Jurnal Meningitis Anak

n engl j med

349;5

www.nejm.org july

31, 2003

The

new england journal

of

medicine

435

original article

Risk of Bacterial Meningitis in Children with Cochlear Implants

Jennita Reefhuis, Ph.D., Margaret A. Honein, Ph.D., Cynthia G. Whitney, M.D., Shadi Chamany, M.D., Eric A. Mann, M.D., Ph.D.,

Krista R. Biernath, M.D., Karen Broder, M.D., Susan Manning, M.D., Swati Avashia, M.D., Marcia Victor, M.P.H., Pamela Costa, M.A., Owen Devine, Ph.D.,

Ann Graham, C.R.N.A., M.P.H., and Coleen Boyle, Ph.D.

From the National Center on Birth De-fects and Developmental Disabilities (J.R.,M.A.H., K.R.B., M.V., P.C., O.D., C.B.), theEpidemiology Program Office (J.R., S.C.,K.B., S.M., S.A.), the National Center for In-fectious Diseases (C.G.W., S.C.), and theNational Immunization Program (K.B.),Centers for Disease Control and Prevention,Atlanta; the Food and Drug Administration,Rockville, Md. (E.A.M., A.G.); the New YorkCity Department of Health and Mental Hy-giene, New York (S.M.); and the Texas De-partment of Health, Austin (S.A.). Addressreprint requests to Dr. Reefhuis at the Na-tional Center on Birth Defects and Devel-opmental Disabilities, Centers for DiseaseControl and Prevention, 1600 Clifton Rd. NE,MS E-86, Atlanta, GA 30333.

N Engl J Med 2003;349:435-45.

Copyright 2003 Massachusetts Medical Society.

background

In June 2002, the Food and Drug Administration received reports of bacterial meningi-tis in patients with cochlear implants for treatment of hearing loss. Implants that in-cluded a positioner (a wedge inserted next to the implanted electrode to facilitate trans-mission of the electrical signal by pushing the electrode against the medial wall of thecochlea) were voluntarily recalled in the United States in July 2002.

methods

We identified patients with meningitis and conducted a cohort study and a nested casecontrol investigation involving 4264 children who had received cochlear implants in theUnited States between January 1, 1997, and August 6, 2002, and who were less than sixyears of age when they received the implants. We calculated the incidence of meningitisin the cohort and assessed risk factors for meningitis among patients and among 199controls, using data from interviews with parents and abstracted from medical records.

results

We identified 26 children with bacterial meningitis. The incidence of meningitis causedby

Streptococcus pneumoniae

was 138.2 cases per 100,000 person-years more than 30times the incidence in a cohort of the same age in the general U.S. population. Postim-plantation bacterial meningitis was strongly associated with the use of an implant witha positioner (odds ratio, 4.5 [95 percent confidence interval, 1.3 to 17.9], with adjust-ment for medical, surgical, and environmental factors) and with the joint presence of ra-diographic evidence of a malformation of the inner ear and a cerebrospinal fluid leak(adjusted odds ratio, 9.3 [95 percent confidence interval, 1.2 to 94.5]). The incidenceof meningitis among patients who had received an implant with a positioner remainedhigher than the incidence among those whose implants did not have a positioner forthe duration of follow-up (24 months from the time of implantation).

conclusions

Parents and health care providers should ensure that all children who receive cochlearimplants are appropriately vaccinated and are then monitored and treated promptly forany bacterial infections after receiving the implant.

The New England Journal of Medicine Downloaded from nejm.org on May 26, 2013. For personal use only. No other uses without permission.

Copyright 2003 Massachusetts Medical Society. All rights reserved.

n engl j med

349;5

www.nejm.org july

31

,

2003

The

new england journal

of

medicine

436

early 10,000 children in the unit-

ed States with severe-to-profound hear-ing loss currently have a cochlear implant,

a surgically implanted device that includes an elec-trode array that is inserted into the cochlea.

1

Withthe use of cochlear implants, people who have se-vere-to-profound hearing loss may perceive soundsand learn to speak.

2

In June 2002, one manufacturer of cochlear im-plants notified the Food and Drug Administration(FDA) of 15 reports of postimplantation bacterialmeningitis in patients of any age who had receivedits implants. Early speculation implicated a posi-tioner, a component used by one manufacturer insome types of implants.

3

This small Silastic wedgeis inserted next to the implanted electrode to facili-tate transmission of the electrical signal by pushingthe electrode against the medial wall of the cochlea.The implants that included a positioner were vol-untarily recalled by the manufacturer in July 2002.

4

Since the initial reports were issued, however, theother two manufacturers of cochlear implants in theUnited States have notified the FDA of additionalcases of meningitis, principally in young children.

Because the magnitude of the problem was un-known, as were the factors, aside from the use of apositioner, that might increase the risk of meningi-tis among implant recipients, the Centers for Dis-ease Control and Prevention (CDC), the FDA, andthe health departments of 36 states, Chicago, NewYork City, and Washington, D.C., undertook an in-vestigation. The purpose was to establish the inci-dence of bacterial meningitis among children withcochlear implants and to identify specific types ofdevices, surgical factors, and patient characteristicsthat might be associated with an increased risk ofbacterial meningitis among recipients of cochlearimplants. The focus of the investigation was youngchildren, because they account for the majority ofknown cases and represent the population that willreceive most cochlear implants in the future.

study design and population

The investigation consisted of two parts: a cohortstudy to determine the incidence of bacterial men-ingitis among young children with cochlear im-plants and a nested casecontrol study to examinerisk factors for meningitis. The study population in-cluded all children who had received a cochlear im-plant in the United States between January 1, 1997,

and August 6, 2002, and who were less than six yearsold at the time of implantation. All three compa-nies that market cochlear implants in the UnitedStates provided warranty lists (estimated to be 95percent complete) of children who met these crite-ria. We combined these lists to define a study popu-lation of 4264 children.

Nineteen cases of bacterial meningitis were iden-tified from reports to the companies, the AdverseEvent Reporting System of the FDA, and the sur-veillance systems of the CDC and state and localhealth departments. Seven additional cases wereidentified by means of a brief questionnaire that wasmailed to the families of all children in the studypopulation. The rate of response to the question-naire was 57.3 percent and did not vary accordingto the year of implantation. The use of differentmethods of ascertainment and the attention paid tothis issue by the media lead us to believe that ourascertainment of cases was relatively complete.

definition and confirmation of cases

A patient with meningitis was defined as a child inthe study population in whom bacterial meningitisdeveloped after the placement of a cochlear implantand before September 15, 2002. The classificationof a case of meningitis as definite, probable, or pos-sible was based on an abstraction of the medicalrecords reviewed by two investigators using crite-ria developed for this study (Table 1).

cohort study

We estimated the incidence of meningitis in thestudy population using the number of cases of men-ingitis as the numerator and the number of person-years between implantation and the diagnosis ofmeningitis or September 15, 2002, as the denom-inator. We calculated stratified rates using varia-bles that were available for the entire cohort: type ofimplant, age at implantation, year of implantation,geographic region where implantation was per-formed, and time since implantation.

casecontrol study

The nested casecontrol study included all childrenwith cases of bacterial meningitis and a randomsample of 200 children who did not have postim-plantation meningitis. These 200 controls were se-lected with the use of a stratified approach so thatthe distributions of the year of implantation andof manufacturers were proportionate to those inthe total cohort, but controls were not individually

n

methods



n engl j med

349;5

www.nejm.org july

31, 2003

risk of bacterial meningitis in children with cochlear implants

437

matched to children with meningitis. Informationon risk factors in children with meningitis and con-trols was obtained from a parental interview that in-cluded questions about the causes of the childshearing loss and medical history; the abstraction ofthe medical records for the cochlear-implant surgeryand hospitalization for meningitis; and data on im-munizations and medical history obtained from theprimary care provider. All available information onmalformations of the inner ear was reviewed, andall such malformations were classified by an oto-laryngologist.

5

statistical analysis

The cohort data were summarized with the use ofthe stratum-specific incidence of meningitis. The in-cidence according to the time since implantationwas calculated for recipients of implants with a po-sitioner and recipients of implants without a posi-tioner, in an analysis that was limited to childrenwho received their implants after the devices witha positioner became available in 1999.

In the casecontrol analysis, we assessed thepossible association of postimplantation bacterialmeningitis with a number of factors in the medicalhistory: the occurrence of meningitis before implan-tation, the placement of tympanostomy tubes, theoccurrence of otitis media, the placement of a ven-triculoperitoneal shunt, other chronic medical con-ditions that could increase the risk of systemic in-fections, ossification of the cochlea, the vaccinationstatus for pneumoccoccal,

Haemophilus influenzae

type b (Hib), and meningococcal vaccines, and ra-diographic evidence of inner-ear malformations.The demographic variables evaluated included sex,age at implantation, year of implantation, race, andgeographic region.

We examined surgery-related factors, includingthe use of a positioner; the incomplete insertion ofthe electrode; a requirement for multiple attemptsin order to insert the electrode; the presence of acerebrospinal fluid leak (defined as a preexistingleak or an intraoperative leak or gusher resultingin rapid leakage of cerebrospinal fluid during sur-gery); the use of antibiotics before, during, or afterthe procedure; more than one implant; and signs ofmiddle-ear inflammation at the time of surgery. Useof a positioner was defined as the use of an implantmodel that included the electrode positioner as astandard component (AB-5100H and AB-5100H-11,Advanced Bionics) or the use of an implant for whichthe positioner was optional (AB-5100, Advanced

Bionics) in a child whose operative report indicatedthat the positioner was used. The postimplantationfactors evaluated were the presence of tympanos-tomy tubes at the time of or after cochlear-implantsurgery, the occurrence of otitis media after surgery,attendance at day care, and exposure to smoking inthe household.

6,7

Because inner-ear malformationsand cerebrospinal fluid leaks were strongly corre-lated with one another, new combination variableswere created. These consisted of both inner-ear mal-formation and cerebrospinal fluid leak, inner-earmalformation but no cerebrospinal fluid leak, andcerebrospinal fluid leak but no inner-ear malfor-mation.

We included in the analyses only the first implantor the first occurrence of meningitis after implanta-tion for children who had received multiple implantsor who had had multiple episodes of meningitis.Risk factors that had an association with any of thecategories of meningitis with a two-sided P value ofless than 0.1 on univariate analysis were includedin the multivariate logistic-regression model forall cases of meningitis.

identification of cases

A total of 29 episodes of postimplantation bacterialmeningitis were identified in 26 children; 3 children

results

* Abnormal cerebrospinal fluid was defined by two or more of the following: a cerebrospinal fluid protein level above 55 mg per deciliter, a cerebrospinal fluid white-cell count above 10 per cubic millimeter after adjustment for hemor-rhagic cerebrospinal fluid (with allowance for 1 white cell per 8000 red cells), and a cerebrospinal fluid glucose level of 40 mg per deciliter (2.2 mmol per liter) or less. Symptoms consistent with the presence of bacterial meningitis included two or more of the following: fever (temperature 38C), stiff neck or nuchal

rigidity, lethargy or altered mental status, and headache.

Table 1. Case Definitions for Definite, Probable, and Possible Bacterial Meningitis.*

Diagnosis Criteria

Definite Isolation of bacteria from cerebrospinal fluid or isolation of bac-teria from blood with abnormal cerebrospinal fluid and symp-toms consistent with the presence of bacterial meningitis

Probable Abnormal cerebrospinal fluid, symptoms consistent with bacte-rial meningitis, and evidence of bacteria in cerebrospinal fluid (on antigen testing, Grams staining, or polymerase chain reaction), or histopathological evidence of bacterial meningitis on autopsy

Possible Abnormal cerebrospinal fluid, symptoms consistent with bacte-rial meningitis, and no evidence suggesting a nonbacterial cause, or death after an unexplained illness with compatible symptoms



n engl j med

349;5

www.nejm.org july

31

,

2003

The

new england journal

of

medicine

438

Tabl

e 2.

Epi

sode

s of

Pos

timpl

anta

tion

Bac

teri

al M

enin

gitis

and

Pre

disp

osin

g Fa

ctor

s in

Chi

ldre

n W

ho R

ecei

ved

a C

ochl

ear

Impl

ant i

n th

e U

nite

d St

ates

bet

wee

n 19

97 a

nd 2

002.

*

Patie

nt

No.

Tim

e fr

omIm

plan

tatio

nto

Dia

gnos

is

Age

at

Tim

e of

Epis

ode

Sex

Cau

sal B

acte

ria

Impl

ant M

odel

at T

ime

ofEp

isod

e

Rad

iogr

aphi

cEv

iden

ce o

f Inn

er-E

arM

alfo

rmat

ion

Cer

ebro

spin

al F

luid

Leak

or

Gus

her

Prev

ious

Men

ingi

tisor

Ven

tric

ulop

eri-

tone

al S

hunt

Otit

is M

edia

at T

ime

of H

ospi

taliz

atio

n fo

r M

enin

gitis

His

tory

of

Rec

urre

ntO

titis

Med

ia

His

tory

of

Tym

pano

s-to

my

Tube

s

Sign

s of

Acu

te O

titis

Med

ia

11

day

2 yr

1 m

oM

Stre

ptoc

occu

s

pneu

mon

iae

CI2

4MN

one

Nei

ther

Nei

ther

Yes

No

No

24

days

16 m

oM

Aci

neto

bact

er

baum

anii

AB

-510

0HC

ochl

ear

dysp

lasi

aG

ushe

rN

eith

erYe

sN

oN

o

34

days

18 m

oM

A. b

aum

anii

AB

-510

0U

nkno

wn

Peri

oper

ativ

e le

akN

eith

erYe

sYe

sN

o

4Ep

isod

e 1

Epis

ode

24

days

2 m

o3

yr 7

mo

3 yr

9 m

oF F

Ente

roco

ccus

Unk

now

nC

I24M

CI2

4MM

ondi

nis

mal

form

atio

nM

ondi

nis

mal

form

atio

nG

ushe

rG

ushe

rN

eith

erM

enin

gitis

Yes

Yes

Yes

Yes

No

No

510

day

s23

mo

F

Esch

eric

hia

coli

AB

-510

0H-1

1N

one

Peri

oper

ativ

e le

akN

eith

erYe

sYe

sN

o

613

day

s2

yrF

S. p

neum

onia

e

AB

-510

0H-1

1N

one

Nei

ther

Nei

ther

No

No

No

714

day

s5

yr 3

mo

M

S. p

neum

onia

e

AB

-510

0H-1

1En

larg

ed v

estib

ular

aq

uedu

ctG

ushe

rN

eith

erYe

sYe

sU

nkno

wn

820

day

s2

yr 1

mo

M

S. p

neum

onia

e

CI2

4RST

Non

eN

eith

erB

oth

Yes

No

No

929

day

s2

yr 4

mo

F

Hae

mop

hilu

s in

fluen

zae

type

bC

I22M

Mon

dini

s m

alfo

rmat

ion

Preo

pera

tive

leak

Men

ingi

tisYe

sYe

sN

o

101.

5 m

o3

yrF

S. p

neum

onia

e

AB

-510

0H-1

1U

nkno

wn

Nei

ther

Nei

ther

Yes

Yes

No

111.

5 m

o3

yr 1

0 m

oM

S. p

neum

onia

e

AB

-510

0HN

one

Nei

ther

Nei

ther

No

No

Yes

12Ep

isod

e 1

Epis

ode

21.

5 m

o9

mo

3 yr

8 m

o4

yr 3

mo

M M

S. p

neum

onia

eH

. inf

luen

zae,

nont

ypea

ble

AB

-510

0HA

B-5

100H

Unk

now

nU

nkno

wn

Unk

now

nU

nkno

wn

Nei

ther

Men

ingi

tisYe

sYe

sYe

sYe

sYe

sYe

s

132

mo

4 yr

3 m

oF

S. p

neum

onia

e

AB

-510

0HC

omm

on c

avity

Nei

ther

Nei

ther

No

No

Yes

142

mo

4 yr

M

S. p

neum

onia

e

AB

-510

0HN

one

Nei

ther

Nei

ther

Yes

Yes

Unk

now

n

15Ep

isod

e 1

Epis

ode

23

mo

15 m

o2

yr 2

mo

3 yr

3 m

oM M

S. p

neum

onia

e

Unk

now

nA

B-5

100H

AB

-510

0HN

one

Non

eN

eith

erN

eith

erN

eith

erM

enin

gitis

Yes

Yes

No

No

Unk

now

nU

nkno

wn

169

mo

3 yr

6 m

oM

H. i

nflu

enza

e,

no

ntyp

eabl

eA

B-5

100H

-11

Unk

now

nU

nkno

wn

Nei

ther

Yes

No

Yes



n engl j med

349;5

www.nejm.org july

31, 2003


439

*Im

plan

t mod

els

AB

-510

0H a

nd A

B-5

100H

-11

(Adv

ance

d B

ioni

cs) i

nclu

de a

pos

ition

er; t

he p

ositi

oner

is o

ptio

nal i

n m

odel

AB

-510

0 (A

dvan

ced

Bio

nics

). P

atie

nt 3

did

not

hav

e a

posi

tione

r pl

aced

; Pat

ient

23

did

have

a p

ositi

oner

pla

ced.

Mod

els

CI2

4M, C

I24R

ST, a

nd C

I22M

are

man

ufac

ture

d by

Coc

hlea

r Cor

pora

tion,

and

mod

el C

40+H

S is

man

ufac

ture

d by

MED

-EL.

Mon

dini

s

mal

form

atio

n is

def

ined

as

a sm

all c

ochl

ea w

ith a

n in

com

plet

e or

no

inte

rsca

lar

sept

um, w

ith a

nor

mal

or

mal

form

ed v

estib

ule

and

sem

icir

cula

r ca

nals

.

Sign

s of

acu

te o

titis

med

ia o

n ad

mis

sion

incl

uded

ear

dra

inag

e, a

bul

ging

tym

pani

c m

embr

ane,

or

the

rece

ipt o

f ant

ibio

tics

for

acut

e ot

itis

med

ia.

In

this

chi

ld, t

he im

plan

t was

rem

oved

dur

ing

the

hosp

italiz

atio

n fo

r m

enin

gitis

.

The

child

die

d of

men

ingi

tis.

M

enin

gitis

occ

urre

d af

ter

a se

cond

impl

ant w

as p

lace

d; b

oth

impl

ants

wer

e m

odel

AB

-510

0H.

1710

mo

6 yr

3 m

oM

S. p

neum

onia

e

CI2

4MC

omm

on c

avity

Nei

ther

Men

ingi

tisN

oN

oN

o

18

10 m

o6

yr 6

mo

F

S. p

neum

onia

e

C40

+HS

Mon

dini

s m

alfo

rmat

ion

Gus

her

Vent

ricu

lope

rito

neal

shun

tN

oN

oN

o

1910

mo

2 yr

2 m

oF

H. i

nflu

enza

e,

nont

ypea

ble

AB

-510

0H-1

1N

one

Nei

ther

Nei

ther

No

No

Yes

20

12 m

o2

yr 1

1 m

oM

S. p

neum

onia

e

AB

-510

0HN

one

Nei

ther

Nei

ther

No

No

Yes

2114

mo

3 yr

4 m

oM

H. i

nflu

enza

e

type

bA

B-5

100H

-11

Non

eN

eith

erN

eith

erN

oN

oU

nkno

wn

2214

mo

6 yr

8 m

oM

Unk

now

nA

B-5

100H

Non

eN

eith

erN

eith

erN

oN

oYe

s

2317

mo

3 yr

2 m

oM

S. p

neum

onia

e

AB

-510

0N

one

Nei

ther

Nei

ther

Yes

No

No

2419

mo

3 yr

1 m

oF

S. p

neum

onia

e

CI2

4RST

Mon

dini

s m

alfo

rmat

ion

Gus

her

Nei

ther

Yes

No

Unk

now

n

2524

mo

5 yr

2 m

oM

Unk

now

nA

B-5

100H

U

nkno

wn

Nei

ther

Nei

ther

No

No

No

2636

mo

5 yr

11

mo

FU

nkno

wn

CI2

2MN

one

Nei

ther

Nei

ther

Yes

No

No



n engl j med

349;5

www.nejm.org july

31

,

2003

The

new england journal

of

medicine

440

had 2 episodes of bacterial meningitis each. Of the29 episodes, 24 met the criteria for definite bacteri-al meningitis, and 5 met the criteria for possible bac-terial meningitis. The cases of possible meningitiswere associated with a cerebrospinal fluid white-cellcount of 300 to 6115 per cubic millimeter, and inall but one case, there was a predominance of neu-trophils. Nine episodes of bacterial meningitis wereperioperative (occurring 30 days after surgery); 20episodes were sporadic (occurring >30 days aftersurgery). Of the 20 sporadic episodes, 8 (40 percent)were in patients who had evidence of acute otitismedia at presentation. One child died, and three hadthe implant removed during their hospitalizationfor meningitis.

Streptococcus pneumoniae

accounted for 15 of the24 episodes of bacterial meningitis with a knowncause (62 percent) (Table 2). In 11 of these 15 cases(73 percent), meningitis was associated with bac-teremia, and in 1 of these cases, pneumonia waspresent. Two children (Patient 7, a five-year-old boy,and Patient 10, a three-year-old girl) (Table 2) hadreceived one dose of 7-valent pneumococcal con-jugate vaccine (PCV7); the serotypes that caused theillness were unknown. One child (Patient 24, a three-year-old girl) had received two doses of PCV7 andhad

S. pneumoniae

meningitis caused by serotype 10A a type that is not included in PCV7. No patientsreceived a 23-valent pneumococcal polysaccharidevaccine.

Two children had meningitis caused by Hib; itis unknown whether the serotype of these isolateswas reconfirmed.

8

One of these children (Patient21, a three-year-old boy) had no clinically significantmedical history and was fully vaccinated againstHib. The second (Patient 9, a two-year-old girl) hadreceived three of four recommended doses of vac-cine; this child had six episodes of meningitis be-fore receiving the cochlear implant, two of whichwere caused by Hib.

cohort analysis

The median follow-up period for children in thecohort in whom postimplantation meningitis didnot develop was 29 months (interquartile range,15 to 45). The incidence of all cases of meningitisin the cohort was 239.3 per 100,000 person-years(95 percent confidence interval, 156.4 to 350.6).Meningitis caused by

S. pneumoniae

accounted for138.2 cases per 100,000 person-years (95 percentconfidence interval, 77.4 to 227.9). Perioperativemeningitis occurred at a rate of 2.1 cases per 1000

procedures. There were differences in the observedrate according to whether the implant type includ-ed a positioner, the type of implant, and the year ofimplantation (Table 3). When the analysis was re-stricted to children who had received an implantbetween 1999 and 2002, the observed incidence ofbacterial meningitis among children with the AB-5100H or AB-5100H-11 implant was higher for theduration of follow-up than the incidence amongthose with implant models that did not include a po-sitioner (Table 4 and Supplementary Appendix 1[available with the full text of this article at http://www.nejm.org]), although the 95 percent confi-dence intervals for the incidence in the two groupsoverlapped.

casecontrol analysis

We reviewed the implant-surgery records of 24 ofthe patients with bacterial meningitis (92 percent)and 186 of 200 controls (93 percent). One controlwas excluded from the analysis because she wasmore than six years old at the time of surgery. Inter-views were conducted with parents of all 26 patientswith meningitis and with parents of 161 of the 199remaining controls (81 percent). We received infor-mation from primary care providers for 20 of thepatients with meningitis (77 percent) and 159 of thecontrols (80 percent).

Univariate analyses showed possible associa-tions between several risk factors and one or moresubcategories of meningitis (Table 5). These riskfactors included a history of placement of a ventric-uloperitoneal shunt; a history of otitis media beforeimplantation; the presence of inner-ear malforma-tions combined with a cerebrospinal fluid leak anda cerebrospinal fluid leak alone; the use of a posi-tioner; incomplete insertion of the electrode; signsof middle-ear inflammation at the time of implanta-tion; and exposure to smoking in the household.The risk factors for sporadic meningitis were dif-ferent from those for perioperative meningitis: theuse of a positioner was more strongly associatedwith the occurrence of sporadic meningitis, and in-complete insertion of the electrode and a cerebro-spinal fluid leak without an inner-ear malformationwere associated with the occurrence of periopera-tive meningitis but not with the occurrence of spo-radic meningitis.

On multivariate modeling, the use of a position-er was significantly associated with the occurrenceof meningitis (odds ratio, 4.5; 95 percent confidenceinterval, 1.3 to 17.9), as was inner-ear malforma-



n engl j med

349;5

www.nejm.org july

31, 2003


441

* Perioperative meningitis was defined as meningitis with an onset within 30 days after surgery, and sporadic meningitis as that with an onset more than 30 days after surgery. CI denotes confidence interval.

Data are for models AB-5100H and AB-5100H-11 (Advanced Bionics); data on children who received model 5100H (Ad-vanced Bionics), for which the use of a positioner was optional, were included in the group without a positioner, because information on the use or nonuse of the positioner was not available for children who were not included in the casecontrol investigation.

No cases were reported among children who received implant models CI24R(CS), who accounted for 1300 person-years, C40+H, who accounted for 301 person-years, or C40+HGB, who accounted for 10 person-years.

A test for trend showed a significant increase over time: chi-square statistic for trend=19.9, with 1 df; P

n engl j med

349;5

www.nejm.org july

31

,

2003

The

new england journal

of

medicine

442

tion with a cerebrospinal fluid leak (odds ratio, 9.3;95 percent confidence interval, 1.2 to 94.5). Otherfactors were associated with similar adjusted andcrude odds ratios, suggesting that they might beindependently associated with the occurrence ofbacterial meningitis; however, the 95 percent confi-dence intervals for these odds ratios were wide.When the analysis was limited to implants that wereplaced between 1999 and 2002, the adjusted oddsratio for meningitis associated with the use of a po-sitioner increased to 5.6 (95 percent confidence in-terval, 1.3 to 33.5). When the analysis was limited topatients with definite cases of meningitis, the adjust-ed odds ratio for meningitis associated with the useof a positioner was 5.3 (95 percent confidence in-terval, 1.4 to 25.1).

The incidence of bacterial meningitis caused by

S. pneumoniae

in our study population of childrenwho received cochlear implants when they were lessthan six years of age was more than 30 times the in-cidence of pneumococcal meningitis among chil-dren in that age group in the general U.S. popula-

tion in 2000 covered by the Active Bacterial CoreSurveillance program of the CDC

9

(4.0 cases per100,000 children [unpublished data]). Even amongchildren who received an implant without a posi-tioner, the incidence of

S. pneumoniae

meningitis was16 times as high as the background rate in the gen-eral population for this age group. Therefore, thehigher incidence of meningitis in this cohort wasnot fully explained by the use of positioners. Ourstudy was not designed to determine the incidenceof meningitis among children with severe-to-pro-found hearing loss who had not received a cochlearimplant, and no data on this incidence are available.The incidence might be higher than that amongchildren in the same age group in the general pop-ulation because of underlying risk factors in suchchildren. For example, among the controls in ourstudy, 8.5 percent had radiographic evidence of in-ner-ear malformations, 3 percent had a cerebro-spinal fluid leak, and 10.1 percent had a history ofmeningitis.

The observed incidence of meningitis increasedover the six-year study period. Although this find-ing could reflect underascertainment in earlier years,it is also possible that there has been a true increase

discussion

* This analysis was limited to children who received an implant after the devices including a positioner became available in 1999; three children in the cohort who had meningitis in 1997 or 1998 were excluded from the analysis. Confidence in-tervals for incidence rates and incidence-rate ratios were calculated according to the exact method under the assumption that the number of cases is a Poisson random variable.

Data are for models AB-5100H and AB-5100H-11 (Advanced Bionics); data on children who received model 5100H (Ad-vanced Bionics), for which the use of a positioner was optional, were included in the group without a positioner, because information on the use or nonuse of the positioner was not available for children who were not included in the casecontrol investigation.

In instances in which the incidence-rate ratio was undefined, the lower limit for the 95 percent confidence interval was

calculated according to exact methods.

Table 4. Incidence of Bacterial Meningitis among Children Who Received Implant Models Including a Positioner as Compared with Those Who Received Other Models, 19992002.*

Months ofFollow-up Implant Models with a Positioner Other Implant Models

Incidence-Rate Ratio(95% CI)

No. ofCases

Incidence/100,000 Person-Yr

(95% CI)No. ofCases

Incidence/100,000 Person-Yr

(95% CI)

1.0 4 5818 (158014,841) 3 1528 (3164472) 3.8 (0.625.9)

1.13.0 5 3676 (11948579) 0 0 (0948) Undefined (2.6infinity)

3.16.0 1 526 (132932) 0 0 (0673) Undefined (0.1infinity)

6.112.0 3 922 (1902697) 2 205 (25739) 4.5 (0.554.0)

12.118.0 2 870 (1063140) 1 121 (3675) 7.2 (0.4423.8)

18.124.0 1 701 (183896) 1 150 (4837) 4.7 (0.1365.6)



n engl j med

349;5

www.nejm.org july

31, 2003


443

in the number of cases in recent years. In part, suchan increase might be attributable to the introductionof the device with the positioner in 1999. In addition,there has been a trend toward earlier placement ofimplants, because improved development of speechand language is associated with earlier implanta-tion,

10-12

and younger children are at higher risk formeningitis.

9

The company that marketed cochlear implantswith a positioner voluntarily recalled those productsin July 2002.

4

In our analysis in the cohort study,the use of the implant models with a positioner wasimplicated as an important risk factor for meningi-tis. In the casecontrol study, the presence of a po-sitioner was significantly associated with the riskof sporadic meningitis but not with the risk of peri-operative meningitis. The attributable fraction

13

ofall cases of meningitis was 50.9 percent (60.7 per-cent for cases in children who received their implantbetween 1999 and 2002), suggesting that about halfthe cases of bacterial meningitis in implant recip-ients might be attributable to the use of the posi-tioner, if a causal relation is assumed.

The reasons for the association between the useof the positioner and the occurrence of meningitis

are unclear. Possible explanations include the larg-er cochleostomy required for the insertion of boththe electrode and the positioner, the presence of anadditional foreign body in the inner ear, trauma tothe fragile anatomy of the inner ear caused by thewedging of the positioner into place,

14

deficient for-mation of a fibrous-tissue seal at the site of the coch-leostomy,

3

and a combination of these factors. Afurther complication for the comparison of types ofimplants is that the choice of a particular implantcould be based on certain characteristics of the pa-tient. For example, product labeling for the implantswith positioners states that their use is contrain-dicated in people with ossification of the cochlea.

Overall, the risk of meningitis decreased rapid-ly after the perioperative period. The incidence ofmeningitis among children with an implant thatincluded a positioner remained higher than thatamong children with other types of implants for theduration of follow-up. However, only a few caseswere identified during each period, and such smallnumbers result in imprecise estimates; continuedfollow-up is needed in order to estimate the inci-dence of meningitis more than 24 months after im-plantation.

* Perioperative meningitis was defined as meningitis with an onset within 30 days after surgery, and sporadic meningitis as that with an onset more than 30 days after surgery. There were 199 controls included in all analyses. Univariate results included in this table are limited to those that were associated with any subcategory of meningitis with a two-sided P value of less than 0.1.

Fishers exact methods were not used, because the expected numbers in the contingency tables were greater than 5. There were no patients with a malformation only. There were no patients with a leak only.

All patients had otitis media before implantation.

Table 5. Risk Factors for Meningitis.*

Variable All Meningitis Univariate Analysis

MultivariateAnalysis(N=26)

UnivariateAnalysis(N=26)

PerioperativeMeningitis

(N=9)

SporadicMeningitis

(N=17)

S. pneumoniae

Meningitis(N=15)

odds ratio (95 percent confidence interval)

Implant with a positioner 4.5 (1.317.9) 5.6 (2.413.4) 2.4 (0.511.4) 9.7 (3.031.1) 6.0 (1.723.1)

Inner-ear malformation and cerebrospinal fluid leakBothMalformation onlyLeak only

9.3 (1.294.5)2.0 (0.115.2)6.2 (0.479.5)

18.4 (3.4120.3)1.0 (0.14.8)4.9 (0.444.9)

39.2 (4.8320.2)

14.0 (1.0138.5)

8.9 (0.784.3)1.8 (0.29.4)

14.7 (1.7119.8)1.8 (0.29.4)

Incomplete insertion of electrode 3.2 (0.519.7) 3.2 (0.810.7) 7.2 (1.040.9) 1.7 (0.28.8) 3.0 (0.513.1)

History of ventriculoperitoneal-shunt placement 7.1 (0.3126.1) 3.9 (0.328.8) 5.9 (0.168.2) 2.9 (0.131.9) 7.2 (0.655.1)

Exposure to smoking in household 2.2 (0.67.8) 2.2 (0.95.2) 3.8 (0.819.8) 1.7 (0.55.2) 2.0 (0.66.8)

Otitis media before implantation 2.7 (0.3127.1) 3.7 (0.816.3) 2.3 (0.521.5)

Signs of middle-ear inflammation at implantation 1.4 (0.28.4) 2.3 (0.67.3) 4.9 (0.727.0) 1.3 (0.16.2) 3.3 (0.712.7)



n engl j med

349;5

www.nejm.org july

31

,

2003

The

new england journal

of

medicine

444

In the casecontrol study, the combination of ra-diographic evidence of an inner-ear malformationand a cerebrospinal fluid leak was associated withan increased risk of all subtypes of meningitis, andthe presence of a cerebrospinal fluid leak alone wasassociated with an increased risk of perioperativemeningitis. Children with cerebrospinal fluid leakshave been targeted for pneumococcal vaccination,because such leaks represent a known risk factorfor pneumococcal meningitis.

15,16

Our analysis didnot identify inner-ear malformations alone as a riskfactor for meningitis, although case reports havenoted such an association, particularly with Mondi-nis malformation.

17-21

Our ability to assess the protective effect of pneu-mococcal vaccinations was limited because of thesmall number of vaccinated children; pneumococ-cal conjugate vaccine, now universally recommend-ed for children less than two years of age, was notlicensed until 2000 late in our study period. As-sessment of Hib vaccination was limited by the factthat almost all children were vaccinated before re-ceiving a cochlear implant, as is universally recom-mended.

22

We found that risk factors and infecting speciesof bacteria in cases of perioperative meningitis dif-fered somewhat from those in cases of sporadicmeningitis. Although it was difficult to interpretthese differences because of the small numbers, theysuggest a difference in the pathophysiology of in-fection. In perioperative cases, bacteria may be in-troduced at the time of the surgery. In sporadic cas-es, because of the association with cerebrospinalfluid leaks and inner-ear malformations, because40 percent of the patients with sporadic cases hadsigns of otitis media at presentation, and becausethe main pathogens (

S. pneumoniae

and

H. influenzae,

with

Neisseria meningitidis

notably absent) are con-sistent with those that cause otitis media,

23,24

wesuspect that bacteria often entered through the mid-dle and inner ear.

The investigation included cases of meningitisoccurring before September 15, 2002; we have sub-sequently received reports of postimplantationmeningitis after this date in six children in the studypopulation. The cases in these six children occurredfrom 6 to 41 months after implantation, and thecausative pathogen in all six was

S. pneumoniae.

Fiveof these children received cochlear implants thatincluded a positioner (AB-5100H in two and AB-5100H-11 in three), and one received a model C40+HGB implant (MED-EL).

Our findings have important implications forthe prevention of meningitis in children with alltypes of cochlear implants. Providers should reviewthe vaccination records of their patients who havereceived or are candidates for receiving cochlearimplants in order to ensure that they have receivedpneumococcal vaccinations according to the age-appropriate schedules for high-risk children

25

(seeSupplementary Appendix 2, available with the fulltext of this article at http://www.nejm.org) andthat they have received age-appropriate Hib vacci-nations. Parents should contact their childs healthcare provider to ensure that all recommended vac-cines have been received.

22

Children should receivevaccinations two or more weeks before surgery.

Both parents and health care providers shouldremain vigilant during the months after surgery forpossible signs and symptoms of meningitis. Sur-geons who place cochlear implants should informthe childs health care provider and parents aboutany inner-ear malformations or other findings dur-ing surgery (e.g., cerebrospinal fluid leak or incom-plete insertion) that could increase the risk of men-ingitis. Prompt diagnosis and treatment of acuteotitis media might prevent the spread of infectionthrough any possible weakness of the fibrous-tissueseal at the cochleostomy site.

Removal of an existing implant containing a po-sitioner might not be beneficial. Because the mech-anism by which the positioner increases the riskof meningitis is unknown, it is unclear whether ex-plantation of the device would lower the risk of men-ingitis; in addition, explantation would place thechild at risk for operative complications, includingperioperative meningitis. Parents and health careproviders should also consider the benefits thatcochlear implants provide for children with severe-to-profound hearing loss. We hope that with appro-priate use of pneumococcal vaccines and no furtheruse of implants with a positioner, the incidence ofbacterial meningitis among children with cochlearimplants will decrease. However, the continued oc-currence of bacterial meningitis underscores theneed for parents and health care providers of all chil-dren with any type of cochlear implant to take ap-propriate precautions.

We are indebted to P. Blake, J. Block, D. Blythe, E. Bresnitz, N.Calonge, M. Crutcher, R. Danila, J. Davies Cole, J. Davis, A. DeMaria,P. Effler, J. Engel, S. Englender, J. Gibson, M. Goldoft, G. Huhn, T.Jones, M. Kohn, C. Lohff, P. Lurie, M. Layton, J. Marr, C. McRill, B.Mosley, P. Mshar, B. Paul, D. Perrotta, G. Pezzino, R. Ratard, R.Rolfs, T. Safranek, E. Simoes, F. Smith, S. Stonecipher, R. Teclaw,D. Vugia, S. Wiersma, and C. Woernle for their assistance; to S.Moore for her administrative support of this investigation; to J.



n engl j med

349;5

www.nejm.org july

31, 2003


445

Moran, A. Schuchat, B. Plikaytis, J. Horan, and C. Moore; and to themembers of the Cochlear Implant Team: K. Abe, A. Adija, F. Arguel-lo, R. Avchen, B. Culpepper, V. Dato, Y. Dominique, M. Dott, S. Epps,J. Farrell, L. Fassett, A. Flores, M. Gaffney, D. Green, K. Griffith, T.Gross, J. Holstrum, L. Jackson, C. Johnson, T. Kerr, K. Kilker, C. Kios-ki, J. Kucik, J. Lai, B. Lash, A. Lee, N. Lee, V. Lee, P. Ludwinski, J. Ma-

lone, K. Mark, J. Mazurek, T. McCauley, A. Mersereau, D. Miller, J.Montero, J. Mosely Hayes, K. Patterson, K. Piper, A. Preiss, J. Rain-bow, B. Ramsey, D. Ramsey, G. Reddy, A. Roach, M. Romney, M. Ros-er, S. Ruuska, K. Stefonek, M. Sotir, P. Thomas, J. Thorell, D. Thor-oughman, M. Tobin-dAngelo, B. Tynan, K. van Naarden Braun, A.Vicari, J. Walls, S. Whitehead, J. Williams, L. Yeung.

references

1.

Cochlear implants. Bethesda, Md.: Na-tional Institute on Deafness and Other Com-munication Disorders, 2003. (Accessed July1, 2003, at http://www.nidcd.nih.gov/health/hearing/coch.asp.)

2.

NIH Consensus Conference. Cochlearimplants in adults and children. JAMA 1995;274:1955-61.

3.

Arnold W, Bredberg G, Gstottner W, etal. Meningitis following cochlear implanta-tion: pathomechanisms, clinical symptoms,conservative and surgical treatments. ORL JOtorhinolaryngol Relat Spec 2002;64:382-9.

4.

Advanced Bionics Corporation. Electrodepositioner. (Accessed July 8, 2003, at http://www.bionicear.com/vaccine/menin_implants2.html.)

5.

Jackler RK, Luxford WM, House WF.Congenital malformations of the inner ear:a classification based on embryogenesis. La-ryngoscope 1987;97:Suppl 40:2-14.

6.

Iles K, Poplawski NK, Couper RT. Passiveexposure to tobacco smoke and bacterialmeningitis in children. J Paediatr Child Health2001;37:388-91.

7.

Nuorti JP, Butler JC, Farley MM, et al.Cigarette smoking and invasive pneumococ-cal disease. N Engl J Med 2000;342:681-9.

8.

Serotyping discrepancies in Haemophi-lus influenzae type b disease United States,19981999. MMWR Morb Mortal Wkly Rep2002;51:706-7.

9.

Schuchat A, Robinson K, Wenger JD, et

al. Bacterial meningitis in the United Statesin 1995. N Engl J Med 1997;337:970-6.

10.

Hammes DM, Novak MA, Rotz LA,Willis M, Edmondson DM, Thomas JF. Ear-ly identification and cochlear implantation:critical factors for spoken language devel-opment. Ann Otol Rhinol Laryngol Suppl2002;189:74-8.

11.

Govaerts PJ, De Beukelaer C, Daemers K,et al. Outcome of cochlear implantation atdifferent ages from 0 to 6 years. Otol Neuro-tol 2002;23:885-90.

12.

Sharma A, Dorman And MF, Spahr AJ.A sensitive period for the development ofthe central auditory system in children withcochlear implants: implications for age ofimplantation. Ear Hear 2002;23:532-9.

13.

Greenland S. Attributable fraction esti-mation. In: Rothman KJ, Greenland S, eds.Modern epidemiology. 2nd ed. Philadelphia:LippincottRaven, 1998:295-7.

14.

Richter B, Aschendorff A, Lohnstein P,Husstedt H, Nagursky H, Laszig R. Clarion1.2 standard electrode array with partialspace-filling positioner: radiological andhistological evaluation in human temporalbones. J Laryngol Otol 2002;116:507-13.

15.

Preventing pneumococcal disease amonginfants and young children: recommenda-tions of the Advisory Committee on Immu-nization Practices (ACIP). MMWR RecommRep 2000;49 (RR-9):1-35.

16.

Kline MW. Review of recurrent bacteri-

al meningitis. Pediatr Infect Dis J 1989;8:630-4.

17.

Ohlms LA, Edwards MS, Mason EO, Ig-arashi M, Alford BR, Smith RJ. Recurrentmeningitis and Mondini dysplasia. Arch Oto-laryngol Head Neck Surg 1990;116:608-12.18. Herther C, Schindler RA. Mondinis dys-plasia with recurrent meningitis. Laryngo-scope 1985;95:655-8.19. Hayashi N, Kino M, Nobori U, et al. Re-current bacterial meningitis: secondary tomalformation of the inner ear. Clin Pediatr(Phila) 1989;28:139-41.20. Phelps PD, King A, Michaels L. Cochleardysplasia and meningitis. Am J Otol 1994;15:551-7.21. Valmari P, Palva A. Recurrent meningitisdue to pneumococci and non-typable Hae-mophilus influenzae in a child with a Mon-dini malformation. Infection 1986;14:36-7.22. Recommended childhood and adoles-cent immunization schedule: United States,2003. MMWR Morb Mortal Wkly Rep 2003;52:Q1-Q4. [Erratum, MMWR Morb MortalWkly Rep 2003;52:191.]23. Berman S. Otitis media in children.N Engl J Med 1995;332:1560-5.24. Klein JO. Otitis media. Clin Infect Dis1994;19:823-33.25. Pneumococcal vaccination for cochlearimplant recipients. MMWR Morb MortalWkly Rep 2002;51:931.Copyright 2003 Massachusetts Medical Society.

images in clinical medicine

The Journal welcomes consideration of new submissions for Images in Clinical Medicine. Instructions for authors and procedures for submissions can be found on the Journal s Web site at http://www.nejm.org. At the discretion of the editor, imag-es that are accepted for publication may appear in the print version of the Journal, the electronic version, or both.



Documents

Jurnal Meningitis Anak