

s fo

r a

6-m

o pe

riod

. si

ckle

ave

whe

n us

ed a

s a

prev

enta

tive

stra

tegy

for

Onl

y a

43%

com

plia

nce

rate

w/ s

oft b

race

. No

decr

ease

(36

vs

34%

) in

inci

denc

e of

bac

k pa

in in

gro

ups

w/ o

r L

BP.

Unl

ike

prev

ious

stu

dies

ther

e w

as n

o in

crea

sew

/o b

raci

ng. N

o de

crea

se in

sic

k le

ave.

In

1 su

bgro

up o

f pa

tient

s (t

hose

w/ p

revi

ous

back

pai

n) b

raci

ng r

educ

ed

in th

e in

cide

nce

of L

BP

in th

e gr

oups

dis

cont

inui

ng

days

lost

to b

ack

pain

fro

m 6

.5 to

1.2

.us

e of

the

belt.

*FU

= f

ollo

w u

p; L

BI

= lo

w-b

ack

inju

ry; L

BP

= lo

w-b

ack

pain

; LSO

= lu

mbo

sacr

al o

rtho

sis.




719

TAB

LE

2Su

mm

ary

of s

tudi

es i

nvol

ving

the

tre

atm

ent

of l

ow-b

ack

pain

and

bra

ce t

hera

py*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Cox

head

, et a

l., 1

981

II32

2 pa

tient

s ra

ndom

ized

to tr

actio

n, e

xerc

ises

, man

ipul

atio

n, &

cor

set.

Patie

nts

had

scia

tic p

ain

w/ o

r w

/o b

ack

No

sign

ific

ant d

iffe

renc

es in

any

gro

up a

t 4 &

16

pa

in. F

acto

rial

stu

dy d

esig

n so

that

16

grou

ps w

ere

pres

ent i

n to

tal.

Tre

atm

ent l

aste

d fo

r 4

wks

& o

utco

me

was

mos

. Man

ipul

atio

n pa

tient

s ha

d gr

eate

r im

prov

e-as

sess

ed a

t 1, 4

, & 1

6 m

os. O

utco

me

mea

sure

s at

1 m

o w

ere

impr

ovem

ent o

n V

AS.

At a

ll tim

e po

ints

pat

ient

sm

ent i

n pa

in. A

ctiv

e ph

ysio

ther

apy

usef

ul in

the

give

n a

bette

r, sa

me,

wor

se s

atis

fact

ion

ques

tionn

aire

. Als

o R

TW

S as

sess

ed a

t 1 &

4 m

os. 9

1% F

U a

t 1 m

o sh

ort t

erm

.&

80%

at 1

6 m

os. P

atie

nts

rece

ivin

g m

anip

ulat

ion

had

mor

e V

AS

impr

ovem

ent a

t 4 w

ks. T

here

is a

tren

d to

-w

ard

mor

e su

bjec

tive

impr

ovem

ent i

n pa

tient

s re

ceiv

ing

mor

e tr

eatm

ents

.M

illio

n, e

t al.,

198

1II

19 p

atie

nts

w/ c

hron

ic L

BP

wer

e ra

ndom

ized

to r

ecei

ve e

ither

a s

oft c

orse

t or

a co

rset

w/ a

n in

set.

Subj

ectiv

e &

A

utho

rs c

oncl

uded

that

the

bene

fit o

f a

lum

bar

sup-

ob

ject

ive

outc

omes

wer

e co

mpa

red

at 4

& 8

wks

. The

sub

ject

ive

was

a 1

5-ite

m q

uest

ionn

aire

that

look

ed a

t po

rt d

oes

not o

ccur

bas

ed o

n in

crea

se in

intr

aab-

pain

& li

mita

tion

in f

unct

ion

as a

nsw

ered

by

a V

AS

(Mill

ion

Scal

e). T

he o

bjec

tive

mea

sure

men

ts w

ere

RO

Mdo

min

al p

ress

ure

as e

vide

nced

by

the

lack

of

im-

& S

LR

. The

re w

ere

no in

terg

roup

dif

fere

nces

w/ r

egar

d to

obj

ectiv

e cr

iteri

a, b

ut th

ere

was

an

impr

ovem

ent i

n pr

ovem

ent i

n th

e gr

oup

w/ t

he s

oft b

inde

r. T

he

pain

& f

unct

ion

as a

sses

sed

by th

e M

illio

n Sc

ale.

stud

y ha

s a

smal

l sam

ple

size

& s

hort

per

iod

of

use

for

the

brac

e. O

nly

cont

rol g

roup

giv

en a

sof

t bi

nder

.W

illne

r, 19

85II

I48

pat

ient

s w

/ LB

Pfr

om 3

con

ditio

ns. P

atie

nts

retu

rned

at a

mea

n of

1 y

r &

wer

e as

ked

if s

ympt

oms

wer

e un

-N

o ou

tcom

e in

stru

men

t use

d. D

iffi

cult

to d

eter

min

ech

ange

d, im

prov

ed, o

r re

solv

ed. P

atie

nts

w/ s

pond

ylo

fare

d th

e be

st w

/ 13/

15 e

xper

ienc

ing

com

plet

e re

lief

why

pat

ient

s w

ere

even

pre

scri

bed

brac

es.

& th

e ot

her

two

sign

ific

ant i

mpr

ovem

ent i

n th

e pa

in. S

pina

l ste

nosi

s pa

tient

s ha

d 2/

7 w

/ com

plet

e re

lief

& 4

/7

w/ s

ome

relie

f. I

n ca

ses

of L

BP

of u

nkno

wn

orig

in 1

7/26

had

no

relie

f of

bac

k pa

in.

Hse

ih, e

t al.,

199

2I

Suba

cute

LB

P(3

–24-

wk)

pat

ient

s w

ere

rand

omiz

ed to

man

ipul

atio

n, m

assa

ge, c

orse

t, or

TM

S. O

utco

me

asse

ssed

A

utho

rs f

ound

bot

h sc

ales

rel

iabl

e fo

r m

easu

ring

by

OD

I &

RM

Q. 6

3 pa

tient

s en

rolle

d in

3-w

k st

udy.

The

man

ipul

atio

n gr

oup

perf

orm

ed b

ette

r th

an m

assa

ge

LB

P. M

anip

ulat

ion

fare

d th

e be

st in

the

shor

t-te

rm

& T

MS

grou

ps, &

cor

set b

ette

r th

an m

assa

ge.

FU.

Val

le-J

ones

, et a

l., 1

992

I21

6 pa

tient

w/ L

BP

of a

ll du

ratio

n w

/o a

ny b

one

inju

ry o

r an

y in

terv

erte

bral

dis

c pa

thol

ogy.

Ran

dom

ized

to lu

m-

Sign

ific

antly

mor

e im

prov

emen

t in

pain

at r

est,

acti-

ba

r br

ace

or a

ctiv

ity m

odif

icat

ion.

Pat

ient

s th

en m

easu

red

pain

& d

isab

ility

on

a V

AS

& p

arac

etam

ol in

take

. vi

ty, &

nig

ht in

bra

cing

gro

up a

fter

Day

7. F

ewer

A

lso

com

pete

d a

subj

ectiv

e as

sess

men

t of

thei

r co

nditi

on. S

tudy

dur

atio

n w

as 3

wks

. 100

% o

f pa

tient

s co

m-

anal

gesi

cs in

bra

ce g

roup

. Bot

h gr

oups

impr

oved

plet

ed tr

ial.

over

tim

e; th

e br

ace

grou

p im

prov

ed f

aste

r.Po

pe, e

t al.,

199

4I

164

patie

nts

w/ s

ubac

ute

LB

Pra

ndom

ized

to 1

of

4 tr

eatm

ents

: 1)

chir

opra

ctic

man

ipul

atio

n; 2

) T

MS;

3)

mas

sage

; N

o di

ffer

ence

s bt

wn

grou

ps. A

ll pa

tient

s en

rolle

d in

-&

4)

cors

et. 2

:1 r

ando

miz

atio

n w

/ 70

patie

nts

in G

roup

1 &

31

in e

ach

of th

e ot

her

4. A

sses

sed

as V

AS

for

to th

e st

udy

wer

e dr

awn

from

a c

hiro

prac

tic c

linic

.pa

in &

fun

ctio

nal R

OM

w/ n

o di

ffer

ence

s am

ong

the

grou

ps a

t 3 w

ks. C

ompl

ianc

e ra

te w

as 8

8% &

ther

e w

ere

no in

terg

roup

dif

fere

nces

in p

ain

or f

unct

ion.

Jelle

ma,

et a

l., 2

002

IIO

bser

vatio

nal s

tudy

exa

min

ing

usag

e ra

tes

& im

prov

emen

t in

sym

ptom

s fo

r ho

me

heal

th w

orke

rs w

/ LB

P. S

emi-

44%

dec

reas

e in

bac

k pa

in. I

ndiv

idua

ls r

ated

ben

efit

rigi

d su

ppor

t wor

n fo

r 6

mos

. Mon

thly

ass

essm

ents

of

pain

(V

AS)

, dis

abili

ty (

Que

bec

Bac

k Pa

in D

isab

ility

as

7/1

0.Sc

ale)

, & s

ubje

ctiv

e be

nefi

t on

a 10

-poi

nt s

cale

. 80%

of

patie

nts

com

plia

nt w

/ the

bra

ce.

*IV

DD

= in

terv

erte

bral

dis

c di

seas

e; R

MQ

= R

olan

d–M

orri

s Q

uest

ionn

aire

; RT

WS

= r

etur

n-to

-wor

k st

atus

; SL

R =

str

aigh

t leg

rai

sing

; spo

ndyl

o =

spo

ndyl

olis

thes

is.




TAB

LE

3Su

mm

ary

of s

tudi

es i

nvol

ving

the

use

of

TE

PF

*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Ole

rud,

et a

l., 1

986

III

Rep

orte

d on

the

use

of th

e T

EPF

in 1

8 pa

tient

s w

/ LB

P. A

n im

prov

emen

t was

not

ed w

/ VA

S, p

ain

diag

ram

s, &

T

he a

utho

rs n

oted

it w

as a

use

ful t

est i

npa

tient

s w

/ a

mob

ility

w/ t

he f

ixat

ion

in p

lace

, but

no

corr

elat

ion

was

mad

e w

/ sub

sequ

ent f

usio

n.

prev

ious

fus

ion

who

stil

l had

pai

n. A

ll 8

patie

nts

in th

is c

ateg

ory

impr

oved

w/ e

xter

nal f

ixat

ion.

Ess

es, e

t al.,

198

9II

I35

pat

ient

s w

/ var

ying

dia

gnos

es f

or L

BP

had

exte

rnal

fix

atio

n fo

r 1–

2 w

ks. T

hey

also

had

a p

erio

d of

tim

e w

/ N

o da

ta g

iven

on

degr

ee o

f pa

in r

elie

f or

who

was

th

e br

ace

loos

ened

(a

“pla

cebo

tria

l”).

27/

35 s

ubje

ctiv

ely

impr

oved

with

bra

cing

& th

ey w

ere

offe

red

op w

/ fu-

sele

cted

for

ext

erna

l fix

atio

n. T

he a

utho

rs a

lso

try

sion

per

form

ed u

sing

a v

arie

ty o

f te

chni

ques

. Som

e FU

is a

vaila

ble

for

23 o

f th

e pa

tient

s w

/ 6 p

atie

nts

repo

rtin

g to

com

pare

ext

erna

l fix

atio

n w

/ rad

iogr

aphi

c ch

ang-

com

plet

e re

lief,

11 s

igni

fica

nt r

elie

f, 5

no c

hang

e, &

1 w

as w

orse

. Fis

her

exac

t tes

t: th

ere

is a

str

ong

asso

ciat

ion

es, d

isco

grap

hy, &

fac

et b

lock

s, d

raw

ing

the

con-

btw

n pa

in r

elie

f w

/ ext

erna

l fix

atio

n &

pos

itive

out

com

e of

sub

sequ

ent f

usio

n.cl

usio

n th

at e

xter

nal f

ixat

ion

is m

ore

relia

ble

than

th

ese

stud

ies.

Ord

eber

g, e

t al.,

199

3II

IFU

stu

dy o

f pa

tient

s fr

om th

e or

igin

al s

erie

s by

Ole

rud;

TE

PF h

ad b

een

pla

ced

for

the

trea

tmen

t of

LB

P. 9

3 liv

ing

Lon

g-te

rm p

rogn

ostic

val

ue o

f th

e te

st n

ot a

s go

od a

s pa

tient

s w

ere

iden

tifie

d &

63

that

wer

e si

gnif

ican

tly im

prov

ed w

ere

avai

labl

e fo

r re

view

by

an in

depe

nden

t re-

the

shor

t-te

rm r

esul

ts. I

n th

e pa

per

it is

unc

lear

why

view

er. O

nly

35 (

56%

) st

ill f

elt t

hey

wer

e im

prov

ed w

/ 13

unch

ange

d &

15

wor

se.

only

63

patie

nts

wer

e av

aila

ble

for

FU &

the

mea

-su

re o

f pa

in w

as s

ubje

ctiv

e th

roug

hout

the

stud

y.So

ini,

et a

l., 1

993

III

42 p

atie

nts

w/ v

ario

us d

iagn

oses

und

erw

ent T

EPF

. Pai

n w

as f

ollo

wed

usi

ng V

AS

& f

unct

ion

usin

g O

DI.

29

pa-

Ove

rall

fusi

on r

ate

was

83%

w/ g

reat

er im

prov

emen

t tie

nts

had

a m

arke

d re

lief

of p

ain

& h

ad A

LIF

w/ t

he e

xter

nal f

ram

e le

ft in

pla

ce f

or a

mea

n of

14

wks

. A

tin

bac

k &

leg

pain

as

wel

l as

in O

DI

in th

e pa

-2

yrs,

22

wer

e av

aila

ble

for

FU &

had

a s

tatis

tical

ly s

igni

fica

nt im

prov

emen

t in

back

(62

vs

30),

leg

(56

vs

tient

s w

/ a s

olid

fus

ion.

29),

& O

DI

(47

vs 2

8). S

ubje

ctiv

ely,

11

wer

e pa

in f

ree

or s

igni

fica

ntly

impr

oved

, 4 h

ad s

ome

impr

ovem

ent,

4 w

ere

unch

ange

d, &

3 w

ere

wor

se.

Soin

i & S

eits

alo,

199

3II

I25

/100

pat

ient

s ha

d 30

com

plic

atio

ns w

/ 18%

hav

ing

pin

trac

t inf

ectio

ns.

Mus

t car

eful

ly c

onsi

der

risk

–ben

efits

of

the

test

.Je

anne

ret,

et a

l., 1

994

II10

1 pa

tient

s w

/ dis

ablin

g L

BP

unde

rwen

t TE

PF. I

n 47

the

pain

was

rel

ieve

d &

ret

urne

d af

ter

loos

enin

g of

the

The

deg

ree

of p

ain

relie

f w

as s

ubje

ctiv

ely

judg

ed &

ex

tern

al f

ixat

or, i

n 2

the

pain

rem

aine

d ab

sent

, & in

52

the

pain

did

not

cha

nge

w/ e

xter

nal f

ixat

ion.

42

(34

th

e co

mpl

icat

ion

rate

was

17%

. Als

o th

e se

lect

ion

w/ a

pos

tive

resu

lt fr

om e

xter

nal f

ixat

ion

& 8

w/o

rel

ief)

wen

t on

to f

usio

n vi

a se

vera

l tec

hniq

ues

& h

ad 2

-yr

of n

onre

spon

ders

for

fus

ion

is n

ot c

lear

. If

good

FU

that

con

firm

ed f

usio

n. O

utco

mes

wer

e ra

ted

good

if p

ain

was

wel

l con

trol

led,

fai

r if

pai

n ha

d de

crea

sed

& f

air

outc

omes

are

see

n as

a p

ositi

ve r

espo

nse

but w

as s

till a

pro

blem

, & p

oor

if p

ain

was

wor

se o

r un

chan

ged.

In

patie

nts

who

had

pai

n re

lief

w/ e

xter

nal

to f

usio

n, e

xter

nal f

ixat

ion

has

a se

nsiti

vity

of

fi

xatio

n, 1

4 (4

1%)

patie

nts

wer

e go

od, 1

2 (3

5.5%

) w

ere

fair

, & 8

wer

e po

or (

23.5

%).

Of

the

8 w

/ no

relie

f on

96

%, s

peci

fici

ty o

f 47

%, P

PVof

76%

, & a

n N

PVex

tern

al f

ixat

ion,

7 w

ere

poor

, & 1

was

goo

d.

of 8

8%. I

f on

ly th

e go

od r

espo

nses

are

judg

ed a

po

sitiv

e ou

tcom

e, th

e te

st h

as a

sen

sitiv

ity 9

3%,

spec

ific

ity o

f 26

%, P

PVof

41%

, & a

n N

PVof

87%

. B

edna

r &R

aduc

an, 1

996

III

61 p

atie

nts

cons

ider

ed f

or lu

mba

r fu

sion

bas

ed o

n po

sitiv

e di

scog

raph

y or

fac

et b

lock

s, r

ando

miz

ed to

fus

ion

(31)

St

atis

tical

ly s

igni

fica

nt r

educ

tion

in p

ain

scor

e &

an-

or

TE

PF (

30)

follo

wed

by

fusi

on if

the

fixa

tor

relie

ved

the

pain

. 19/

31 (

61%

) ha

d si

gnif

ican

t rel

ief

of p

ain

fol-

alge

sic

use

in th

e st

udy

grou

p. S

tudy

was

des

igne

dlo

win

g fu

sion

in G

roup

I &

19/

21 (

90%

) of

Gro

up I

I ha

d si

gnif

ican

t pai

n re

lief.

The

oth

er 9

in G

roup

II

had

to d

eter

min

e th

e ef

fica

cy o

f ex

tern

al f

ixat

ion

inno

pai

n re

lief

& w

ere

excl

uded

fro

m f

urth

er c

onsi

dera

tion

for

fusi

on.

conj

unct

ion

w/ o

ther

dia

gnos

otic

test

s.va

n de

r Sc

haaf

, et a

l., 1

999

III

133

patie

nts

w/ 1

yr

of in

capa

cita

ting

LB

Pof

no

clea

r ca

use.

VA

S &

wor

king

cap

acity

wer

e as

sess

ed b

efor

e &

Pa

tient

sat

isfa

ctio

n in

the

fusi

on g

roup

was

hig

h. A

u-fo

r 7–

10 d

ays

duri

ng T

EPF

. Pat

ient

s w

/ a s

igni

fica

nt r

elie

f of

pai

n fo

llow

ing

fixa

tion

& r

etur

n of

pai

n w

/ the

th

ors

conc

lude

d th

at w

hen

pain

is r

elie

ved

w/ e

x-ri

gid

fram

e lo

osen

ed w

ere

then

off

ered

fus

ion.

55

patie

nts

had

fusi

on &

at 2

2-m

o FU

had

a c

ontin

ued

redu

ctio

n te

rnal

fix

atio

n, th

e su

bseq

uent

fus

ion

over

the

in p

ain

(77

vs 4

2) th

at w

as s

igni

fica

nt c

ompa

red

w/ b

asel

ine

& th

e co

ntro

l non

fusi

on g

roup

.te

sted

leve

ls w

ill li

kely

be

succ

essf

ul in

rel

ievi

ngpa

in. T

here

was

a 9

.8%

infe

ctio

n ra

te &

a 9

%

root

irri

tatio

n ra

te.

Bed

nar,

2001

III

100

patie

nts

unde

rwen

t TE

PF f

or 1

–2 w

ks w

/ VA

S &

Pro

lo s

core

rec

orde

d pr

ior

to &

at t

he c

ompl

etio

n of

tria

l. Po

oled

dat

a fr

om 5

oth

er s

erie

s &

fou

nd d

efin

itive

60

had

ade

quat

e pa

in r

elie

f &

wer

e of

fere

d fu

sion

. 49

had

PLF

w/ i

nstr

umen

tatio

n. 5

5% h

ad im

prov

emen

t in

long

-ter

m r

elie

f in

onl

y 36

% o

f pa

tient

s (1

73/4

82).

th

eir

Prol

o sc

ore,

37%

no

chan

ge, &

8%

wer

e w

orse

. T

hese

incl

ude

all p

atie

nts

subj

ecte

d to

ext

erna

l fi

xatio

n, n

ot ju

st “

resp

onde

rs.”

Axe

lsso

n, e

t al.,

200

3II

IO

bser

vatio

nal s

tudy

of

26 p

atie

nts

w/ L

BP

who

had

TE

PF p

lace

d fo

r 8

days

. Pat

ient

s ha

d a

vari

ety

of d

iagn

oses

D

iffi

cult

to d

eter

min

e w

hich

pat

ient

s ha

d a

“pos

itive

&

wer

e se

lect

ed f

or f

usio

n ba

sed

on a

dec

reas

e in

pai

n in

5 A

DL

& 2

fun

ctio

nal w

alki

ng te

sts.

20

patie

nts

resp

onse

” to

ext

erna

l fix

atio

n. T

he o

utco

me

crite

ria

wen

t on

to f

usio

n &

14

had

good

, 4 f

air,

& 2

poo

r ou

tcom

es. T

he c

oncl

usio

n w

as th

at a

tria

l of

exte

rnal

fix

atio

n w

ere

a pa

tient

-bas

ed s

atis

fact

ion

surv

ey. W

ide

led

to im

prov

ed s

elec

tion

of p

atie

nts

who

had

fus

ion

for

LB

Pba

sed

on th

e fa

ct th

at 7

0% o

f pa

tient

s ha

d a

good

va

riet

y o

f di

agno

ses

wer

e in

the

stud

y w

hich

ou

tcom

e.m

akes

it im

poss

ible

to s

ubdi

vide

bas

ed o

n ca

use.

*A

DL

= a

ctiv

ities

of

daily

livi

ng; A

LIF

= a

nter

ior

lum

bar

inte

rbod

y fu

sion

.


using a VAS and were also assessed for ROM after 3 weeksof treatment. There were no differences among the groups.Because of the relatively small treatment groups (~ 30patients in three of the four groups) and selected patientpopulation (from a chiropractic practice), this paper is con-sidered to provide Class II medical evidence suggestingthat braces are not an effective treatment for low-back pain.Valle-Jones and colleagues40 randomized 216 patients withnonspecific low-back pain of varying duration to lumbarbrace therapy or activity modification for 3 weeks. Out-come measures included a VAS score for pain and disabil-ity. Patients were also asked to record usage of pain med-ication. Brace-treated patients were found to have moreimprovement in pain at rest, pain with activity, and pain atnight between Days 7 and 21. In addition, brace-treatedpatients took 50% the analgesics of those in the activitymodification group over the 3-week period. This paper isconsidered to provide Class I medical evidence supportingthe efficacy of braces for the short-term amelioration oflow-back pain. With regard to the chronic low-back painpopulation, the evidence provided by this report would beconsidered Class III because of the inclusion of manypatients with relatively acute symptoms. Hseih and associ-ates15 studied 63 patients with low-back pain of less than 6months’ duration. Patients were randomized to either ma-nipulation, massage, lumbar corset, or TMS treatment for3 weeks. Functional outcomes were assessed by ODI andRoland–Morris questionnaires. Patients treated with cor-sets had significantly better outcomes than those in themassage group on both scales, but they were not function-ally improved compared with those in the manipulation orTMS groups. This paper provides Class I medical evidencesupporting the role of short-term lumbar brace therapy inpatients with acute or subacute low-back pain. No infer-ences can be drawn regarding the effect of braces amongthe chronic low-back pain population.

Two randomized controlled studies published in 1981provide information on lumbar brace therapy for low-backpain. Coxhead and coworkers10 performed a randomizedstudy of 322 sciatica patients with or without low-backpain randomized to different treatment modalities includ-ing traction, exercises, manipulation, corset brace, andcombinations of these treatments for a total of 16 treatmentgroups. Treatments lasted for 4 weeks and outcome wasassessed at 1, 4, and 16 months by VAS, return-to-workstatus, and patient satisfaction criteria. No benefit, short orlong term, was detected for the use of lumbar corset braces.Because the population was composed of patients with sci-atica, the medical evidence in this report against the effica-cy of brace therapy for low-back pain is considered ClassIII. In a smaller study of 19 patients with chronic low-backpain, Million, et al.,25 randomized patients to either a soft or rigid lumbar brace group for 4 weeks. A 15-item ques-tionnaire about pain and functional limitation on a VAS(Million Scale) demonstrated a significant improvement (p , 0.01) for patients wearing a rigid brace at 4 and 8weeks. Rigid lumbar bracing may therefore have someshort-term benefit compared with soft bracing for theshort-term treatment of low-back pain. Because there wasno control group in this study, the paper is considered toprovide Class III medical evidence regarding the efficacyof brace therapy for low-back pain.

One study reported in the literature specifically exam-



721

TAB

LE

4Su

mm

ary

of s

tudi

es i

nvol

ving

bra

ce t

hera

py a

nd l

umba

r fu

sion

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

John

sson

, et a

l., 1

992

I &

III

*C

ompa

riso

n of

11

patie

nts

w/ r

igid

ext

erna

l ort

hosi

s fo

r 5

mos

(G

roup

I)

& 1

1 w

/ ort

hosi

s fo

r 3

mos

(G

roup

II)

. M

ovem

ent w

as a

sses

sed

by s

agitt

al, v

ertic

al, &

tran

s-A

ll pa

tient

s ha

d a

PLF

for

Gra

de I

or

II s

pond

ylo

& ta

ntal

um m

arke

rs w

ere

plac

ed a

t the

tim

e. P

atie

nts

wer

eve

rse

tran

slat

ion.

The

mot

ion

subs

ided

btw

n th

e 3

follo

wed

w/ R

SAat

var

ious

tim

e po

ints

up

to 1

yr.

In G

roup

I, 8

/11

wer

e fu

sed

base

d on

no

tran

slat

ion

on

& 6

-mo

exam

s.R

SAco

mpa

red

to 2

/11

in G

roup

II.

A

xels

son,

et a

l., 1

995

III

All

patie

nts

who

wer

e to

und

ergo

a f

usio

n fo

r L

BP

had

a tr

ial o

f ri

gid

or s

emir

igid

bra

ce th

erap

y fo

r at

leas

t 3 w

ks.

The

out

com

e m

easu

re u

sed

is n

onva

lidat

ed. T

he p

op-

Pain

impr

ovem

ent w

as r

ecor

ded

& a

sig

nifi

cant

pai

n re

spon

se w

as ju

dged

to b

e an

impr

ovem

ent o

f $

50%

. ul

atio

n st

udie

d is

sel

ecte

d. T

he a

utho

rs c

ompa

red

Patie

nts

w/ a

sol

id r

adio

grap

hic

fusi

on a

t 1 y

r w

ere

incl

uded

in th

e st

udy

(50

patie

nts)

& th

en a

t 2 y

rs ju

dged

th

e pe

rcen

tage

s of

fav

orab

le r

espo

nses

w/ b

raci

ng

for

pain

rel

ief

w/ p

ain

free

, sig

nifi

cant

impr

ovem

ent,

slig

ht im

prov

emen

t, un

chan

ged,

or

wor

se &

as

satis

fied

&

w/ o

p us

ing

a ch

i-sq

uare

ana

lysi

s &

fou

nd n

oor

uns

atis

fied

. 31

patie

nts

wer

e im

prov

ed w

/ bra

cing

& 2

0 of

thes

e ha

d a

good

out

com

e at

2 y

rs, w

hile

11

did

corr

elat

ion

of r

espo

nse

to p

reop

bra

ce &

pai

n no

t. 19

pat

ient

s di

d no

t hav

e si

gnif

ican

t rel

ief

w/ b

raci

ng &

13

of th

ese

had

a fa

vora

ble

outc

ome

at 2

yrs

whi

le

relie

f af

ter

solid

fus

ion.

6 di

d no

t. If

bra

cing

is u

sed

as a

pre

op te

st f

or s

ucce

ss a

fter

fus

ion,

the

sens

itivi

ty is

61%

, spe

cifi

city

is 3

5%,

PPV

is 6

5%, &

the

NPV

is 3

2%.

*C

lass

I e

vide

nce

for

resu

lts o

f 3-

ver

sus

5-m

onth

imm

obili

zatio

n; C

lass

III

evi

denc

e fo

r re

sults

of

brac

e ve

rsus

no

brac

e th

erap

y.


ined the use of preoperative brace therapy as a predictivetest for outcome following lumbar fusion. Axelsson, etal.,6 placed all patients who were to undergo a lumbarfusion for low-back pain in either a rigid or a semirigidbrace for at least 3 weeks. Pain improvement was record-ed and a significant response was judged to be an im-provement in pain of at least 50%. Only 50 patients witha solid radiographic fusion at 1 year were included in thestudy. Two years following surgery, these same patientswere subjectively examined for pain relief and satisfac-tion. Thirty-one patients had improved pain scores and 20of these had a good outcome at 2 years (pain free or sig-nificant improvement), whereas 11 patients had poor out-comes despite a favorable response to preoperative lum-bar bracing. Nineteen patients did not have significantrelief and 13 of these reported a favorable outcome at 2years. If lumbar bracing is used as a preoperative “prog-nostic test” for success after solid fusion, the sensitivity is61%, the specificity is 35%, the PPV is 65%, and the NPVis 32%. Therefore, the use of lumbar bracing as a preop-erative adjunct to predict the outcome of fusion is not rec-ommended. Because of the reliance on patient satisfactionscores, the select population studied (only patients withsolid radiographic fusion), and the use of several types ofbraces in the preoperative period, the medical evidencederived from this study is considered Class III.

Although originally described as a technique for reduc-tion and fixation of thoracolumbar fractures,24 TEPF hasbeen used to predict the response to lumbar fusion in pa-tients with low-back pain.4,12,29–31,37 Several studies haveexamined the ability of TEPF to predict outcome follow-ing lumbar fusion. Axelsson, et al.,4–6 described their expe-rience in treating 26 patients with low-back pain (with avariety of diagnoses) who underwent TEPF for 8 days.Patients were selected for lumbar fusion if they reported adecrease in pain while performing five activities of dailyliving and demonstrated improvement in two functionalwalking tests. Twenty patients reported improvement andwere treated with lumbar fusion. Fourteen reported a goodoutcome, four reported a fair outcome, and two patientsreported their outcome as poor. The authors concludedthat 70% of their patients achieved good outcomes andtherefore that a trial of TEPF led to better selection ofpatients with low-back pain who might benefit from lum-bar fusion surgery.5 Because no comparison was madewith a control group (patients who did not have pain reliefwith TEPF or those who did not undergo TEPF) wasdescribed, no meaningful conclusions regarding the effi-cacy of TEPF in this population can be drawn.

Bednar7 noted that only 55% of 49 patients who experi-enced pain relief after TEPF and subsequent instrumenta-tion-based PLF had improvement in their Prolo Scalescore. Thirty-seven percent reported no change in theirProlo Scale scores following surgery and 8% were worse.Other authors have provided Class III medical evidencefor the use of TEPF as a preoperative test to predict out-come from lumbar fusion. In all of the studies cited, onlypatients with a positive response to TEPF were selectedfor lumbar fusion. The methods used to evaluate theresponse to TEPF and to evaluate the success of surgeryvary from study to study.12–14,30,34,36,37,39,41

One report includes outcome data obtained in patientswho were nonresponders to a trial of TEPF but who none-

theless underwent lumbar fusion surgery.16 One hundredone patients with disabling low-back pain underwentplacement of external fixation. Forty-seven patients hadrelief of pain due to TEPF that recurred after loosening ofthe external fixator device. Two patients had lasting reliefof their pain even when the fixator was removed. Fifty-two patients did not have pain relief with TEPF. Forty-twopatients underwent successful lumbar fusion (a variety oftechniques were used); 34 experienced pain relief withTEPF and eight underwent surgery despite an absent re-sponse to TEPF. Outcomes were determined using a pa-tient satisfaction scale. Of patients who had pain reliefwith external fixation, 14 (41%) reported good results, 12(35.5%) reported fair results, and eight (23.5%) reportedpoor results. Of the eight patients with no relief, sevenreported poor results and one reported a good result. Ifgood and fair outcomes are grouped as a positive res-ponse, TEPF-related pain relief has a 96% sensitivity,47% specificity, 76% PPV, and 88% NPV for predictingresponse to fusion. If only good outcomes are consideredas positive, the corresponding values are 93, 26, 41, and87%. The medical evidence provided by this report is con-sidered Class III because the outcome scale used is non-validated, the criteria used to determine the occurrence ofsuccessful fusion are unclear, and the indications for fu-sion among patients who did not respond to TEPF werenot described.

The complication rate of TEPF is not low. Soini andSeitsalo35 reported a 25% complication rate, includinginfection, nerve root irritation, and spinal fluid leakage. Inanother report of TEPF the authors cited infection rates of10% or greater and complication rates approaching 20%.14

Because of the significant complication rate and theuncertainty of TEPF in predicting a good outcome fol-lowing lumbar fusion, TEPF is not recommended as a rou-tine screening modality for the success of lumbar fusionsurgery in patients suffering from low-back pain.

Several authors have advocated the use of brace thera-py following lumbar fusion surgery;9,19 however, there areno published studies that compared outcomes followinglumbar fusion with and without the supplemental use of alumbar orthosis. Johnsson, et al.,20 have suggested a mini-mum 5-month period of bracing following noninstru-mented lumbar fusion. They noted that patients assessed6 months following surgery with RSA had a higher fusionrate (eight of 11 patients) at 1 year compared with 3months (two of 11). The authors found that sagittal andvertical translation decreased significantly as measuredby RSA between 3 and 6 months following surgery. Theyinterpreted this result as evidence that healing of a nonin-strumented lumbar fusion occurs over a 6-month timeperiod. They presented no evidence, however, regardingthe effect of lumbar bracing on the rate of lumbar spinalfusion or functional outcome.

Summary

Although conflicting reports have been presented in theliterature regarding the utility of lumbar braces for the pre-vention of low-back pain, most Class III medical evidencesuggests that these supports used prophylactically do notreduce the incidence of low-back pain or decrease the




amount of time lost from work in the general workingpopulation. Among workers with a history of a back in-jury, their use appears to decrease the number of workdays lost due to back pain. Lumbar braces appear to be aneffective treatment for acute low-back pain in some popu-lations. They do not appear to be effective in the chroniclow-back pain population. If a brace is used, rigid bracesoffer some benefit over soft braces.

There are no data to suggest that relief of low-back painwith preoperative external bracing predicts a favorableoutcome following lumbar spinal fusion. No informationis available on the benefit of bracing for improving fusionrates or clinical outcomes following instrumented lumbarfusion for degenerative disease.


The most relevant questions for the spine surgeon maybe related to the ability of a trial of brace therapy to pre-dict functional outcomes following lumbar fusion surgeryand the ability of postoperative braces to improve func-tional and radiographic outcomes. The first issue could beresolved by randomizing half of a population of patientsalready selected for surgery to a trial of bracing prior tolumbar fusion surgery. The results of surgery could thenbe compared with the results of bracing. Depending on themethods of patient allocation and outcome measures,Class I or II medical evidence could be derived. To deter-mine the efficacy of postoperative bracing, a comparisonbetween patients undergoing similar lumbar fusion proce-dures, randomized to brace therapy or no such therapy,would provide high-quality evidence to support or refutethe hypothesis that bracing improves fusion rates and/orfunctional outcome following lumbar fusion procedures inthe treatment of lumbar degenerative disease.

References

1. Alexander A, Woolley SM, Bisesi M, et al: The effectiveness ofback belts on occupational back injuries and worker perception.Professional Safety 40:22–26, 1995

2. Axelsson P, Johnsson R, Stromqvist B: Effect of lumbar ortho-sis on intervertebral mobility. A roentgen stereophotogrammet-ric analysis. Spine 17:678–681, 1992

3. Axelsson P, Johnsson R, Stromqvist B: Lumbar orthosis withunilateral hip immobilization. Effect on intervertebral mobilitydetermined by roentgen stereophotogrammetric analysis. Spine18:876–879, 1993

4. Axelsson P, Johnsson R, Stromqvist B, et al: External pedicularfixation of the lumbar spine: outcome evaluation by functionaltests. J Spinal Disord 12:147–150, 1999

5. Axelsson P, Johnsson R, Stromqvist B, et al: Temporary exter-nal pedicular fixation versus definitive bony fusion: a prospec-tive comparative study on pain relief and function. Eur SpineJ 12:41–47, 2003

6. Axelsson P, Johnsson R, Stromqvist B, et al: Orthosis as prognostic instrument in lumbar fusion: no predictive value in 50 cases followed prospectively. J Spinal Disord 8:284–288,1995

7. Bednar DA: Failure of external spinal skeletal fixation to im-prove predictability of lumbar arthrodesis. J Bone Joint SurgAm 83:1656–1659, 2001

8. Bednar DA, Raducan V: External spinal skeletal fixation in themanagement of back pain. Clin Orthop Relat Res 322:131–139, 1996

9. Connolly PJ, Grob D: Bracing of patients after fusion for de-generative problems of the lumbar spine—yes or no? Spine 23:1426–1428, 1998

10. Coxhead CE, Inskip H, Meade TW, et al: Multicentre trial ofphysiotherapy in the management of sciatic symptoms. Lancet1:1065–1068, 1981

11. Dillingham TR: Lumbar supports for prevention of low backpain in the workplace. JAMA 279:1826–1828, 1998

12. Esses SI, Botsford DJ, Kostuik JP: The role of external spinalskeletal fixation in the assessment of low-back disorders. Spine14:594–601, 1989

13. Faraj AA: External fixation in lumbar segmental instability.Acta Orthop Belg 69:9–12, 2003

14. Faraj AA, Akasha K, Mulholland RC: Temporary external fix-ation for low back pain: is it worth doing? Eur Spine J 6:187–190, 1997

15. Hsieh CY, Phillips RB, Adams AH, et al: Functional outcomesof low back pain: comparison of four treatment groups in a ran-domized controlled trial. J Manipulative Physiol Ther 15:4–9, 1992

16. Jeanneret B, Jovanovic M, Magerl F: Percutaneous diagnosticstabilization for low back pain. Correlation with results afterfusion operations. Clin Orthop Relat Res 304:130–138, 1994

17. Jellema P, Bierma-Zeinstra SM, Van Poppel MN, et al: Fea-sibility of lumbar supports for home care workers with lowback pain. Occup Med (Lond) 52:317–323, 2002

18. Jellema P, van Tulder MW, van Poppel MN, et al: Lumbar sup-ports for prevention and treatment of low back pain: a system-atic review within the framework of the Cochrane Back ReviewGroup. Spine 26:377–386, 2001

19. Johnsson R: The use of orthoses in lumbar spine fusion. ActaOrthop Scand Suppl 251:92–93, 1993

20. Johnsson R, Stromqvist B, Axelsson P, et al: Influence of spinalimmobilization on consolidation of posterolateral lumbosacralfusion. A roentgen stereophotogrammetric and radiographicanalysis. Spine 17:16–21, 1992

21. Kraus JF, Brown KA, McArthur DL, et al: Reduction of acutelow back injuries by use of back supports. Int J Occup En-viron Health 2:264–273, 1996

22. Lahad A, Malter AD, Berg AO, et al: The effectiveness of fourinterventions for the prevention of low back pain. JAMA272:1286–1291, 1994

23. Lantz SA, Schultz AB: Lumbar spine orthosis wearing. I. Re-striction of gross body movements. Spine 11:834–837, 1986

24. Magerl F: Stabilization of the lower thoracic and lumbar spinewith external skeletal fixation. Clin Orthop Relat Res 189:125–141, 1984

25. Million R, Haavik-Nilsen K, Jayson MIV, et al: Evaluation oflow back pain and assessment of lumbar corsets with and with-out back supports. Ann Rheum Dis 40:449–454, 1981

26. Minor SD: Use of back belts in occupational settings. PhysTher 76:403–408, 1996

27. Mitchell LV, Lawler FH, Bowen D, et al: Effectiveness andcost-effectiveness of employer-issued back belts in areas ofhigh risk for back injury. J Occup Med 36:90–94, 1994

28. Nachemson AL: Orthotic treatment for injuries and diseases ofthe spinal column. Phys Med Rehabil 1:11–24, 1987

29. Olerud S, Sjostrom L, Karlstrom G, et al: Spontaneous effect ofincreased stability of the lower lumbar spine in cases of severechronic back pain. The answer of an external transpeduncularfixation test. Clin Orthop Relat Res 203:67–74, 1986

30. Ordeberg G, Enskog J, Sjostrom L: Diagnostic external fixationof the lumbar spine. Acta Orthop Scand Suppl 251:94–96,1993

31. Pavlov PW, De Kleuver M, Spruit M, et al: Evaluating externalspinal skeletal fixation used predictively to select candidates forlumbar arthrodesis: nonfixation controls. J Bone Joint SurgAm 84:1714, 2002

32. Pope MH, Phillips RB, Haugh LD, et al: A prospective ran-



723


domized three-week trial of spinal manipulation, transcuta-neous muscle stimulation, massage, and corset in the treatmentof subacute low back pain. Spine 19:2571–2577, 1994

33. Reddell CR, Congleton JJ, Dale Huchingson R, et al: An eval-uation of a weightlifting belt and back injury prevention training class for airline baggage handlers. Appl Ergon 23:319–329, 1992

34. Sahni IK, Hipp JA, Kirking BC, et al: Use of percutaneoustranspedicular external fixation pins to measure intervertebralmotion. Spine 24:1890–1893, 1999

35. Soini J, Seitsalo S: The external fixation test of the lumbarspine. Thirty complications in 25 of 100 consecutive patients.Acta Orthop Scand 64:147–149, 1993

36. Soini J, Slatis P, Kannisto M, et al: External transpedicular fix-ation test of the lumbar spine correlates with the outcome ofsubsequent lumbar fusion. Clin Orthop Relat Res 293:89–96,1993

37. Soini JR, Harkonen HI, Alaranta HT, et al: External fixation testin low back pain. Function analyzed in 25 patients. Acta Or-thop Scand 65:87–90, 1994

38. Thompson L, Pati AB, Davidson H, et al: Attitudes and backbelts in the workplace. Work 4:22–27, 1994

39. Tiusanen H, Seitsalo S, Osterman K, et al: The role of externaltranspedicular fixation in anterior interbody lumbar fusion. JSpinal Disord 9:433–438, 1996

40. Valle-Jones JC, Walsh H, O’Hara J, et al: Controlled trial of aback support (‘Lumbotrain’) in patients with non-specific lowback pain. Curr Med Res Opin 12:604–613, 1992

41. van der Schaaf DB, van Limbeek J, Pavlov PW: Temporaryexternal transpedicular fixation of the lumbosacral spine. Spine24:481–485, 1999

42. van Poppel MN, de Looze MP, Koes BW, et al: Mechanisms ofaction of lumbar supports: a systematic review. Spine 25:2103–2113, 2000

43. van Poppel MN, Koes BW, van der Ploeg T, et al: Lumbar sup-ports and education for the prevention of low back pain in indus-try: a randomized controlled trial. JAMA 279:1789–1794, 1998

44. van Tulder MW, Jellema P, van Poppel MN, et al: Lumbar sup-ports for prevention and treatment of low back pain. CochraneDatabase Syst Rev:CD001823, 2000

45. Walsh NE, Schwartz RK: The influence of prophylactic or-thoses on abdominal strength and low back injury in the work-place. Am J Phys Med Rehabil 69:245–250, 1990

46. Willner S: Effect of a rigid brace on back pain. Acta OrthopScand 56:40–42, 1985






Diagnostic Recommendations


Guidelines. Use of intraoperative SSEP or DSEP moni-toring is recommended as an adjunct in those circum-stances during instrumented lumbar spinal fusion proce-dures in which the surgeon desires immediate intraoperativeinformation regarding the potential of a neurological injury.The occurrence of a postoperative neurological deficit ishighly correlated with intraoperative changes in thesemonitoring modalities. An abnormal SSEP or DSEP dur-ing surgery, however, often does not correlate with a post-operative neurological injury because of a high false-pos-itive rate.

Use of intraoperative evoked EMG recordings is rec-ommended in those circumstances in which the operatingsurgeon wishes to confirm the lack of a neurological in-jury during pedicle screw placement. A normal evokedEMG response is highly predictive of the lack of a neuro-logical injury. An abnormal EMG response during the sur-

gical procedure may or may not be associated with a clin-ically significant injury.

Options. Intraoperative evoked EMG recording is rec-ommended as an option during lumbar spinal fusion sur-gery in those situations in which the operating surgeon de-sires immediate information regarding the integrity of thepedicle wall, as a normal evoked EMG response is corre-lated with an intact pedicle wall.

Therapeutic Recommendations



Options. Intraoperative SSEP, DSEP, EMG, and/or evok-ed EMG monitoring are recommended only as adjunctiveoptions during instrumented lumbosacral fusion proce-dures for degenerative spinal disease. The use of any ofthese modalities has not been convincingly demonstrated toinfluence patient outcome favorably.

Rationale

Intraoperative electrophysiological monitoring of spin-al cord and nerve root function is used in a variety of clin-ical scenarios. Various techniques are thought to be usefulfor the detection and prevention of neurological deficits



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 15: electrophysiological monitoring and lumbar fusion


Department of Neurosurgery, University of Wisconsin, Madison, Wisconsin; Department ofNeurosurgery, Mount Sinai Medical School, New York, New York; Department of Neurosurgery,University of Washington, Seattle, Washington; Department of Neurosurgery, Indiana University,Indianapolis, Indiana; Departments of Orthopedic Surgery and Neurosurgery, University ofCalifornia at Los Angeles, California; Department of Neurosurgery, University of Alabama atBirmingham, Alabama; Department of Neurosurgery, Emory University, Atlanta, Georgia; Bone and Joint Clinic of Houston, Texas; and Department of Neurosurgery, Brown University, Providence,Rhode Island

KEY WORDS • lumbar spine • fusion • electromyography • practice guidelines

725

Abbreviations used in this paper: CMAP = compound muscleaction potential; CT = computerized tomography; DSEP = der-matomal sensory evoked postential; EMG = electromyography;NMEP = neurogenic motor evoked potential; NPV = negative pre-dictive value; PPV = positive PV; SSEP = somatosensory evokedpotential.


during surgery to repair aortic aneurysms, correct scoliot-ic or traumatic spinal deformities, and during resection ofspinal cord tumors.11,13,29,30,38,45 Intraoperative monitoring forroutine lumbosacral fusion for degenerative disease is alsobeing performed. The primary justification for the use ofthese surgical adjuncts is the perception that the safety andefficacy of pedicle screw fixation are enhanced through theuse of intraoperative electrophysiological monitoring. Thepurpose of this review is to examine the medical evidenceconcerning intraoperative monitoring to answer the follow-ing questions. 1) Does intraoperative electrophysiologicalmonitoring of the nerve roots or spinal cord increase thesafety of lumbar or lumbosacral instrumentation? 2) Doesthe use of intraoperative electrophysiological monitoring ofthe spinal cord and nerve roots influence patient outcomesfollowing lumbar spinal surgery for degenerative disease?

Search Criteria

A computerized search of the database of the NationalLibrary of Medicine from 1966 to March 2003 was con-ducted using the search terms “electrophysiology andspinal surgery,” or “EMG and spinal surgery,” or “evokedpotentials and spinal surgery.” The search was restricted tothe English language and yielded a total of 1068 citations.The titles and abstracts of each of these references werereviewed and papers not concerned with the use of moni-toring for lumbosacral fusion were discarded. Referenceswere identified that provided either direct or supportingevidence relevant to the use of monitoring for lumbar orlumbosacral fusion procedures. These papers were obtain-ed and reviewed, and relevant references from the bibli-ographies of these papers were also identified. All papersproviding Class II or better medical evidence regarding theuse of electrophysiological monitoring for lumbar or lum-bosacral fusion procedures are summarized in Table 1.Additional information is provided by other references list-ed in the bibliography.


There are two potential uses for electrophysiologicalmonitoring: an intraoperative diagnostic test for nerve in-jury during screw placement or a treatment tool to improvepostoperative outcome. Because two different uses of mon-itoring are being considered, medical evidence was classi-fied (as Class I, II, or III) separately when the evidence wasused to support monitoring as a diagnostic study or whenmonitoring was used as a treatment modality. The level ofevidence provided by a particular paper is therefore re-ferred to as diagnostic, treatment, or both in Table 1.

Somatosensory Evoked Potential Monitoring

Intraoperative SSEP monitoring has been used in spinalsurgery since the 1940s.2,12,30 Initially, SSEPs were usedfor assessing spinal cord function during scoliosis surgery,resection of spinal cord tumors, spinal cord decompres-sion, or during vascular procedures with the potential toalter spinal cord blood flow.4,11,13,20,21,23,24,30,33,40–42,45,46 Morerecently, the use of this technique has been extended topatients being treated with lumbosacral fusion for degen-erative disease. Balzer, et al.,1 reported their results from a

group of 44 patients who were treated with lumbosacralfusion. All patients were studied with SSEP as well asspontaneous EMG and evoked EMG monitoring. Thirty-eight of these patients were surgically treated for degener-ative spinal disease; postoperative S-1 radiculopathydeveloped in two. Intraoperative SSEPs were reported tobe abnormal in one of these patients, despite repositioningof a screw. Intraoperative SSEPs were also reported to beabnormal in another patient who did not suffer a neuro-logical deficit. In this patient, recordings normalized fol-lowing adjustment of a cross-link. The sensitivity of SSEPmonitoring for the detection of new deficits in this serieswas 50%.1 Although the NPV of normal SSEPs in thisseries was high (97%), the incidence of safe screw place-ment irrespective of SSEP monitoring was also high(95%). Bose and colleagues3 monitored continuous SSEPsas well as continuous EMG in a group of 61 patients beingtreated with instrumented posterolateral fusion proceduresof the lumbosacral spine. These authors reported thatSSEP monitoring was less useful than EMG monitoringfor the detection of nerve injuries. Lencke, et al.,22 simi-larly used continuous SSEP as well as evoked EMG mon-itoring during placement of 233 lumbar pedicle screws.They reported that no useful information was provided bythe SSEP monitoring. Reidy and colleagues37 noted thatSSEPs were relatively insensitive to pedicle fracture dur-ing thoracic pedicle screw placement when comparedwith EMG monitoring. Mochida, et al.,28 compared SSEPmonitoring with evoked EMG monitoring and found thatthe use of evoked EMG monitoring was more sensitiveand specific for diagnosing intraoperative injury to nerveroots than were SSEPs.

Manninen26 reported his experience with continuousSSEP recording in a series of 309 patients who were treat-ed with spinal surgery, 169 of whom underwent surgery fordegenerative lumbar disease. He noted a 4.4% rate of false-positive changes in responses, a 1.1% incidence of false-negative changes in responses, a sensitivity of 57%, and aspecificity of 95%. When the lumbar group was consideredalone, the sensitivity of SSEP monitoring for detecting newdeficits was 50%. Nishijima and colleagues31 noted signif-icant SSEP changes in 11 of 58 patients they treated withthoracolumbar or lumbar surgery, yet none of these patientswere noted to have a postoperative deficit. Meyer and col-leagues27 compared their results with SSEP monitoring in agroup of patients with traumatic injuries of the thora-columbar spine. They compared two groups of patientswho underwent surgical stabilization of the spine: 150 withmonitoring and 145 without monitoring. These authorsfound that new neurological injuries occurred less fre-quently in the monitored group. They reported that sixpatients had significant changes in SSEP responses, onlyone of whom had a neurological deterioration (sensitivity100%, specificity 96%, PPV 16%, NPV 100%). In thecomparison group, 10 of 145 patients had new postopera-tive neurological deficits. There were, however, significantdifferences between the patient groups. The nonmonitoredgroup was made up of a historical cohort combined withmore severely injured patients in whom there were no reli-able SSEPs preoperatively. The relevance of these findingsto the current discussion is questionable because of dispar-ities between the treatment groups, because the patient pop-ulation treated was a trauma population, and because the





Intraoperative monitoring

727

TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

ele

ctro

phys

iolo

gica

l m

onit

orin

g*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Her

ron,

et a

l., 1

987

III

30 p

atie

nts

w/ l

umba

r st

enos

is h

ad in

trao

p D

SEP.

Im

prov

emen

ts in

mon

itori

ng

DSE

Ps im

prov

e af

ter

deco

mpr

essi

on b

ut p

rogn

ostic

impo

rtan

ce is

unc

lear

.w

ere

seen

pos

tdec

ompr

essi

on in

all

patie

nts.

No

sign

ific

ant d

iffe

renc

e bt

wn

mag

nitu

de o

f im

prov

emen

t btw

n pa

tient

s w

/ goo

d &

fai

r or

poo

r re

sults

. In

2pa

tient

s, la

ck o

f im

prov

emen

t led

to f

urth

er d

ecom

pres

sion

. Con

tral

at r

espo

nses

impr

oved

som

ewha

t w/ u

nila

t pro

cedu

res.

Mey

er, e

t al.,

198

8II

I tr

eatm

ent

295

patie

nts

unde

rwen

t op

for

trau

ma

& w

ere

retr

ospe

ctiv

ely

anal

yzed

. 150

w/

SSE

Pm

onito

ring

is s

ensi

tive

for

dete

ctio

n of

ner

ve r

oot i

njur

y. U

se o

f

II d

iagn

osis

SSE

Ps, 1

45 n

ot m

onito

red.

New

def

icits

mor

e co

mm

on in

non

mon

itore

d gr

oup.

m

onito

ring

may

impr

ove

safe

ty o

f tr

aum

a su

rger

y in

thor

acic

& th

o-

Non

mon

itore

d gr

oup

wer

e pa

tient

s in

who

m r

espo

nses

cou

ld n

ot b

e el

icite

d &

ra

colu

mba

r sp

ine.

Sen

sitiv

ity 1

00%

, spe

cifi

city

97%

, PPV

17%

, NPV

hist

oric

al c

ases

. 5 f

alse

-pos

itive

SSE

Pal

erts

.10

0%, L

R(1

) 33

, LR

(2)

0.O

wen

, et a

l., 1

988

III

NM

EPs

mon

itore

d in

111

pat

ient

s; s

ucce

ssfu

l in

90%

. No

men

tion

of r

esul

ts. N

o N

ME

Ps c

an b

e m

easu

red

in m

ost p

atie

nts.

pa

tient

had

neu

rolo

gica

l inj

ury.

Mix

ed p

atie

nt p

opul

atio

n.G

epst

ein

& B

row

ning

, 198

9II

IO

bser

vatio

nal s

tudy

of

41 p

atie

nts

w/ S

SEPs

w/ l

amin

ecto

my

for

HN

Por

ste

nosi

s.

Dec

ompr

essi

on o

f ne

rve

root

s is

ass

ocia

ted

w/ i

mpr

oved

SSE

Pla

tenc

ies.

Impr

ovem

ents

in la

tenc

y co

rrel

ated

w/ f

unct

iona

l out

com

e at

3 m

os b

ut n

ot 1

yr.

All

patie

nts

had

impr

oved

late

ncy.

C

ohen

, et a

l., 1

991a

III

trea

tmen

tV

ery

sele

ct g

roup

of

12 p

atie

nts

w/ p

reop

wea

knes

s m

onito

red

w/ D

SEPs

for

de-

D

SEP

may

impr

ove

afte

r ne

rve

root

dec

ompr

essi

on. I

mpr

ovem

ent i

s a

com

pres

sive

op.

9 h

ad “

com

plet

e no

rmal

izat

ion”

of

DSE

Ps &

rec

over

ed s

tren

gth.

fa

vora

ble

prog

nost

ic s

ign.

3 ha

d “v

aria

ble”

nor

mal

izat

ion

& d

id n

ot r

egai

n st

reng

th.

Coh

en, e

t al.,

199

1bII

I tr

eatm

ent

Pude

ndal

ner

ve D

SEP

perf

orm

ed in

154

pat

ient

s w

/ sac

ral f

ixat

ion.

1 p

atie

nt lo

st

Pude

ndal

ner

ve D

SEP

mon

itori

ng is

pos

sibl

e in

mos

t pat

ient

s. U

tility

is

III

diag

nosi

sre

spon

ses.

No

patie

nt h

ad a

def

icit.

uncl

ear.

Nis

hijim

a, e

t al.,

199

2II

I58

pat

ient

s w

/ lum

bar

spin

e op

(10

fus

ion)

mon

itore

d w

/ SSE

Ps w

/dou

ble

stim

uli.

Use

fuln

ess

of S

SEP

mon

itori

ng f

or lu

mba

r su

rger

y un

clea

r.L

oss

of 2

nd r

espo

nse

in 1

1 pa

tient

s. N

o pa

tient

had

det

ecta

ble

post

op d

efic

it.

No

men

tion

of a

ltera

tion

of te

chni

que

in r

espo

nse

to m

onito

ring

abn

orm

aliti

es.

Hir

aizu

mi,

et a

l., 1

993

III

5 pa

tient

s w

/ sev

ere

clau

dica

tion

had

preo

p &

intr

aop

mon

itori

ng. S

ympt

oms

cor-

Sacr

al E

Pm

easu

rem

ents

can

be

abno

rmal

in p

atie

nts

w/ s

ever

e cl

audi

ca-

rela

te w

/ sac

ral E

Pm

easu

rem

ents

. 4 p

atie

nts

had

op &

all

had

impr

ovem

ent i

n tio

n. I

mpr

ovem

ent c

an o

ccur

aft

er d

ecom

pres

sion

.am

plitu

de. C

linic

al r

esul

ts e

xcel

lent

in 2

pat

ient

s &

goo

d in

2.

Rob

inso

n, e

t al.,

199

3II

dia

gnos

is26

pat

ient

s w

/ lum

bar

frac

ture

s st

udie

d w

/ fem

oral

ner

ve S

SEPs

. 5 p

atie

nts

had

Fem

oral

ner

ve S

SEPs

can

det

ect i

njur

y to

upp

er lu

mba

r ro

ots.

Sen

sitiv

ity

III

trea

tmen

tsi

gnif

ican

t SSE

Pch

ange

s. 1

had

pos

top

defi

cit d

espi

te im

med

iate

act

ion

take

n 10

0%, s

peci

fici

ty 8

7%, P

PV40

%, N

PV10

0%, L

R(1

) 7.

7, L

R(2

) 0.

by

the

surg

eon.

Util

ity f

or in

crea

sing

saf

ety

uncl

ear.

Tole

ikis

, et a

l., 1

993

II d

iagn

osis

80 p

atie

nts

had

lum

bar

op w

/ PS

fixa

tion

mon

itore

dw

/ DSE

Ps. 6

9 ha

d “m

inim

al”

DSE

Pca

n de

tect

ner

ve r

oot i

njur

y. S

ensi

tivity

66%

, spe

cifi

city

88%

, II

I tr

eatm

ent

chan

ges

duri

ng o

p &

non

e ha

d ne

w d

efic

it. 1

1 ha

d D

SEPs

dis

appe

ar d

urin

g op

. PP

V18

%, N

PV98

%, L

R(1

) 5.

5, L

R(2

) 0.

39. U

tility

unc

lear

.2

had

neur

olog

ical

def

icits

. 1 r

elat

ed to

dee

p ve

nous

thro

mbo

sis,

1 to

dur

al te

ar.

Ow

en, e

t al.,

199

4II

I80

pat

ient

s m

onito

red

w/ m

echa

nica

lly e

licite

d E

MG

. 17.

8% o

f ne

rve

root

s st

udie

d M

echa

nica

lly e

licite

d E

MG

can

det

ect n

erve

roo

t irr

itatio

n. I

mpo

rtan

ce

dem

onst

rate

d so

me

firi

ng. I

n 1,

fir

ing

led

to r

emov

al o

f PS

. No

post

op d

efic

its

uncl

ear.

repo

rted

.C

aste

llo, e

t al.,

199

5II

I30

pat

ient

s ha

d de

com

pres

sive

op

for

HN

Por

lat r

eces

s st

enos

is w

/ SSE

P. R

es-

SSE

Ps m

ay im

prov

e af

ter

deco

mpr

essi

on o

f ne

rve

root

.po

nses

gen

eral

ly im

prov

ed a

fter

dis

cect

omy

or f

oram

inot

omy.

No

clin

ical

out

- co

me

give

n.G

lass

man

, et a

l., 1

995

III

trea

tmen

tPr

ospe

ctiv

e st

udy

of 5

12 P

Ss p

lace

d in

90

patie

nts.

All

PSs

stim

ulat

edw

/ EM

G &

CM

AP

thre

shol

d of

.15

mA

likel

y re

pres

ents

intr

aped

icul

ar s

crew

.II

I di

agno

sis

CM

AP.

All

patie

nts

unde

rwen

t CT

for

PS p

ositi

on. 1

1 PS

s ad

just

ed b

ased

on

EM

G (

repo

sitio

ning

did

not

impr

ove

thre

shol

d) &

4 P

Ss a

djus

ted

base

d on

im-

agin

g w

/ nor

mal

EM

G th

resh

old

(.15

mA

). N

orm

al E

MG

thre

shol

d pr

edic

ted

no o

bvio

us P

S m

alpo

sitio

n w

/ an

NPV

of 9

8%. D

owng

rade

d to

Cla

ss I

II d

ue to

m

etho

dolo

gica

l iss

ues

of C

Tas

sess

men

t.L

enck

e, e

t al.,

199

5II

I23

3 PS

spl

aced

in lu

mba

r or

sac

ral p

edic

les

in 5

4 pa

tient

s w

/ con

tinuo

us S

SEP

&A

utho

rs p

ropo

se th

resh

old

guid

elin

es f

or a

sses

smen

t of

pedi

cle

wal

l th

en tr

igge

red

EM

G w

/ thr

esho

lds

mea

sure

d. 2

29 s

crew

s w

ere

wel

l pla

ced

& 4

in

tegr

ity (

4 m

A, 4

–8 m

A, .

8 m

A).

ques

tiona

ble.

Onl

y 1

post

op d

efic

it in

a p

atie

nt w

hose

PS

was

not

rep

lace

d &

w

hose

def

icit

may

not

hav

e an

ythi

ng to

do

w/ t

he P

S. N

o C

Tda

ta.

Con

tinu

ed




TAB

LE

1 C

onti

nued

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Mag

uire

, et a

l., 1

995

III

273

reco

rdin

gs m

ade

from

ner

ve r

oots

, dri

ll bi

ts, &

PSs

dur

ing

plac

emen

t. C

urre

nt

Aut

hors

sug

gest

6-m

Ath

resh

old

as b

eing

indi

cativ

e of

ped

icle

wal

l th

resh

olds

mor

e re

liabl

e th

an v

olta

ge th

resh

olds

for

mea

sure

men

t pur

pose

s. 1

3in

tegr

ity.

of 2

39 in

stru

men

tatio

n m

easu

re ,

6 m

A. 8

of

13 w

ere

inco

rrec

tly p

lace

d ra

dio-

grap

hica

lly. 2

of

13 r

epos

ition

ed a

fter

dir

ect i

nspe

ctio

n; 3

of

13 le

ft in

pla

ce

afte

r in

spec

tion.

1 p

atie

nt w

/ . 6

-mA

thre

shol

d ha

d a

clea

rly

mis

plac

ed s

crew

on

intr

aop

radi

ogra

phs.

No

CT

data

.C

lem

ents

, et a

l., 1

996

III

trea

tmen

t25

pat

ient

s re

ceiv

ed 1

12 P

Ss w

/ EM

G &

CM

AP.

PSs

pal

pate

d or

insp

ecte

d vi

sual

ly.

EM

G &

CM

AP

can

dete

ct p

edic

le v

iola

tion.

Eff

ect o

n ou

tcom

e un

clea

r.II

dia

gnos

is12

PSs

w/ s

timul

atin

g th

resh

old

,11

mA

brok

e th

roug

h co

rtex

. All

cort

ical

Fo

r ne

rve

inju

ry, s

ensi

tivity

100

%, s

peci

fici

ty 9

0%, P

PV8%

, NPV

viol

atio

ns n

oted

by

palp

atio

n or

insp

ectio

n. 2

pat

ient

s ha

d a

new

pos

top

(pai

n 10

0%, L

R(1

) 10

, LR

(2)

0.on

ly)

defi

cit.

Dar

den,

et a

l., 1

996

III

trea

tmen

t13

2 pa

tient

s pr

ospe

ctiv

ely

mon

itore

d w

/ EM

G. P

atie

nts

w/o

EM

G r

espo

nse

at

EM

G c

an d

etec

t ner

ve in

jury

. Sen

sitiv

ity 9

0%, s

peci

fici

ty 4

5%, P

PVII

dia

gnos

is40

Vha

d a

10.6

% r

ate

of n

ew r

adic

ulop

athy

, pat

ient

s w

/ pos

itive

res

pons

e le

ad-

40%

, NPV

91%

, LR

(1)

1.64

, LR

(2)

0.22

. Unc

lear

why

Gro

up 1

did

in

g to

no

corr

ectiv

e ac

tion

had

a 15

% n

ew r

adic

ulop

athy

rat

e, &

no

patie

nts

w/

wor

se th

an G

roup

3. M

ajor

ity o

f E

MG

ala

rms

led

to n

o ad

just

men

ts.

a po

sitiv

e E

MG

res

pons

e w

hose

PS

was

red

irec

ted

had

a ne

w d

efic

it. C

Tco

n-fi

rmed

mal

posi

tion

in 2

.2%

of

type

1 &

4.9

% o

f ty

pe 2

pat

ient

s.M

ochi

da, e

t al.,

199

7II

I“T

rain

spi

nal s

timul

atio

n” u

sed

in 3

4 pa

tient

s w

/ tho

raci

c or

thor

acol

umba

r fu

sion

. “T

rain

spi

nal s

timul

atio

n,”

a fo

rm o

f C

MA

P, w

as f

ound

to b

e se

nsiti

ve

10%

late

ncy

dela

y th

at d

id n

ot im

prov

e w

as a

ssoc

iate

d w

/ pos

top

defi

cits

. SSE

Ps

for

the

dete

ctio

n of

pos

top

defi

cits

. SSE

Ps w

ere

not.

did

not d

etec

t any

cha

nges

.Ts

ai, e

t al.,

199

7II

ID

SEPs

in 3

3 pa

tient

s w

/ lum

bar

deco

mpr

essi

on. 5

8% o

f pa

tient

s ha

d re

liabl

e re

s-D

SEP

mon

itori

ng is

not

fea

sibl

e in

man

y pa

tient

s w

/ lum

bar

nerv

e ro

ot

pons

es. A

ll pa

tient

s ha

d im

prov

emen

t in

resp

onse

s po

stde

com

pres

sion

. Clin

ical

co

mpr

essi

on. I

mpr

ovem

ent i

n D

SEP

crite

ria

does

not

cor

rela

te w

/ re

sults

var

iabl

e &

did

not

cor

rela

te w

/ DSE

Pre

sults

.cl

inic

al o

utco

me.

Bal

zer,

et a

l., 1

998

III

trea

tmen

t44

pat

ient

s ha

d op

w/ S

SEP

& E

MG

w/ C

MA

P. 6

inst

ance

s of

EM

G a

bnor

mal

ities

SS

EP

& E

MG

mon

itori

ng w

ere

sens

itive

for

intr

aop

nerv

e ro

ot in

jury

. II

dia

gnos

isno

ted,

all

repo

rted

to s

urge

on. 2

of

thes

e pa

tient

s ha

d po

stop

def

icits

. 2 in

stan

ces

N

orm

al m

onito

ring

par

amet

ers

wer

e pr

edic

tive

of n

o ne

w in

jury

. Eff

ect

of S

SEP

chan

ges

note

d, 1

pat

ient

had

pos

top

defi

cit.

of m

onito

ring

on

ultim

ate

outc

ome

uncl

ear.

Sens

itivi

ty s

peci

fici

ty,N

PV,

PPV

, LR

( 1),

LR

(2)

for

SSE

Pw

ere

100%

, 98%

, 100

%, 5

0%, 4

3, &

0.

The

sam

e va

lues

for

EM

G w

ere

100%

, 90%

, 100

%, 3

3%, 1

0.5,

& 0

. D

arde

n, e

t al.,

199

8II

I tr

eatm

ent

Pros

pect

ive

com

pari

son

of im

peda

nce

mea

sure

men

ts, E

MG

& C

MA

Pto

det

ect

Impe

danc

e m

easu

rem

ents

unr

elia

ble

for

asse

ssin

g pe

dicl

e w

all v

iola

-II

I di

agno

sis

pedi

cle

wal

l vio

latio

n in

20

patie

nts.

98

pedi

cle

hole

s &

PSs

test

ed w

/ all

mo-

tion.

EM

G s

ensi

tive

for

dete

ctin

g pe

dicl

e w

all v

iola

tion.

dalit

ies

& a

ll pe

dicl

e w

alls

wer

e in

spec

ted

visu

ally

&/o

r by

pal

patio

n. 2

ped

i-cl

es v

iola

ted,

bot

h ha

d ab

norm

al E

MG

val

ues.

Im

peda

nce

mea

sure

men

ts u

n-re

liabl

e. N

o cl

inic

al d

ata

give

n. N

o po

stop

ass

essm

ent o

f PS

pos

ition

.M

anni

nen,

199

8II

dia

gnos

ticSS

EP

in 1

49 lu

mba

r op

s (d

isc

& s

teno

sis)

. SSE

Pch

ange

s in

8 &

def

icits

in 4

pa-

Sens

itivi

ty o

f SS

EP

duri

ng lu

mba

r su

rger

y fo

r di

agno

sing

new

def

icit

is

tient

s. 2

fal

se-n

egat

ive

SSE

Pst

udie

s &

5 f

alse

pos

itive

s.

50%

, spe

cifi

city

is 9

6%, N

PV98

%, P

PV25

%, L

R(1

) 12

.5, L

R(2

) 0.

52.

Nor

cros

s-N

echa

y, e

t al.,

199

9II

I70

pat

ient

s m

onito

red

w/ S

SEPs

for

lum

bar

sten

osis

op

(a s

ubse

t had

fus

ion)

. 12

SEPs

can

det

ect i

ntra

oper

ativ

e ne

rve

inju

ries

. Sen

sitiv

ity 1

00%

, spe

ci-

patie

nts

had

intr

aop

dete

rior

atio

n of

SSE

Ps. 9

had

no

clin

ical

seq

uela

. 3 h

ad s

ig-

fici

ty 8

6%, P

PV25

%, N

PV10

0%, L

R( 1

) 7.

14, L

R(2

) 0.

nifi

cant

neu

rolo

gica

l def

icits

pos

top.

Op

“adj

ustm

ents

” w

ere

mad

e &

hig

h-do

se

ster

oids

wer

e gi

ven

to a

ll w

/ SSE

Pde

teri

orat

ion.

Rei

dy, e

t al.,

200

1II

Pros

pect

ive

stud

y of

95

PSs

in th

orac

ic s

pine

w/ E

MG

& S

SEP.

CT

& c

linic

al F

UE

MG

(C

MA

P) &

SSE

Ps d

o no

t im

prov

e ac

cura

cy o

r sa

fety

of

thor

acic

pe

rfor

med

. No

chan

ges

in S

SEPs

. 90%

of

scre

ws

accu

rate

ly p

lace

d.PS

pla

cem

ent.

Sens

itivi

ty o

f E

MG

for

det

ectin

g pe

dicl

e br

each

was

50

%, s

peci

fici

ty 8

3%, N

PV94

%, P

PV22

%, L

R(1

) 2.

9, L

R(2

) 0.

6.

Bos

e, e

t al.,

200

2II

I tr

eatm

ent

61 p

atie

nts

had

op w

/ SSE

P&

EM

G. 1

4 ev

ents

in 1

3 (2

1%)

patie

nts.

Eve

nts

re-

EM

G &

SSE

Pse

nsiti

ve f

or n

erve

inju

ry. E

ffec

t of

mon

itori

ng o

n ul

ti-II

dia

gnos

ispo

rted

to s

urge

on w

ho m

ade

chan

ges.

The

re w

as 1

pos

top

defi

cit i

n pa

tient

w/

mat

e ou

tcom

e is

unc

lear

. Sen

sitiv

ity 1

00%

, spe

cifi

city

80%

, NPV

“sus

tain

ed n

euro

toni

c di

scha

rges

” du

ring

ret

ract

ion

of th

ecal

sac

.10

0%, P

PV8%

, LR

(1)

5, L

R(2

) 0.

Hag

high

i, 20

02II

I41

pat

ient

s w

/ CM

APs

for

dec

ompr

essi

ve o

ps. N

o ch

ange

s w

/ dec

ompr

essi

on o

f C

MA

Ps d

o no

t pro

vide

use

ful p

rogn

ostic

info

rmat

ion

afte

r de

com

pres

-ne

rve

root

s in

6 p

atie

nts

w/ r

adic

ulop

athy

.si

on o

f ne

rve

root

s.

*FU

= f

ollo

w u

p; H

NP

= h

erni

ated

nuc

leus

pul

posu

s; L

R =

late

ncy

resp

onse

; PS

= p

edic

le s

crew

.


medical and surgical treatment paradigm used in the studyis outdated and not currently practiced.

Norcross-Nechay, et al.,32 monitored 70 patients theytreated with lumbar decompression, instrumentation, andfusion with continuous SSEP monitoring and noted signif-icant changes in 12 cases. In all 12 instances, immediateintraoperative “adjustments” were made and high-dose ste-roids (1.5 mg/kg dexamethasone intravenously) were givento those patients whose responses did not return to norm-al. Three patients had persistent deficits postoperatively.Therefore, the sensitivity of the SSEP monitoring for newinjury was 100%, the specificity was 86%, the PPV was25%, and the NPV was 100%. Similarly, Robinson, et al.,36

monitored femoral nerve SSEPs in 26 patients undergoingsurgery for lumbar fractures and found that five patientshad significant intraoperative recording changes, only oneof whom had a postoperative deficit. The sensitivity, speci-ficity, PPV, and NPV in this study (100%, 87%, 25%, and100%, respectively) were similar to those noted by Nor-cross-Nechay, et al.32 Other supportive series from the sco-liosis literature also indicate a significant false-positive rateand rare false-negative reports.17,36,46

Based on this information, it appears that SSEPs areable to detect many instances of nerve injury during tho-racolumbar and lumbar surgery. Somatosensory evokedpotential monitoring may be used to provide information tothe surgeon regarding the occurrence of an intraoperativeneurological injury. The majority of patients who awakenwith deficits will have had abnormal SSEPs intraoperative-ly. There is, however, no evidence to suggest that a surgicalresponse to SSEP changes influences patient outcome. Fur-thermore, there is a significant incidence of false alarmsassociated with the use of SSEPs that may cause the sur-geon to make “adjustments” that are unnecessary and po-tentially harmful. These data do not, therefore, support theuse of SSEP monitoring to improve outcome during thora-columbar and lumbar fusion procedures in patients whoundergo operation for degenerative disease.

Dermatomal Sensory Evoked Potential Monitoring

Dermatomal sensory evoked potential monitoring hasbeen reported to be a more specific means of intraopera-tive nerve root monitoring than SSEP monitoring,43,44 andhas been used to provide feedback to the surgeon duringlumbar fusion procedures. Toleikis, et al.,43 monitored 80patients they treated with lumbar spinal surgery and pedi-cle screw fixation and reported that DSEP responses werelost in 11 patients. Two of these 11 patients had postoper-ative deficits, one of whom had a deficit related to deepveinous thrombosis but unrelated to screw placement. Thesecond patient with a deficit was noted to lose responsesimmediately following a dural tear. Cohen, et al.,8 fol-lowed pudendal DSEPs in 154 patients. One patient wasnoted to have a deterioration in the DSEP response. Thispatient did not experience a postoperative neurologicaldeficit. It appears that DSEPs are capable of detectingnerve injury and are likely similar to SSEPs in terms ofdiagnostic accuracy. There are, however, fewer publica-tions to review concerning this monitoring modality. Theneurological abnormalities described in the available ser-ies were either irrelevant to surgical manipulation, falselypositive, or simply a marker of damage already done.

These data do not support the use of DSEP monitoring forimprovement of patient outcome following lumbar fusionfor degenerative disease.

Electromyographic Monitoring

Electromyography provides information regarding themotor function of the nerve roots and spinal cord. It hasbeen used in many forms as a monitoring method duringdecompression and instrumentation procedures in thelumbar spine. Electromyographic monitoring techniquescan be divided into those measuring mechanically elicitedaction potentials or those elicited by electrical stimulationof tissue or instruments. Various terms are used to des-cribe these responses depending on the stimulus used toprovoke a response from the nerve and the techniquesused to measure the response. Some of the more commonterms include evoked or elicited EMG, the CMAP, theNMEP, and persistently electrified pedicle screw instru-mentation. The following discussion concerns the use ofall modalities of EMG. Important differences in recordingand stimulating techniques used in the studies will be dis-cussed individually.

Balzer, Rose, and colleagues1,39 published a series of pa-pers detailing their use of continuously electrified surgicalinstruments during the placement of pedicle screws duringlumbar spinal surgery. These authors recorded both sponta-neous (mechanically elicited) and evoked EMG activity(using a variable voltage threshold of stimulation deter-mined intraoperatively) in patients being treated with instru-mented lumbar fusion procedures. In six of 44 cases re-ported by Balzer, et al.,1 persistent spontaneous EMGactivity was noted and “immediate measures” were under-taken to alleviate nerve root irritation. Two of these patientshad new radicular complaints following surgery despiteintraoperative adjustments. Bose and colleagues3 used a si-milar protocol (although they used a current threshold of 7mA as a cutoff for pedicle breach as opposed to an intraop-erative voltage measurement). In a series of 61 consecutivepatients, “significant neurophysiological events” occurredin 13 patients (21%). Sustained neurotonic discharges inresponse to retraction of the thecal sac were seen in five of40 patients being treated with interbody fusion. One of thesepatients had a postoperative deficit (temporary paraparesis).The sensitivity, specificity, PPV, and NPV for EMG moni-toring in this series were 100%, 94%, 25%, and 100%, res-pectively.

Clements, et al.,6 prospectively studied the efficacy ofevoked and spontaneous EMG in a series of 25 patientswho were treated with instrumented lumbar fusion proce-dures. One hundred twelve screws were placed in patients.These authors found that a pedicle breach was predictedwith 100% accuracy when the stimulating threshold wasless than 11 mA. Pedicle breach was confirmed by visual-ization or palpation of the medial pedicle wall at the timeof surgery. One patient with a pedicle breach experiencedpostoperative radiculopathy. Because the abnormal EMGresponses were all noted after the pedicle had already beenbreached and because all pedicle breaches were detectedby visualization and palpation, it is unclear if the use ofEMG contributed useful data to the surgeon. Owen, etal.,34 recorded mechanically elicited EMG in 89 patientsbeing treated with surgery for stenosis (80 lumbar and



729


nine cervical). They noted abnormal responses in 85(17.8%) of 476 nerve roots monitored. No informationwas provided regarding the clinical consequences of thisfrequently noted activity.

Darden and colleagues9,10 published two papers thatprovide useful information regarding EMG monitoringduring placement of pedicle screws. In a 1996 paper,10

they prospectively evaluated the insertion of pediclescrews in a series of 132 consecutive patients being treat-ed with instrumented lumbar spinal fusion procedures.They analyzed their data by using arbitrary threshold cut-offs of 20 and 40 V. Patients were divided into threegroups: Group 1, consisting of patients in whom no posi-tive EMG responses to stimulation were seen; Group 2,consisting of patients in whom a positive response wasnoted but no corrective action was undertaken; and Group3, patients in whom corrective action was undertaken fol-lowing a positive response. Using a 40 V threshold as acutoff, 35% of patients were in Group 1 (changes were re-ported to the surgeon in 65% of cases). In approximatelytwo thirds of the cases with abnormal responses, the sur-geon elected not to take any corrective action and in onethird of the cases the surgeon took corrective actions. Tenpercent of patients in Group 1 had postoperative deficits.Fifteen percent of patients in Group 2 had postoperativedeficits, and none of the Group 3 patients had a postopera-tive deficit. The sensitivity, specificity, PPV, and NPV forthis threshold level for predicting a neurological deficitwere 90%, 45%, 40%, and 91%. Using a 20 V stimulus,68% of patients were in Group 1, 16% of patients were inGroup 2, and 16% were in Group 3. The sensitivity was64%, the specificity 83%, the PPV 67%, and the NPV 90%.Computerized tomography scans were obtained in somepatients. In 2.1% of Group 1 and 3.3% of Group 2 patientsstudied with CT scanning demonstrable screw malpositionwas revealed.10 The authors recommend that screws withstimulation thresholds less than 20 V be subject to correc-tive action. It is noteworthy that there was a 10% rate ofnew injuries in the group of patients with threshold valuesgreater than 40 V. In a 1998 follow-up paper,9 these au-thors demonstrated that elicited EMG (using a 20-Vthreshold) was more specific and sensitive for pedicle wallviolation than screw hole impedance measurements.

Another potential clinical outcome measure that may beaffected by the use of electrophysiological monitoring isthe assessment of pedicle wall integrity. Several studieswere identified in which the absence of a pedicle breachor screw malposition was used as the primary outcomemeasure. The technique for placing a lumbar pediclescrew and the desired anatomical outcome is the same forcases involving screw fixation for scoliosis or degenera-tive disease. Therefore, it is reasonable to include dataconcerning lumbar pedicle screw placement for these dis-orders in this review. Lencke, et al.,22 performed a pro-spective study of evoked EMG monitoring during lumbarpedicle screw placement in a mixed patient populationmade up primarily of adolescents with scoliosis. In thisstudy, 233 screws were placed, 229 of which were be-lieved to be within the pedicle by clinical and EMG crite-ria. The majority (93%) of the well-placed screws hadthresholds greater than 8 mA and all had thresholdsgreater than 4 mA. Three of the four screws that wereknown to be malpositioned had thresholds less than 4 mA.

There was one patient with a screw threshold potential of4.7 mA who awoke with radicular symptoms. These au-thors concluded that a threshold potential of greater than 8mA is highly predictive of intraosseous placement. Thispaper is weakened by the lack of postoperative imagingstudies with which to assess screw location. Maguire, etal.,25 reviewed their experience with evoked EMG moni-toring and pedicle screw placement in a series of 29patients treated with 144 pedicle screws. They stimulateddrill bits, probes, and screws. Using a 6-mA threshold,they noted that stimulation of five screws with subthresh-old parameters resulted in a measurable response. All fivescrews were thought to be in an acceptable position basedon intraoperative radiographs. These screws were ex-plored and two were eventually repositioned. The otherthree screws were thought to be in good position and wereleft alone. Conversely, one screw was found to have a nor-mal EMG response (threshold . 6 mA) but was clearlymalpositioned based on intraoperative radiographs. Nopatient suffered a root injury.25 This study also lacked anobjective, consistent postoperative assessment of screwposition.

Glassman, et al.,15 published a prospective analysis ofintraoperative EMG monitoring compared with postoper-ative CT assessment of screw position in a series of 90patients treated with 512 pedicle screws. The authorsreported an NPV of 98% for evoked EMG thresholdsgreater than 15 mA. There were 11 screws that were repo-sitioned based primarily on EMG threshold, and fourscrews were repositioned based on radiographic criteriadespite a threshold greater than 15 mA. Fifteen screwswere noted to have thresholds less than 15 mA and wereexplored. Nine of the 15 demonstrated minimal penetra-tion of the pedicle wall; all penetrations were thought tobe clinically insignificant. The remaining six pedicleswere explored and found to be intact. None of the screwswas repositioned. Therefore, 25 screws were found tohave stimulation thresholds less than 15 mA. Of these 25,11 were repositioned. Twenty-two screws were found tobe in good position, and three screws (12%) were thoughtto be malpositioned.15 This compares with 10 screws thatwere malpositioned despite normal EMG responses (2%).This paper is considered to provide Class III medical evi-dence due to methodological concerns related to CTacquisition and grading criteria. The authors used CTscans with 5-mm cuts and used a grading system that con-sidered screws that could not be well visualized to be inthe “well-positioned” group. The clinical consequences ofscrew malposition in this series were not discussed. Reidy,et al.,26 performed a prospective assessment of thoracicpedicle screw placement by using EMG and postoperativeCT assessment of screw position. They found, in contrastto the previous study, that the use of intraoperative EMGrecordings to predict screw position was not useful in theclinical setting. In this study in which a 7-mA thresholdwas used, the sensitivity of EMG for detection of screwbreakout was 50%, the specificity was 83%, the PPV22%, and the NPV 94%.37 It should be emphasized thatthis study was performed in thoracic pedicles and there-fore may not be directly relevant to this discussion be-cause the use of EMG monitoring in the thoracic spinemay be less reliable compared with the lumbar spine.Overall, the medical evidence derived from the literature




indicates that the lack of a recordable EMG response fol-lowing pedicle screw stimulation is highly predictive ofacceptable placement of the pedicle screw. There is noevidence to suggest, however, that the safety or efficacy oflumbar pedicle screw fixation is improved with the use ofintraoperative EMG monitoring.

Monitoring During Decompression

Review of the literature revealed a series of papers inwhich the issue of using electrophysiological monitoring todetermine the adequacy of nerve root decompression or topredict improvement in function following surgery wasexamined. For example, Castello and colleagues5 followedSSEPs during nerve root decompression in patients treatedfor a herniated disc. They found that SSEPs generallyimproved more with decompression of the lateral recessthan with discectomy alone. There was no correlation withclinical outcome, however.5 Gepstein and Brown14 record-ed SSEPs in a group of 41 patients they treated with lum-bar nerve root decompression. Although there was an earlycorrelation of improved SSEP latency with improved clin-ical outcome at 3 weeks following surgery, this associationwas not present 1 year following surgery. These authorsreported performing further exploration in six patientsbecause of a failure to obtain improvements in SSEP laten-cy. In all six patients, a second compressive lesion wasfound. Haghighi16 recorded motor evoked potentials in sixpatients with radiculopathy who underwent lumbar decom-pression and fusion. No improvements in the potentialswere noted and no clinical outcome data were presented.

Cohen, et al.,7 retrospectively reviewed a very selectgroup of 12 patients (in a total population of 150) withweakness due to compression of lumbar nerve roots. Nineof the patients had complete normalization of DSEPs fol-lowing decompression and had improved strength postop-eratively. The other three patients had “variable improve-ment” in DSEPs and did not recover strength. Clearly, achronically damaged nerve may not return to normal func-tion even after adequate decompression. Therefore, al-though complete normalization of responses may be apositive predictor of improved function, the value of in-formation provided by a “variable improvement” is lessclear. Herron and colleagues18 recorded DSEPs in 30 pa-tients who were treated with decompression for lumbarstenosis. Patient outcomes were graded as good in 21patients, fair in seven, and poor in one patient (based on anonvalidated outcome scale). Patients with good out-comes had a mean decrease in latency of 9.9 msec com-pared with a mean decrease in latency of 8.2 msec in thefair group and 6 msec in the patient with the poor out-come. There was a noted improvement in latencies on theunoperated side in patients with unilateral symptoms aswell (mean 3.4 msec, range 0–7.6 msec). In two patients,failure of improvement in DSEPs led to further explo-ration, and a compressive lesion was found in one. Al-though this paper suggests a modest association betweenimprovement in DSEP latency and outcome, there was noclearcut threshold of improvement to guide the surgeon.Tsai, et al.,44 also followed DSEP potentials during lumbardecompression surgery in 33 patients and correlated intra-operative changes in DSEP responses to clinical outcome.The DSEPs were recordable in 19 (68%) of 33 patients.

The authors reported complete normalization in DSEPlatencies in every surgically treated patient who had adetectable baseline abnormality (13 of 19). There was nocorrelation between normalization of DSEP responses andclinical outcome.44

Summary

Based on the medical evidence provided by the litera-ture reviewed, there does not appear to be support for thehypothesis that any form of intraoperative monitoringimproves patient outcomes following lumbar decompres-sion or fusion procedures for degenerative spinal disease.Evidence does indicate that a normal evoked EMG res-ponse is predictive for intrapedicular screw placement(high NPV for breakout). The presence of an abnormalEMG response does not, however, exclude intrapedicularscrew placement (low PPV). The majority of clinicallyapparent postoperative nerve injuries are associated withintraoperative changes in SSEP and/or DSEP monitoring.For this reason, changes in DSEP/SSEP monitoring ap-pear to be sensitive to nerve root injury. There is a high–false positive rate, however, and changes in DSEP andSSEP recordings are frequently not related to nerve injury.A normal study has been shown to correlate with the lackof a significant postoperative nerve injury. There is nosubstantial evidence to indicate that the use of intraopera-tive monitoring of any kind provides useful information tothe surgeon in terms of assessing the adequacy of nerveroot decompression at the time of surgery.

Future Research Directions

The main issues addressed in this review are the contri-bution of intraoperative monitoring to the clinical safetyand effectiveness of lumbosacral fusion procedures fordegenerative spinal disease. No studies reviewed provid-ed more than suggestive evidence on these issues. A ran-domized prospective study comparing clinical and radio-graphic outcomes in similar groups of patients undergoinglumbosacral fusion either with or without intraoperativemonitoring would provide Class I evidence supporting orrefuting the hypothesis that the added expense associatedwith the use of intraoperative monitoring is justified by aclinical benefit.

References

1. Balzer JR, Rose RD, Welch WC, et al: Simultaneous soma-tosensory evoked potential and electromyographic recordingsduring lumbosacral decompression and instrumentation. Neu-rosurgery 42:1318–1325, 1998

2. Bieber E, Tolo V, Uematsu S: Spinal cord monitoring duringposterior spinal instrumentation and fusion. Clin Orthop RelatRes 229:121–124, 1988

3. Bose B, Wierzbowski LR, Sestokas AK: Neurophysiologicmonitoring of spinal nerve root function during instrumentedposterior lumbar spine surgery. Spine 27:1444–1450, 2002

4. Brown RH, Nash CL Jr, Berilla JA, et al: Cortical evoked po-tential monitoring. A system for intraoperative monitoring ofspinal cord function. Spine 9:256–261, 1984

5. Castello PH, Place HM, Hemler DE, et al: Quantification oflumbar nerve root decompression using somatosensory-evokedpotentials. J Spinal Disord 8:444–450, 1995



731


6. Clements DH, Morledge DE, Martin WH, et al: Evoked andspontaneous electromyography to evaluate lumbosacral pediclescrew placement. Spine 21:600–604, 1996

7. Cohen BA, Major MR, Huizenga BA: Predictability of adequa-cy of spinal root decompression using evoked potentials. Spine16 (8 Suppl):S379–S384, 1991

8. Cohen BA, Major MR, Huizenga BA: Pudendal nerve evokedpotential monitoring in procedures involving low sacral fixa-tion. Spine 16 (8 Suppl):S375–S378, 1991

9. Darden BV II, Owen JH, Hatley MK, et al: A comparison ofimpedance and electromyogram measurements in detecting thepresence of pedicle wall breakthrough. Spine 23:256–262, 1998

10. Darden BV II, Wood KE, Hatley MK, et al: Evaluation of pedi-cle screw insertion monitored by intraoperative evoked elec-tromyography. J Spinal Disord 9:8–16, 1996

11. Dawson EG, Sherman JE, Kanim LE, et al: Spinal cord monitor-ing. Results of the Scoliosis Research Society and the EuropeanSpinal Deformity Society survey. Spine 16 (8 Suppl):S361–S364, 1991

12. Dawson GD: Cerebral responses to electrical stimulation ofperipheral nerves in man. J Neurol Neurosurg Psychiatry 10:137–140, 1947

13. Dinner DS, Luders H, Lesser RP, et al: Intraoperative spinalsomatosensory evoked potential monitoring. J Neurosurg 65:807–814, 1986

14. Gepstein R, Brown MD: Somatosensory-evoked potentials inlumbar nerve root decompression. Clin Orthop Relat Res 245:69–71, 1989

15. Glassman SD, Dimar JR, Puno RM, et al: A prospective analy-sis of intraoperative electromyographic monitoring of pediclescrew placement with computed tomographic scan confirma-tion. Spine 20:1375–1379, 1995

16. Haghighi SS: Monitoring of motor evoked potentials with highintensity repetitive transcranial electrical stimulation duringspinal surgery. J Clin Monit Comput 17:301–308, 2002

17. Harper CM Jr, Daube JR, Litchy WJ, et al: Lumbar radiculopa-thy after spinal fusion for scoliosis. Muscle Nerve 11:386–391,1988

18. Herron LD, Trippi AC, Gonyeau M: Intraoperative use of der-matomal somatosensory-evoked potentials in lumbar stenosissurgery. Spine 12:379–383, 1987

19. Hiraizumi Y, Transfeldt EE, Fujimaki et al: Electrophysiologicevaluation of intermittent sacral nerve dysfunction in lumbarspinal canal stenosis. Spine 18:1355–1360, 1993

20. Jellinek D, Jewkes D, Symon L: Noninvasive intraoperativemonitoring of motor evoked potentials under propofol anesthe-sia: effects of spinal surgery on the amplitude and latency ofmotor evoked potentials. Neurosurgery 29:551–557, 1991

21. Keith RW, Stambough JL, Awender SH: Somatosensory corti-cal evoked potentials: a review of 100 cases of intraoperativespinal surgery monitoring. J Spinal Disord 3:220–226, 1990

22. Lenke LG, Padberg A, Russo MH, et al: Triggered electromyo-graphic threshold for accuracy of pedicle screw placement. An ani-mal model and clinical correlation. Spine 20:1585–1591, 1995

23. Loder RT, Thomson GJ, LaMont RL: Spinal cord monitoring inpatients with nonidiopathic spinal deformities using somatosen-sory evoked potentials. Spine 16:1359–1364, 1991

24. Macon JB, Poletti CE, Sweet WH, et al: Conducted somatosen-sory evoked potentials during spinal surgery. Part 2: clinicalapplications. J Neurosurg 57:354–359, 1982

25. Maguire J, Wallace S, Madiga R, et al: Evaluation of intrapedi-cular screw position using intraoperative evoked electromyog-raphy. Spine 20:1068–1074, 1995

26. Manninen PH: Monitoring evoked potentials during spinal sur-gery in one institution. Can J Anaesth 45:460–465, 1998

27. Meyer PR Jr, Cotler HB, Gireesan GT: Operative neurologicalcomplications resulting from thoracic and lumbar spine internalfixation. Clin Orthop Relat Res 237:125–131, 1988

28. Mochida K, Komori H, Okawa A, et al: Evaluation of motorfunction during thoracic and thoracolumbar spinal surgery basedon motor-evoked potentials using train spinal stimulation. Spine22:1385–1393, 1997

29. More RC, Nuwer MR, Dawson EG: Cortical evoked potentialmonitoring during spinal surgery: sensitivity, specificity, relia-bility, and criteria for alarm. J Spinal Disord 1:75–80, 1988

30. Nash CL Jr, Lorig RA, Schatzinger LA, et al: Spinal cord mon-itoring during operative treatment of the spine. Clin OrthopRelat Res 126:100–105, 1977

31. Nishijima Y, Okada M, Yasuaki Y, et al: Intraoperative moni-toring for thoracolumbar or lumbar surgery with somatosenso-ry evoked potentials after double stimuli. Spine 17:1304–1308,1992

32. Norcross-Nechay K, Mathew T, Simmons JW, et al: Intra-operative somatosensory evoked potential findings in acute andchronic spinal canal compromise. Spine 24:1029–1033, 1999

33. Nuwer MR, Dawson EG, Carlson LG, et al: Somatosensoryevoked potential spinal cord monitoring reduces neurologicdeficits after scoliosis surgery: results of a large multicenter sur-vey. Electroencephalogr Clin Neurophysiol 96:6–11, 1995

34. Owen JH, Kostuik JP, Gornet M, et al: The use of mechanical-ly elicited electromyograms to protect nerve roots during sur-gery for spinal degeneration. Spine 19:1704–1710, 1994

35. Owen JH, Laschinger J, Bridwell K, et al: Sensitivity and speci-ficity of somatosensory and neurogenic-motor evoked poten-tials in animals and humans. Spine 13:1111–1118, 1988

36. Papastefanou SL, Henderson LM, Smith NJ, et al: Surface elec-trode somatosensory-evoked potentials in spinal surgery: impli-cations for indications and practice. Spine 25:2467–2472, 2000

37. Reidy DP, Houlden D, Nolan PC, et al: Evaluation of electro-myographic monitoring during insertion of thoracic pediclescrews. J Bone Joint Surg Br 83:1009–1014, 2001

38. Robinson LR, Slimp JC, Anderson PA, et al: The efficacy offemoral nerve intraoperative somatosensory evoked potentialsduring surgical treatment of thoracolumbar fractures. Spine18:1793–1797, 1993

39. Rose RD, Welch WC, Balzer JR, et al: Persistently electrifiedpedicle stimulation instruments in spinal instrumentation. Tech-nique and protocol development. Spine 22:334–343, 1997

40. Stechison MT, Panagis SG, Reinhart SS: Somatosensory evok-ed potential. Monitoring during spinal surgery. Acta Neuro-chir 135:56–61, 1995

41. Szalay EA, Carollo JJ, Roach JW: Sensitivity of spinal cordmonitoring to intraoperative events. J Pediatr Orthop 6:437–441, 1986

42. Tamaki T, Tsuji H, Inoue S, Kobayashi H. The prevention ofiatrogenic spinal cord injury utilizing the evoked spinal cordpotential. Int Orthop 4:313–317, 1981

43. Toleikis JR, Carlvin AO, Shapiro DE, et al: The use of der-matomal evoked responses during surgical procedures that useintrapedicular fixation of the lumbosacral spine. Spine 18:2401–2407, 1993

44. Tsai RY, Yang RS, Nuwer MR, et al: Intraoperative dermatomalevoked potential monitoring fails to predict outcome from lum-bar decompression surgery. Spine 22:1970–1975, 1997

45. Wiedemayer H, Fauser B, Sandalcioglu IE, et al: The impact ofneurophysiological intraoperative monitoring on surgical deci-sions: a critical analysis of 423 cases. J Neurosurg 96:255–262, 2002

46. Wilber RG, Thompson GH, Shaffer JW, et al: Postoperativeneurological deficits in segmental spinal instrumentation. Astudy using spinal cord monitoring. J Bone Joint Surg Am66:1178–1187, 1984






Recommendations

Standards. The use of autologous bone or rhBMP-2bone graft substitute is recommended in the setting of anALIF in conjunction with a threaded titanium cage.


Options. 1) Recombinant human BMP-2 in combinationwith HA and tricalcium phosphate may be used as a sub-stitute for autograft bone in some cases of PLF. 2) Severalformulations of calcium phosphate exist and are recom-mended as bone graft extenders, especially when used incombination with autologous bone.

Rationale

Successful arthrodesis following lumbar fusion requiresosseous bridging between the vertebral bodies, which isusually achieved by placing graft material between the

vertebral bodies, which then heal over time. The standardgraft material is harvested autogenous bone, which maybe limited by availability and may be associated with do-nor-site morbidity. Allograft bone may also be used forvarious applications; however, availability, cost, risk ofdisease transmission, and lack of osteoinductive capacitylimit the utility of allograft in some applications. For thesereasons, bone graft substitutes have been developed forapplication in the lumbar spine. These substitutes havevariable mechanical properties and biological activities,and they may or may not be efficacious for specific situa-tions. The purpose of this review is to examine the med-ical evidence regarding the use of bone graft substitutes inlumbar spinal surgery.

Search Criteria

An electronic search of the database of the National Li-brary of Medicine from 1966 to November 2003 was per-formed using the search terms “bone graft substitute” as akey word and then again as the search focus. The searchwas repeated using search terms “bone substitutes,” “trical-cium phosphate,” “calcium phosphate,” “bone morpho-genetic protein,” and “hydroxyapatite” combined with“spine” and “lumbar.” The search was limited to the Eng-lish language and to reports on humans. The results of thesearches were combined, and a total of 54 articles were



Guidelines for the performance of fusion procedures for degenerative disease of the lumbar spine. Part 16: bone graft extenders and substitutes



KEY WORDS • fusion • lumbar spine • bone graft • bone substitute •treatment outcome • practice guidelines

733

Abbreviations used in this paper: AGF = autologous growth fac-tor; ALIF = anterior lumbar interbody fusion; AWGC = apatite andwollanosite–containing glass ceramic; BMP = bone morphogeneticprotein; HA = hydroxyapatite; ODI = Oswestry Disability Index;PLF = posterolateral fusion; RCT = randomized controlled trial;rhBMP = recombinant human BMP.


identified and reviewed. The reference lists of each of thesepapers was reviewed, and further references were identifiedand subsequently submitted for review. The vast majority ofreferences found included animal data and were thereforeeliminated. There were also several papers dealing with cer-vical interbody fusion and scoliosis. Ultimately, six paperswere identified as providing Class III or better data regard-ing the use of bone substitutes in lumbar fusion for degen-erative disease. These papers are described in Table 1.

Scientific Rationale

Bone graft substitutes and extenders may be classifiedinto two main categories, the first consisting of biologicalagents that induce the formation of bone from native tis-sues. The best known member of this category is rhBMP-2. Other examples of this category include other membersof the BMP family and autogenous growth factor concen-trates. The second class of bone substitutes comprises cal-cium phosphate salts of varying composition used to pro-vide a scaffold for the growth of new bone. Members ofthis second category include b-tricalcium phosphate, hy-droxyapatite, and wollanosite.

Recombinant rhBMP-2 is the best studied of all the bio-logical agents. Three recent clinical series have describedthe use of this substance in humans undergoing fusion forlumbar degenerative disease. Burkus, et al.,2,3 investigatedthe use of rhBMP-2 as a substitute for autograft when usedin combination with a titanium cage for an anterior lum-bar interbody fusion. These investigators performed anRCI comparing rhBMP-2 with autograft in a group of 279well-matched patients with lumbar degenerative disease.2

Fusion status was assessed with both plain radiography(static and dynamic) and computerized tomography scan-ning studies reviewed by radiologists blinded as to bonegraft material used. Clinical outcome measures usedincluded the ODI for low-back pain, patient satisfaction,and visual analog scale scores for leg, back, and graft-sitepain. Follow-up duration was at least 2 years, and theauthors achieved greater than 90% follow up in both treat-ment groups. The authors reported significant improve-ments in ODI, back pain, leg pain, and patient satisfactionscores in both treatment groups. There was a slightly high-er fusion rate in the rhBMP-2 group (94.5% comparedwith 88.7%, probability value not significant). There wereadvantages to the use of rhBMP-2 in terms of a slightlyshorter operating room time (24 minutes) and slightlydecreased blood loss (44 ml). There was also an advantagefor the rhBMP-2 group in terms of donor-site pain. Thisstudy was well designed and used appropriate radiograph-ic and clinical outcome measures. As such, it providesClass I medical evidence supporting the use of rhBMP-2as a bone graft substitute in ALIF involving a titaniumcage. The use of rhBMP-2 in this study was also associat-ed with decreased donor-site pain and shorter operatingroom times.

The same group of authors performed a secondary an-alysis of these data and data derived from another clinicalseries.3 In this second paper, the authors asserted that therhBMP-2 was associated with higher fusion rates thanautograft bone in titanuim cage–augmented ALIF.Interpretation of the data reported in this second papermust be performed with caution. Because much of the

data used for this secondary analysis were derived fromuncontrolled cohort studies and because this was areanalysis of previously published data, the data used tosupport rhBMP-2 as superior to autograft is consideredClass III medical evidence. Given the fact that a Class Istudy established that rhBMP-2 and autograft bone wereessentially equivalent with regard to the success of fusion,the impact of this second paper is minimal. The patients inthe rhBMP-2 group experienced advantages in operatingtime, graft-site morbidity, and blood loss.2

Boden and colleagues1 examined the role of rhBMP-2in combination with b-tricalcium phosphate and HA as abone graft substitute for PLF. They performed a pilotstudy involving 25 patients randomized to instrumenta-tion-based PLF with autograft (five), instrumented PLFwith rhBMP-2 plus carrier (11), or noninstrumented PLFwith rhBMP-2 plus carrier (nine). Patients were followedfor a mean of 17 months. Fusion was assessed with dy-namic plain radiography, augmented by computerized to-mography scanning in cases of uncertainty regardingfusion status. The ODI and portions of the SF-36 ques-tionnaire were used to assess clinical outcomes. Theauthors reported fusion rates of 100% in both rhBMP-2groups, compared with only 40% in the autograft group.The authors also reported improvements in functional out-come in the rhBMP-2 groups compared with the autograftgroup. They concluded that rhBMP-2 may be an effectivealternative to autograft. Although this study was a ran-domized trial, there are several methodological concernsthat limit the quality of its medical evidence. Its smallpopulation resulted in differences among the treatmentgroups. The autograft group had significantly fewer pa-tients with education beyond high school, had a signifi-cantly larger percentage of patients with diabetes, moresmokers, and more patients pursuing Workers’ Compen-sation claims. In the autograft treatment group the fusionrate was only 40%, a much lower rate than that reported innumerous other studies. In this study, there was no signif-icant advantage noted for the instrumented rhBMP-2group in terms of operative time, blood loss, or hospitalstay compared with the instrumented autograft group. Be-cause of the small size of the study and the methodologi-cal concerns resulting from the small sample size, thisstudy is considered to provide Class III medical evidencesupporting the use of rhBMP-2, in combination with a car-rier, in the performance of a PLF, with or without place-ment of instrumentation.

Other biological bone growth stimulators have been usedfor the treatment of lumbar degenerative disease in hu-mans. Lowery, et al.,6 performed a retrospective review of19 patients in whom AGF was used as a bone graft exten-der. Ultraconcentrated platelets derived from the patient’sown blood combined with thrombin created a gel that wasmixed with autograft bone (iliac crest graft in 14 and localautograft in five). Seven patients underwent 360˚ lumbarfusion procedures, eight were treated with PLF, and fourunderwent ALIF. All patients received supplemental pedi-cle screw instrumentation. All patients were considered tohave successful fusion, and no graft-specific complicationswere identified. This study provides Class III medical evi-dence that the use of AGF as a graft extender is safe. Itsefficacy as a graft extender has not been compared withautograft alone or other graft extenders.




The use of processed calcium phosphate salts as bonegraft extenders in the setting of lumbar fusion for degen-erative disease has also been reported. Kasai, et al.,4

recently reported their experience using a mixture of localautograft bone combined with varying amounts ofAWGC. The ceramic was used as a graft extender and wascombined with local autograft bone in a 2:1, 1:1, or 1:2ratio. Thirty-five patients with degenerative disease andstenosis who underwent decompression followed by non-instrumented two-level PLF were randomized to one ofthree groups (defined by the ratio of AWGC to autograft).Fusion rates in the three groups were equivalent (82–83%).Because there was no control group (autograft alone), thispaper provides Class III medical evidence supporting thesafety of AWGC as a bone graft extender for noninstru-mented PLF.

Linovitz and Peppers5 described their experience withthe use of b-tricalcium phosphate ceramic as a graft exten-der at twelve levels in seven patients who underwent inter-body fusion for degenerative disease of the lumbar spine.These patients underwent interbody fusion procedures thatinvolved placement of allograft bone as a structural graftand the ceramic used as an osteoconductive matrix. Nobone marrow aspiration was performed; the ceramic wasmixed with venous blood instead. In all seven patientsfusion was achieved at all operated levels when assessedwith plain radiography 3 to 6 months following surgery.This paper provides Class III medical evidence supportingthe use of b-tricalcium phosphate as a bone graft extenderfor interbody fusion procedures. Thalgott and colleagues7,8

performed retrospective analyses of patients treated witheither ALIF or PLF with processed coralline HA. In twothirds of the patients treated with a PLF demineralizedbone matrix was used in combination with the HA.Interestingly, in patients treated with a combination of HAand demineralized bone matrix there was a slightly lowerfusion rate than in those treated with HA alone. All groupsof patients had fusion rates comparable to historical con-trols. These papers provide Class III medical evidencewith regard to the use of coralline HA as a bone graftextender or substitute.

Summary

Despite the large volume of animal data regarding theuse of synthetic bone graft substitutes or extenders, thereare very few data regarding the use of these substances forfusion in lumbar degenerative disease. The best availabledata indicate that rhBMP-2 is a viable alternative to auto-graft bone for interbody fusion procedures. This samesubstance may also be a viable alternative to autograft forPLF; however, definitive medical evidence is not yetavailable. There is little, if any, medical evidence to sup-port the use of other biological agents at the present time.As promising new compounds are brought to market,well-designed cohort studies and randomized trials will berequired to determine the actual usefulness of these com-pounds in clinical practice. It is important not to general-ize the results obtained with one preparation or applica-tion to different preparations or applications.

The use of synthetic calcium phosphate ceramics asgraft extenders appears to be reasonable in certain situa-


Bone graft substitutes

735

TAB

LE

1Su

mm

ary

of s

tudi

es i

nvol

ving

lum

bar

fusi

on a

nd b

one

graf

t ex

tend

ers

and

subs

titu

tes*

Aut

hors

& Y

ear

Cla

ssD

escr

iptio

nC

omm

ent

Low

ery,

et a

l., 1

999

III

AG

F us

ed in

com

bina

tion

w/ a

utog

raft

& c

eram

ic g

raft

ext

ende

r in

19

patie

nts.

Pat

ient

s di

d w

ell o

vera

ll.A

GFs

app

ear

to b

e sa

fe w

hen

used

as

a gr

aft e

xten

-de

r in

lum

bar

fusi

ons.

Bod

en, e

t al.,

200

2II

ISm

all p

ilot R

CT

com

pari

ng r

hBM

P-2

to a

utog

raft

for

PL

F. D

owng

rade

d to

Cla

ss I

II b

ecau

se o

f sm

all s

ize

&rh

BM

P-2

used

in c

ombi

natio

n w

/ HA

–TC

Pre

sults

in

sign

ific

ant d

iscr

epan

cies

in th

e ch

arac

teri

stic

s of

gro

ups

(sm

okin

g, d

iabe

tes,

Wor

kers

’Com

pens

atio

n).

high

er f

usio

n ra

tes

than

aut

ogra

ft w

/ or

w/o

inst

ru-

men

tatio

n. r

hBM

P-2

may

be

used

as

a bo

ne g

raft

su

bstit

ute.

Bur

kus,

et a

l., 2

002

IM

ultic

ente

r R

CT

com

pari

ng r

hBM

P-2

w/ a

utog

raft

in A

LIF

w/ s

peci

fic

(LT

) ca

ge. R

esul

ts in

term

s of

fus

ion

&

rhB

MP-

2 is

a v

iabl

e al

tern

ativ

e to

aut

ogra

ft f

or

outc

ome

equi

vale

nt w

/ 91%

FU

.A

LIF

w/ t

itani

um c

ages

.L

inov

itz &

Pep

pers

, 200

2II

ISe

ries

of

7 pa

tient

s tr

eate

d w

/ int

erbo

dy f

usio

ns u

sing

b2-

TC

P(s

uppl

emen

ted

w/ v

enou

s bl

ood

only

) in

com

-b

2-T

CP

may

be

used

as

a bo

ne g

raft

sub

stitu

te in

bi

natio

n w

/ str

uctu

ral a

utog

raft

. Fus

ion

cons

ider

ed p

rese

nt in

all

case

s.co

mbi

natio

n w

/ str

uctu

ral a

llogr

aft &

ven

ous

bloo

d in

an

inte

rbod

y ap

plic

atio

n.B

urku

s, e

t al,

2003

III

Com

bine

d an

alys

is o

f se

vera

l ser

ies

com

pari

ng r

hBM

P-2

w/ a

utog

raft

for

AL

IF w

/ spe

cifi

c (L

T)

cage

. Cla

imed

rh

BM

P-2

is a

via

ble

alte

rnat

ive

to a

utog

raft

for

th

at r

hBM

P-2

grou

p di

d be

tter

than

AL

IF g

roup

.A

LIF

w/ t

itani

um c

ages

.K

asai

, et a

l., 2

003

III

RC

Tco

mpa

ring

3 g

roup

s tr

eate

d w

/ non

inst

rum

ente

d PL

F us

ing

vari

ous

mix

ture

s of

loca

l aut

ogra

ft &

AW

GC

AW

GC

app

ears

to b

e ef

fect

ive

as a

bon

e gr

aft e

xten

-(2

:1, 1

:1, 1

:2).

The

3 g

roup

s ha

d co

mpa

rabl

e fu

sion

rat

es. T

here

was

no

cont

rol g

roup

.de

r w

hen

used

w/ l

ocal

aut

ogra

ft.

*FU

= f

ollo

w u

p; T

CP

= tr

ical

cium

pho

spha

te.


tions. The medical evidence available regarding their useis limited and of poor quality. Further study will be req-uired to establish their utility for use in spinal fusion.


As demonstrated by the trials used to evaluate rhBMP-2, clinical RCTs are feasible for the evaluation of bonegraft substitutes. The existence of a true gold standard forcomparison allows for high-quality data to be generatedsupporting or refuting the efficacy of any bone graft sub-stitute. Prior to undertaking such trials, however, the effi-cacy and safety of any given product must be demonstrat-ed across many animal species. The reader is remindednot to generalize results from one application to another(such as using ALIF data to support use in a PLF applica-tion) or to generalize results from one preparation toanother (such as using rhBMP-2 data to support the use ofAGF).

References

1. Boden SD, Kang J, Sandhu H, et al: Use of recombinant humanbone morphogenetic protein-2 to achieve posterolateral lum-bar spine fusion in humans: a prospective, randomized clinicalpilot trial: 2002 Volvo Award in clinical studies. Spine 27:2662–2673, 2002

2. Burkus JK, Gornet MF, Dickman CF, et al: Anterior lumbar

interbody fusion using rhBMP-2 with tapered interbody cages.J Spinal Disord 15:337–349, 2002

3. Burkus JK, Heim SE, Gornet MF, et al: Is INFUSE bone graftsuperior to autograft bone? An integrated analysis of clinicaltrials using the LT-CAGE lumbar tapered fusion device. J SpinDisord Tech 16:113–122, 2003

4. Kasai Y, Takegami K, Uchida A: Mixture ratios of local boneto artificial bone in lumbar posterolateral fusion. J Spin DisordTech 16:31–37, 2003

5. Linovitz RJ, Peppers TA: Use of an advanced formulation ofbeta-tricalcium phosphate as a bone extender in interbody lum-bar fusion. Orthopedics 25 (5 Suppl):S585–S589, 2002

6. Lowery GL, Kulkarni S, Pennisi AE: Use of autologous growthfactors in lumbar spinal fusion. Bone 25 (2 Suppl):S47–S50,1999

7. Thalgott JS, Giuffre JM, Fritts K, et al: Instrumented postero-lateral lumbar fusion using coralline hydroxyapatite with orwithout demineralized bone matrix, as an adjunct to autologousbone. Spine J 1:131–137, 2001

8. Thalgott JS, Klezl Z, Timlin M, et al: Anterior lumbar interbodyfusion with processed sea coral (coralline hydroxyapatite) aspart of a circumferential fusion. Spine 27:E518–E527, 2002






Health & Medicine

Guidelines for the preformance of fusion procedures for degenerative disease of the lumbar spine