Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
PREDICTION of PATIENT OUTCOMES after TRAUMATIC SPINAL CORD INJURY using ACUTE
CLINICAL and RADIOLOGICAL VARIABLES
by
Jefferson R. Wilson
A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy
Institute of Medical Sciences University of Toronto
© Copyright by Jefferson R. Wilson (2013)
ii
Prediction of Patient Outcomes after Traumatic Spinal Cord Injury
using Acute Clinical and Radiological Variables
Jefferson R. Wilson
Doctor of Philosophy
Institute of Medical Sciences
University of Toronto
2013
Abstract
There is a pressing unmet need, in the acute setting after traumatic spinal cord injury (SCI), to
predict both short and long-term clinical outcomes for individuals affected. Possessing this
ability would not only facilitate improved communication and treatment planning in the clinical
realm, but also would allow for enhanced study of SCI patients in the research realm. Presently,
there is little available to scientifically guide clinicians and researchers alike with respect to this
topic. In this thesis, it is hypothesized that clinical outcomes can accurately be predicted after
SCI based on acute patient, injury and radiological features. In order to investigate this
hypothesis, a series of investigations have been undertaken, using a combination of two large
prospective SCI datasets, to explore the impact of acute clinical and radiological factors on long-
term functional outcome, long-term neurological outcome and in-hospital complication
occurrence. First, after performing a comprehensive systematic review of existing literature, a
clinico-radiographic model was created and internally validated using a combination of pre-
specified clinical and magnetic resonance imaging (MRI) variables obtained in the first 3 days
after injury, to predict functional outcome at 1 year. Second, the acute radiological finding of
facet dislocation (FD) was evaluated and found to predict reduced motor neurological recovery at
1 year, even after adjusting for individuals’ baseline degree of injury severity. Third, a
iii
combination of clinical variables predicting in-hospital complication development after cervical
SCI were identified and used to develop a model to predict the occurrence of these complication
events. Finally, the impact of age as an effect modifier governing the relationship between acute
neurologic status and long-term functional outcome was evaluated. This analysis demonstrated
older age to have the greatest negative impact on functional outcome in patients with lesions of
intermediate severity. Although future validation studies are required, the findings presented in
this thesis have provided new insights into relationships between acute predictor and outcome
variables in the context of traumatic SCI. Further, these analyses have demonstrated the
feasibility, and potential advantages, of using a combination of acute clinical and radiological
variables, to predict outcome after SCI.
iv
Acknowledgments
My sincerest gratitude goes to my supervisor Dr. Michael Fehlings. It was five years ago, while a
junior resident on your spine service at Toronto Western Hospital, that I was first inspired to
formally pursue spinal cord injury related research. Since then, you have been an unyielding
source of encouragement, advice and wisdom. Thank you for your continued mentorship.
I would also like to thank the members of my post-graduate advisory committee including Dr.
Aileen Davis, Dr. Abhaya Kulkarni and Dr. Alex Kiss; your contributions to, and careful critique
of this work have truly been invaluable.
This work would also not have been possible without the contributions of Dr. Robert Grossman
and Dr. Ralph Frankowski from the North American Clinical Trials Network for the Treatment
of Spinal Cord Injury, who were, and continue to be, outstanding collaborators.
I would also like to acknowledge and thank all of my friends and mentors from the Division of
Neurosurgery at the University of Toronto who have provided encouragement, reassurance and
camaraderie throughout my years away from clinical training. A special thanks to Dr. James
Rutka, Dr. Charles Tator, Dr. Andres Lozano, Dr. Abhaya Kulkarni, Dr. David Cadotte, Dr. Nir
Lipsman and Dr. Mike Ellis.
I am very grateful to the following organizations who contributed financially toward the
successful completion of this work: The Christopher and Dana Reeve Foundation, The Cervical
Spine Research Society, The United States Department of Defense and the University of Toronto
Surgeon Scientist Program.
Finally, I owe a great debt of gratitude to my family for their never ending love and support.
-JRW 2013
v
Table of Contents
Acknowledgments .......................................................................................................................... iv
Table of Contents ............................................................................................................................ v
List of Tables .................................................................................................................................. x
List of Figures ............................................................................................................................... xii
List of Appendices ....................................................................................................................... xiii
List of Abbreviations ................................................................................................................... xiv
Preamble ......................................................................................................................................... 1
Thesis Structure .............................................................................................................................. 2
1 Chapter 1: Overview of Current Epidemiology, Pathophysiology, Treatment and
Assessment Standards for Traumatic Spinal Cord Injury .......................................................... 3
1.1 Introduction ......................................................................................................................... 4
1.2 Epidemiology and Cost ....................................................................................................... 4
1.3 Pathophysiology of SCI ...................................................................................................... 6
1.4 Current Approach to Clinical and Radiological Assessment .............................................. 7
1.4.1 Neurological Assessment of SCI patients ............................................................... 7
1.4.2 Radiological Assessment ........................................................................................ 9
1.5 Time periods after SCI ...................................................................................................... 10
1.6 Outcome Assessment after SCI ........................................................................................ 10
1.7 Treatments Available ........................................................................................................ 13
1.7.1 Supportive/Critical Care Therapy for SCI ............................................................ 13
1.7.2 Pharmacologic Therapy for SCI ........................................................................... 14
1.7.3 Surgery for SCI ..................................................................................................... 16
1.7.4 Cellular based Therapy ......................................................................................... 17
2 Chapter 2: Clinical and Radiological Predictors of Neurological outcome, Functional
outcome and Complication development after Traumatic Spinal Cord Injury: A Systematic
Review ...................................................................................................................................... 18
vi
2.1 Abstract ............................................................................................................................. 19
2.2 Introduction ....................................................................................................................... 20
2.3 Common Systematic Review Methods ............................................................................. 20
2.4 Literature Review Section 1: The impact of clinical variables on outcome after SCI...... 23
2.4.1 Results ................................................................................................................... 23
2.4.2 Discussion ............................................................................................................. 40
2.4.3 Summary: The impact of clinical variables on outcome after SCI ....................... 47
2.5 Literature Review Section 2: The impact of radiological variables on outcome after
SCI .................................................................................................................................... 48
2.5.1 Results ................................................................................................................... 48
2.5.2 Discussion ............................................................................................................. 54
2.5.3 Summary: The impact of radiological variables on clinical outcomes after SCI . 59
2.6 Limitations of systematic review ...................................................................................... 60
2.7 Conclusion ........................................................................................................................ 61
2.8 Rationale, Overarching Hypothesis and Specific Aims .................................................... 62
2.8.1 Overarching Hypothesis ........................................................................................ 64
2.8.2 Specific Aims ........................................................................................................ 65
3 Chapter 3: Description of Datasets and Approach to Missing Data ........................................ 66
3.1 Introduction ....................................................................................................................... 67
3.2 Surgical Timing in Acute Spinal Cord Injury Study (STASCIS) Database ..................... 67
3.3 North American Clinical Trials Network for SCI (NACTN) Database............................ 67
3.4 Combined/Harmonized Dataset ........................................................................................ 68
3.5 Approach to missing data within the combined dataset .................................................... 70
4 Chapter 4: A Clinical Prediction Model for Long-Term Functional Outcome after
Traumatic Spinal Cord Injury Based on Acute Clinical and Imaging Factors......................... 72
4.1 Abstract ............................................................................................................................. 73
4.2 Introduction ....................................................................................................................... 74
vii
4.3 Methods: ........................................................................................................................... 74
4.3.1 Data Source ........................................................................................................... 74
4.3.2 Predictor Variables ................................................................................................ 76
4.3.3 Outcome and Follow-up ....................................................................................... 77
4.3.4 Statistical methods ................................................................................................ 79
4.4 Results: .............................................................................................................................. 80
4.4.1 Study Population ................................................................................................... 80
4.4.2 Functional Outcomes ............................................................................................ 83
4.4.3 Model Development and Validation ..................................................................... 84
4.5 Discussion: ........................................................................................................................ 85
4.5.1 Study Limitations .................................................................................................. 89
4.6 Conclusion: ....................................................................................................................... 90
5 Chapter 5: The Impact of Facet Dislocation on Neurological Outcome after Cervical
Spinal Cord Injury .................................................................................................................... 91
5.1 Abstract ............................................................................................................................. 92
5.2 Introduction ....................................................................................................................... 93
5.3 Methods ............................................................................................................................. 94
5.3.1 Patient Population ................................................................................................. 94
5.3.2 Outcome Variables ................................................................................................ 95
5.3.3 Statistical Analysis ................................................................................................ 95
5.4 Results ............................................................................................................................... 97
5.4.1 Patient Population ................................................................................................. 97
5.4.2 Treatment .............................................................................................................. 98
5.4.3 Outcome ................................................................................................................ 99
5.5 Discussion ....................................................................................................................... 100
5.5.1 Study Strengths and Limitations ......................................................................... 102
viii
5.6 Conclusion ...................................................................................................................... 102
6 Chapter 6: Predicting Inpatient Complications after Traumatic Cervical Spinal Cord Injury
through the use of Acute Clinical Variables .......................................................................... 104
6.1 Abstract ........................................................................................................................... 105
6.2 Introduction: .................................................................................................................... 106
6.3 Methods: ......................................................................................................................... 107
6.3.1 Study Population: ................................................................................................ 107
6.3.2 Candidate Complication Predictor Variables: .................................................... 107
6.3.3 Complication Categorization and primary outcome: .......................................... 108
6.3.4 Statistical Analysis: ............................................................................................. 108
6.4 Results: ............................................................................................................................ 109
6.5 Discussion: ...................................................................................................................... 113
6.5.1 Study Limitations: ............................................................................................... 115
6.6 Conclusion: ..................................................................................................................... 115
7 Chapter 7: Defining Age Related Differences in Functional Outcome after Traumatic
Spinal Cord Injury .................................................................................................................. 117
7.1 Abstract ........................................................................................................................... 118
7.2 Introduction ..................................................................................................................... 119
7.3 Methods ........................................................................................................................... 120
7.3.1 Study Population ................................................................................................. 120
7.3.2 Age as a predictor variable.................................................................................. 121
7.3.3 Outcome and follow-up ...................................................................................... 121
7.3.4 Statistical Methods .............................................................................................. 122
7.4 Results ............................................................................................................................. 122
7.4.1 Patient Population ............................................................................................... 122
7.4.2 Univariable analysis ............................................................................................ 124
7.4.3 Multivariable analysis ......................................................................................... 125
ix
7.5 Discussion ....................................................................................................................... 129
7.5.1 Study Limitations ................................................................................................ 130
7.6 Conclusion ...................................................................................................................... 131
8 Chapter 8: Summary of Findings, General Discussion, Thesis Limitations and Future
Directions ............................................................................................................................... 132
8.1 Summary of Findings and Unifying Discussion ............................................................. 133
8.1.1 Clinical Variables ................................................................................................ 134
8.1.2 Demographic Variables....................................................................................... 136
8.1.3 Radiological Variables ........................................................................................ 137
8.1.4 Outcome variables............................................................................................... 139
8.2 Thesis Limitations ........................................................................................................... 140
8.3 Future Directions ............................................................................................................ 142
References ................................................................................................................................... 144
Appendices .................................................................................................................................. 166
x
List of Tables
Table 1-1: Epidemiology and Costs of Spinal Cord Injury ............................................................ 5
Table 1-2: The International Standards for Neurological Classification of Traumatic Spinal Cord
Injury. .............................................................................................................................................. 8
Table 1-3: Current Evidence for the Treatment of SCI. ............................................................... 16
Table 2-1: Inclusion and Exclusion Criteria for Systematic Review ............................................ 21
Table 2-2: Evidentiary Table Summarizing Clinical Predictors of Neurological Outcome ......... 25
Table 2-3: Evidentiary Table Summarizing Clinical Predictors of Functional Outcome ............. 32
Table 2-4: Evidentiary Table Summarizing Clinical Predictors of Complications ...................... 37
Table 2-5: Evidentiary table summarizing radiological predictors of outcome............................ 50
Table 3-1: Variables included in Combined/Harmonized Dataset.. ............................................. 69
Table 4-1: Pre-specified Predictor Variables utilized in the development of prediction models . 77
Table 4-2: Functional Independence Measure (FIM) motor score ............................................... 78
Table 4-3: Patient Characteristics at hospital admission .............................................................. 81
Table 4-4: Mean Follow-up FIM motor score by baseline characteristics ................................... 83
Table 4-5: Parameter Estimates for Models predicting a) FIM motor score and b) Functional
Independence, at 1 year follow-up from original sample and bootstrap replicates ...................... 84
Table 4-6: Predicted Estimates for Hypothetical Patients ............................................................ 87
Table 5-1: Baseline Patient Characteristics .................................................................................. 96
Table 5-2: Clinical Outcomes at 1 year Follow-up, FD vs. Non-FD Cohort ................................ 99
Table 5-3: Clinical Outcomes at Follow-up, Unilateral FD vs. Bilateral FD patients ................ 100
xi
Table 5-4: Results of Multivariable Analysis (Outcome variable: Change in ASIA motor score at
1-year follow-up) ........................................................................................................................ 100
Table 6-1: Results of univariable analysis comparing patients who experienced at least one
complication to those who experienced no complications ......................................................... 111
Table 6-2: Results from the multivariable logistic regression model predicting complication
development. ............................................................................................................................... 112
Table 6-3: Predicted Probability estimates for hypothetical SCI patients .................................. 113
Table 7-1: Patient characteristics at acute hospital admission .................................................... 123
Table 7-2: Functional and Neurological Outcomes at Follow-up............................................... 125
Table 7-3: Results of linear regression analysis relative to outcome variable FIM motor score 126
Table 7-4: Results of multivariable analysis adjusted for additional variables .......................... 127
xii
List of Figures
Figure 1-1: Pathophysiology of traumatic SCI. .............................................................................. 7
Figure 4-1: Flow chart of the study design ................................................................................... 82
Figure 4-2: Predictive Model Equations ....................................................................................... 86
Figure 5-1: Number of Facet Dislocations per disc space level ................................................... 97
Figure 5-2: Treatment Flow for patients with Facet Dislocation .................................................. 98
Figure 6-1: Number of Events per Complication Category ........................................................ 110
Figure 7-1: Diagram depicting study goal. ................................................................................. 120
Figure 7-2: Interaction plot demonstrating the modification effect of age on the relationship
between acute injury severity and functional outcome ............................................................... 128
Figure 8-1: Summary of Main Relationships between Acute Predictors and Outcomes Explored
in Multivariable Analyses throughout Thesis ............................................................................. 134
xiii
List of Appendices
Appendix 1: International Standards for the Neurological Classification of Spinal Cord.......... 166
Appendix 2: Emerging Therapies for Acute Traumatic Spinal Cord Injury............................... 168
Appendix 3: Determination of Level of Evidence for papers included in Systematic Review .. 181
Appendix 4: Summary of Data in Combined Dataset ................................................................ 187
xiv
List of Abbreviations
AD Autonomic Dysreflexia
AIS grade ASIA Impairment Scale grade
AMS ASIA Motor Score
AP Antero-posterior
ASIA American Spinal Injury Association
ASS ASIA Sensory Score
AUC Area under the Curve
CT Computerized Tomography
DU Decubitus Ulcer
DVT Deep Venous Thrombosis
FD Facet Dislocation
FIM Functional Independence Measure
HD Herniated Disc
ICU Intensive Care Unit
ISNCSCI International Standards for the Neurological Classification of SCI
ISS Injury Severity Score
LEMS Lower Extremity Motor Score
MAR Missing at Random
MCAR Missing Completely at Random
MCC Maximal Canal Compromise
mJOA modified Japanese Orthopedic Association
MNAR Missing Not at Random
MPSS Methylprednisolone Sodium Succinate
MRI Magnetic Resonance Imaging
MSCC Maximal Spinal Cord Compression
NACTN North American Clinical Trials Network
NASCIS National Acute Spinal Cord Injury Study
NLI Neurological Level of Injury
OPLL Ossification of Posterior Longitudinal Ligament
OR Odds Ratio
ROC Receiver Operator Curve
SCI Spinal Cord Injury
SCIM Spinal Cord Independence Measure
STASCIS Surgical Timing in Acute SCI Study
UEMS Upper Extremity Motor Score
ZPP Zone of Partial Preservation
1
Preamble
Clinicians’ ability to accurately predict future health outcomes for patients suffering from injury
or disease is of obvious importance. From the standpoint of the health care provider, all of the
important events that occur in the early stages after injury or diagnosis, such as therapeutic
decision making, discharge/disposition planning and patient/family counseling, are potentially
facilitated by attaining an adept understanding of the most likely clinical course for the patient in
question. From the standpoint of the patient or family affected, having access to such information
early in the disease process facilitates the management of expectations throughout the pathway
of care, allowing for mental preparedness and resource planning and allocation. From the
research perspective, achieving a sound understanding of the acute factors predicting future
health events and outcomes is also imperative for purposes of planning and completing clinical
trials, decision analysis projects and cost-effectiveness analyses.
As will be expounded upon in the following sections, traumatic spinal cord injury (SCI) remains
one of the most devastating conditions encountered in clinical medicine. That said, there is
substantial variability in reported outcomes with surviving patients populating a spectrum
between minimally impaired to severely affected. In the long-term, while some patients progress
to become functionally independent, others are relegated to a life of severe disability, dependent
on others for the majority of their daily care needs. It follows from this, that identifying, and
quantifying, the relative impact of factors that influence recovery across this spectrum, is of
immense interest from both the clinical and research perspective. Further, since a large
proportion of the decision making and counseling occurs in acute period after SCI, it is of
particular interest to investigate predictive factors which are of relevance during this acute
period. To date, the majority of the existing literature on this topic comes from outside the acute
realm. Consequently, there is a need to further characterize how acute variables can be used by
the clinician and researcher alike to develop accurate outcome predictions for patients with SCI.
Given this background, the preeminent goal of this thesis is to investigate the use of clinical,
demographic and radiological variables obtained during the acute period after SCI, for purposes
of predicting both short-term and long-term clinical outcomes. As described below, a multi-
faceted definition of outcome has been employed in order to attain a more comprehensive
understanding of how these acute variables affect patients’ future status.
2
Thesis Structure
This thesis is organized according to the “multiple paper format” using primarily unaltered peer
reviewed material. As opposed to the “single continuous paper format”, the chosen structure
allows for multiple self-contained research aims to be addressed under the overall umbrella of
the main thesis topic. Also, since different partitions of the same datasets were used depending
on the specific aim, this format allowed for a description of the data considered and the specific
methods used in each case. In certain instances, when constructing a particular chapter based on
a specific peer reviewed paper, the writing or the content of the chapter may have been altered
from the original paper, in order to ensure homogeneity within the document overall.
Chapter 1 primarily serves as an introduction to the topic of SCI, reviewing basic epidemiology,
standards for clinical and radiological assessment, outcome assessment and the current treatment
options and evidence. Chapter 2 is a modified version of a paper published in “Journal of
Neurosurgery: Spine” and consists of a systematic review of the acute clinical and radiological
variables predicting outcome after SCI. The final section of chapter 2 outlines the overall
hypothesis and the specific aims of this thesis. Chapter 3 is a brief section which provides a basic
summary of the datasets used in this thesis. Chapter 4 is a reformatted version of a paper
published in “Journal of Neurotrauma” describing the development and internal validation of a
clinico-radiographic model to predict functional outcome. Chapter 5 is a reformatted version of a
paper published in the journal "Spine” describing the impact of facet dislocations on neurological
outcomes after cervical SCI. Chapter 6 is a reformatted version of a paper published in “Journal
of Neurosurgery: Spine” which explores the combination of clinical variables predicting acute
inpatient complications. Chapter 7 is a reformatted version of a paper, currently under peer
review, which explores the impact of increasing age on long-term clinical outcomes. Finally,
chapter 8 is a general discussion which collates the results presented in each chapter and
summarizes the main conclusions of this thesis as well as opportunities for future investigation.
It should be noted that explicit permission from the publisher was obtained prior to use of all
previously published material contained in this thesis. Copyright acknowledgements can be
found at the conclusion of this document.
3
1 Chapter 1 Overview of Current Epidemiology, Pathophysiology, Treatment and Assessment Standards for Traumatic Spinal Cord Injury
Section 1.7 is modified from the following:
Wilson JR, Forgione N, Fehlings MG. Emerging therapies for acute traumatic spinal
cord injury. CMAJ. Epub: Dec 10, 2012.1
4
1.1 Introduction
As defined by Chin and colleagues, spinal cord injury (SCI) is “an insult to the spinal cord
resulting in a change, either temporary or permanent, in its normal motor, sensory, or autonomic
function”2. Although SCI can result from traumatic and non-traumatic etiologies alike, this thesis
exclusively relates to traumatic SCI. The goal of this chapter is to provide a basic overview of
the epidemiology, standards for initial assessment and outcome measurement, as well as a brief
description of current treatment options for the management of traumatic SCI. This is included as
a general primer on the topic of SCI to provide context and serve as frame of reference for the
reader throughout the remainder of this thesis.
1.2 Epidemiology and Cost
In Canada, approximately 85,000 individuals live with a diagnosis of SCI, with an incidence rate
of approximately 40 cases per million per year3-7
. At present, SCI is diagnosed in approximately
5% of all major trauma patients presenting to hospital and in about 25% of patients with a
diagnosed injury to the vertebral column8. Internationally, the epidemiology of SCI has been
shown to vary between countries9-11
. Annual incidence rates as low as 14.5 per million per year
have been reported in Australia, while rates as high as 57.8 per million per year have been
reported in Portugal12,13
. While some of these differences may truly reflect differences in the
frequency and overall burden of SCI between regions, it is more likely that they reflect
differences in case reporting standards internationally.
Regardless of the region or country considered, studies report similar findings related to
characteristics of injured populations. From a mechanistic standpoint, the majority of injuries
arise from either motor vehicle accidents (accounting for 30-60% of injuries) or falls (accounting
for 20-60% injuries), with the remaining injuries secondary to violence, sports related/diving
accidents, or work related accidents14-18
. Regarding the level of injury, there is consistency in the
finding that most injuries (55-75%) occur at cervical levels, with the remaining roughly equally
distributed between thoracic and lumbosacral levels9,10,19
. In the majority of published
epidemiological studies, the age distribution of SCI appears bimodal with the first peak occurring
in adolescence/young adulthood between the ages of 15 and 30 (related to an increase in injuries
secondary to violence, sports accidents, and motor vehicle accidents) and the second peak
occurring in those greater than 70 years old (related to an increase in fall related SCI in the
5
elderly)6,7,20
. With the overall aging of the population and the demographic restructuring of
society, it is likely that the proportion of injuries seen amongst the elderly will continue to rise in
the coming years21,22
. Finally, a higher incidence of injuries amongst men as compared to
women has consistently been observed at a ratio of approximately 2.5 to 8:114,20,23-25
.
Table 1-1: Epidemiology and Costs of Spinal Cord Injury
Parameter Value
World Wide Incidence: Range 14- 57/million/year
Canadian Incidence: Range 22-52/million/year
World Wide Prevalence: 680-750/million
Age Distribution: Bimodal Distribution with early peak amongst young adults
and a late peak amongst the elderly
Gender: Male predominance with a male to female ratio of 2.5 to 8:1
Common Injury Etiology: Motor Vehicle Accidents most common amongst younger
patients, falls most common amongst elderly patients
Neurological Level of
Injury:
Cervical: 55-75%
Thoracic:15-30%
Lumbosacral:10-20%
Severity of Injury: AIS grade A (45%), AIS grade B (15%), AIS grade C (10%)
and AIS grade D (30%)
Case Fatality Rate During
Acute Hospitalization:
4.4-16.7%
Annual Mortality Rate
After Acute discharge:
First year: 3.8%
Second year: 1.6
Years thereafter: 1.2%
Annual Cost for:
Complete SCI
Incomplete SCI
121,600 CDN*
42,100 CDN *
*Cost in the first year after SCI (2002 costs).
AIS: ASIA Impairment Scale
While the case fatality rate associated with the initial traumatic event remains high (upwards of
15%), life expectancy post injury has improved dramatically over the past several decades,
primarily as a result of improved supportive management and improved medical treatment for
post-injury complications26
. That said, SCI patients continue to die at higher rates than their non-
injured counterparts27-29
. In a large longitudinal analysis, DeVivo et al showed that even at 3
years after injury, the annual mortality rate amongst SCI patients is about 2.5 times higher than
that expected amongst the general population30
. Those that do survive are often left with
significant disability requiring lifelong assistance with basic daily care needs and activities of
6
daily living. It has been estimated that SCI patients, on average, require 30 times the amount of
home care services as compared to individuals in the general population31
. Given the intensity
and duration of care required for these patients, it not surprising that SCI has been ranked among
the most expensive medical conditions to treat, second only to respiratory distress syndrome in
the pre-term infant32
. Quantifying this economic burden, a Canadian study estimated that the
average direct healthcare costs in the first year for a patient with a complete injury were
$121,600 (2002 CDN dollars), whereas the average costs for a patient with an incomplete injury
over the same period were $ 42,100 (2002 CDN dollars)33
. At the societal level, in Canada, it
has been estimated that the total direct and indirect costs attributable to SCI approach $4 billion
annually (2012 CDN dollars)34
.
1.3 Pathophysiology of SCI
Although a detailed description of the pathophysiology of SCI falls outside the purview of this
thesis, a brief overview is provided for contextualization purposes. It is now well recognized that
the initial compressive force applied to the spinal cord at the time of injury (known as the
primary injury) initiates a deleterious cascade of patho-biological processes collectively referred
to as secondary injury mechanisms35-37
. Initially, an increased region of spinal cord ischemia
develops as a consequence of micro-vessel thrombosis and vasospasm38-41
. This leads to
physiologic derangements at the neuronal level which include cell membrane peroxidation and
destruction, ionic imbalances, cellular swelling and the release of toxic levels of the excitatory
neurotransmitter glutamate42-45
. These events begin within seconds of the primary injury and
continue for up to several weeks, ultimately expanding the region of neural tissue damage and
worsening clinical outcomes46
. Many of the therapeutic strategies currently in use, or under
investigation for future use, aim to maximize individuals’ potential for neurologic recovery by
averting different elements of this secondary injury cascade46
.
7
Figure 1-1: Pathophysiology of traumatic SCI: Schematic illustrating the time related
progression of secondary injury related pathological mechanisms after the primary injury event.
These secondary events increase neural tissue damage and negatively impact long-term clinical
outcomes.
1.4 Current Approach to Clinical and Radiological Assessment
1.4.1 Neurological Assessment of SCI patients
In 1982, in order to introduce standardized methods which clinicians could use to precisely
evaluate and document the extent of neurological deficit in the post SCI setting, the American
Spinal Injury Association (ASIA) produced the first version of the International Standards for
Neurological Classification of Spinal Cord Injury (ISNCSCI)47
. Now in its 5th
iteration, the
international standards have been uniformly adopted by the SCI community for purposes of
evaluating the baseline neurological status and for documenting neurological recovery over
follow-up48,49
. Of note, there are three main components to the International Standards exam:
ASIA motor score (AMS), ASIA sensory score (ASS) and ASIA impairment scale (AIS) grade.
AMS records muscle power in 10 myotomes bilaterally (each with a maximum power of 5)
resulting in a cumulative score with a minimum value of 0 and a maximum value of 100 (25
points attributed to each extremity). For ASIA sensory score (ASS), pinprick and light touch
sensation are assessed in 28 dermatomes bilaterally(C2-S5), with each dermatome receiving a
score of 0, 1 or 2 depending on whether the sensation is absent, abnormal or normal respectively.
8
This results in cumulative scores for both pinprick and light touch with minimum values of 0 and
maximum values of 112. Finally, based on the complete motor and sensory assessment, which
includes a digital rectal examination, the AIS grade is determined according the criteria outlined
in Table 1-2. According to epidemiologic studies, AIS grade A injuries are generally the most
common injuries encountered, followed in sequential decreasing order by AIS D, AIS B and AIS
C injuries9.
Table 1-2: The International Standards for Neurological Classification of Traumatic Spinal
Cord Injury. Important terminology within this classification is described below. (Full details
can be found in Appendix #1)
Parameter Diagnostic Criteria/Explanation
ASIA Impairment Scale (AIS) grade A No sensory or motor function is preserved in
the sacral segments S4-5
ASIA Impairment Scale (AIS) grade B Sensory but not motor function is preserved
below the neurological level of injury and
includes the sacral segments S4-5. No motor
function is preserved more than three levels
below the neurological level of injury
ASIA Impairment Scale (AIS) grade C Motor function is preserved below the
neurological level and less than half of the key
muscle below the neurological level of injury
have a muscle grade less than 3
ASIA Impairment Scale (AIS) grade D Motor function is preserved below the
neurological level and at least half of the key
muscles below the neurological level of injury
have a muscle grade less than 3
ASIA Impairment Scale (AIS) grade E Normal sensory and motor function in a patient
who previously had deficits
ASIA Motor Score Ordinal Score between 0-100 reflecting the
power of 10 key muscle groups (each scored
out of 5) on each side of the body
ASIA Sensory Score Ordinal Score reflecting pain and light touch
sensation in 28 dermatomes on each side of the
body.
Neurological Level of Injury (NLI) The most caudal segment of the cord with
intact sensation and anti-gravity muscle
function on both sides of the body
Zone of Partial Preservation (ZPP) In AIS grade A patient only. The myotomes
and dermatomes caudal to the neurological
level of injury that remain partially innervated
Modified from Kirshblum et al, J Spinal Cord Med, 201149
9
In terms of defining relevant terminology, according to the neurological standards, a patient with
a complete injury has no sensory or motor function in the distal most sacral segment, whereas a
patient with an incomplete injury does have evidence of sensory or motor preservation within
this segment (“sacral sparing”). Also important to delineate is the neurological level of injury,
which is defined as the most caudal segment of the spinal cord with normal sensory and motor
function on both sides of the body. Finally, the zone of partial preservation, applies only to
complete or AIS grade A injuries, and refers to those myotomes and dermatomes caudal to the
neurological level of injury that remain partially innervated.
From a psychometric standpoint, the neurological standards have demonstrated high intra- and
inter-rater reliability, with intraclass correlation coefficients >0.90 consistently seen throughout
the literature50-53
. While it is difficult to evaluate concurrent validity in the absence of a
measurement gold standard, these measures have shown to be correlative with other clinical,
radiologic and electrophysiologic measures which approximate injury severity and hence have
shown substantial evidence of convergent construct validity54
.
1.4.2 Radiological Assessment
Traditionally, the standard radiological test to screen trauma patients suspected of harboring a
spinal injury was X-Ray55,56
. However recently, X-Ray has largely been supplanted by CT due to
findings that plain radiographs alone may miss a large proportion of spine fractures57,58
. In the
acute setting, while CT allows for accurate diagnosis of spinal fractures or dislocations, it is
poorly suited for visualizing soft tissues. As a result, MRI is the preferred modality for
diagnosing injuries to soft tissue structures, such as ligaments and intervertebral discs, and for
visualizing the spinal cord and nerve roots. Relating specifically to the evaluation of the spinal
cord, MRI may be used to identify the exact anatomic location of injury, to assess the degree and
nature of spinal compression, and to assess for the presence or absence of abnormal
intramedullary signal characteristics associated with specific pathological states (i.e. hemorrhage
or edema). These MRI findings are important not only for purposes of initial evaluation and
treatment planning, but also for long-term prognostication. The current state of evidence
surrounding the use of acute imaging variables for purposes of outcome prediction is reviewed in
depth in the subsequent chapter.
10
Overall, the current recommendation is that patients be assessed in the acute period following
injury with both CT, for evaluating the bony injury, and MRI for evaluating the soft tissues and
neural elements59,60
. Although interesting from a research standpoint, at present, other imaging
modalities, such diffusion tensor imaging (DTI)61-63
, magnetic resonance spectroscopy
(MRS)64,65
and functional magnetic resonance imaging (fMRI)66-68
, have no standard role in the
routine clinical assessment and management of SCI patients.
1.5 Time periods after SCI
In general, SCI is classified as acute, sub-acute or chronic depending on the time elapsed from
the injury event. For purposes of this thesis, the time windows used by the International
Campaign for Cures of spinal cord injury Paralysis were used to define these periods. The first 3
days after injury represent the acute period during which the patient arrives at the trauma center,
initial clinical and radiological assessments occur, acute medical and surgical therapy is initiated
and enrollment in trials investigating acute therapeutics takes place. The sub-acute stage
represents the period between 3 days and 1 year after injury, during which the majority of
neurological recovery occurs, as is explained below. The chronic stage is attained at 1 year after
injury, at which point the majority of patient recovery has taken place. Throughout this thesis,
discussions and analyses surrounding acute clinical and radiological predictors refer to those that
have been obtained within the first 3 days after injury. The one exception to this rule pertains to
Chapter 6 of this thesis on the topic of “acute complications”. In this one instance, acute refers to
complications events that occur during the period from acute hospital admission to acute care
discharge.
1.6 Outcome Assessment after SCI
There are many frameworks for discussing and reporting on health outcomes. Perhaps the most
widely accepted at present is the World Health Organization (WHO) International Classification
of Functioning, Disabilities and Health which encourages us to consider a bio-psycho-social
definition of health status that includes physical, psychological and social components69
.
However for purposes of this thesis, which explores the predictors of outcome after SCI,
neurologic outcome, functional outcome and complications will serve as the central focus for
investigation and discussion. These particular outcomes have been chosen as they are felt to be at
the very core of well-being for SCI patients70,71
. Moreover these represent outcomes of particular
11
relevance and interest to clinicians and researchers, having been the central outcomes considered
in most of the large SCI clinical studies performed to date. Other outcomes such as quality of
life, pain, disability and mental health are of immense importance however will not be explored
at significant length in this thesis.
Traditionally, neurological improvement at follow-up has been the main outcome of interest in
the majority of SCI clinical trials, as defined by change in AIS grade, AMS or ASS between
hospital admission and follow-up70
. Other neurological outcome measures such as Frankel scale,
NASCIS motor and sensory scores and modified Benzel scale have been considered in past
trials, however use of these has largely been abdicated in favor of the international standard
ASIA indices discussed above72,73
.
The timing of follow-up neurological exam has largely been dictated by the temporal profile for
neurological recovery after injury. While the majority of recovery is known to occur within the
first 4-6 months after injury, a small proportion of patients have been shown to experience
improvement up to 5 years out from injury74
. In an analysis of AIS grade conversion rates from
the multicenter Sygen study which followed patients for 1 year, investigators observed that 77%
of patients who improved at least 2 grades did so before 4 months post injury and 92% did so
before 6 months post injury24
. However, in a separate study, Kirshblum and colleagues
demonstrated that approximately 5% of patients who were AIS grade A at 1 year converted to
incomplete status by year 575
. Taking all these points together, for purposes of defining terms of
reference for this thesis, throughout the subsequent analyses, 1 year follow-up was considered to
be “long-term” outcome. This was felt to best balance the desire to capture the full extent of
individuals’ recovery with the practical challenges of following large groups of patients over
long periods of time.
Several measures of neurological outcome, including AIS grade conversion, have shown to be
poorly predictive of future functional capacity76
. In light of these findings, the International
Campaign for Cures of spinal cord injury Paralysis has prioritized the use of neurologic and
functional outcome assessment tools in current and future SCI clinical trials70
. When considering
functional outcome in the context of SCI, while a number of different indices have arisen over
the years, two main measures are generally considered central: the Functional Independence
Measure (FIM) and the Spinal Cord Independence Measure (SCIM). FIM is a generic
12
neurological functional outcome assessment tool that has been widely used in the clinical
neurosciences throughout the last 2 decades. It consists of 2 subscales: motor and socio-
cognitive, with the socio-cognitive subscale often considered less relevant in SCI outcome
assessment. The FIM motor subscale consists of 13 items which assess individuals’ level of
functional independence across four different domains which include self-care, sphincter control,
transfers and locomotion. The FIM socio-cognitive subscale consists of 5 items which assess
individuals’ communication and social cognition. The performance level for each item is strictly
defined and ranges in value from 1 to 7, where 1 indicates complete dependence in an activity
and 7 indicates complete independence. A score of 6 or greater for each item indicates that a
patient is capable of performing that activity independently, without supervision or help.
Patients’ total cumulative score ranges from 18 to 126 when considering both subscales, or from
13 to 91 when considering just the motor subscale, with a higher value reflecting superior
functional status in either case. The psychometric profile of FIM has been intensively studied in
the context of SCI, with findings of good inter and intra-rater reliability and an acceptable body
of evidence establishing its convergent construct validity in relation to other indices that reflect
severity of impairment post injury77-80
. While FIM is likely the best researched of any functional
outcome measure used in SCI, one frequently cited limitation is, that because it is a generic tool,
it is less responsive or sensitive to subtle changes in functional status within this population81
. In
addition, ceiling effects may also be an issue for patients with less severe injuries when using
FIM. These ceiling effects are especially evident for the socio-cognitive subscale and less so for
the motor subscale82
. While formally speaking, FIM is a measure of functional independence,
hereafter and throughout this thesis, it will be referred to as a measure of functional outcome.
Developed by Catz and colleagues, SCIM is an SCI specific functional outcome measure that has
gained acceptance in the field over the last 5-10 years83
. The most recent SCIM version (version
III) contains 19 items which assess function over domains of self-care, respiration and sphincter
management and mobility84
. The individual items are weighted according to clinical relevance
and when summated result in a total combined score out of 100, with a higher value reflecting
superior functional status. Like the FIM, SCIM has shown demonstrate good inter and intra-rater
reliability85
. From the standpoint of concurrent validity SCIM has shown to correlate highly with
the FIM and, because SCIM is specific for SCI, it has been shown to be more responsive to
subtle changes in function as compared to FIM81,85-87
. Overall both SCIM and FIM are well
13
studied and appropriate for functional outcome measurement in the context of traumatic SCI.
While SCIM overcomes some of the responsiveness issues identified with FIM, it is a relative
new comer to the field and hence databases or studies commenced before its introduction may
not have incorporated it as a standard follow-up measure. It should be noted that many additional
less frequently utilized measures of functional status have been studied in the context of SCI
with some assessing self-care capacities (i.e. Modified Barthel Index, Quadriplegia Index of
Function, Self-Care Assessment Tool) while others are focused on mobility and ambulation (i.e.
Walking Index for Spinal Cord Injury, 10 Meter Walking Test)88-91
.
Apart from neurologic and functional outcomes after SCI, it is also of interest to understand
deleterious patient outcomes such as complications. Complications occur with a high frequency
after SCI and are known not only to increase morbidity and mortality, but also to negatively
impact neurologic and functional recovery92
. While a standard method of complication reporting
has yet to be uniformly adopted by the clinical and research SCI communities, complication
events can be broadly defined as either acute complications (during acute hospital admission) or
sub-acute/chronic complications (sustained after acute discharge). This distinction is important
since the type and frequency of complications affecting patients differ between these time
periods as will be discussed further in the subsequent chapters.
1.7 Treatments Available
In general, the main acute therapies for SCI can be discussed across four broad categories:
supportive, pharmacologic, surgical and cellular based. While not discussed at significant length
here, the importance of neuro-rehabilitation in helping individuals achieve their maximum
potential for neurological and functional recovery cannot be overemphasized.
1.7.1 Supportive/Critical Care Therapy for SCI
Perhaps the most significant advances in the management of SCI have related to ensuring
adequate perfusion and oxygenation of the injured spinal cord as well to the prevention of
secondary complications during the acute setting. The deleterious effects of hypotension in the
setting of injury to the central nervous system are well established93,94
. As a result, avoiding
hypotension and maintaining adequate blood pressure targets are of paramount importance in
helping to optimize patients’ likelihood of recovery. According to the most recent American
14
Association of Neurological Surgeons guidelines for the management of acute SCI, it is
recommended that in addition to avoiding hypotension, mean arterial pressure should be
maintained between 85-90mmHg for the first 7-14 days after injury95
. In addition to
optimization of blood pressure parameters, it is also recommended that injured patients,
particularly those with severe cervical injuries, be admitted to an intensive care unit setting for 7-
14 days where continuous cardiac, hemodynamic and respiratory monitoring can be undertaken.
In this setting, the commonly encountered early medical complications that contribute substantial
morbidity to SCI patients can promptly be identified and averted96
. The standardized adoption of
such physiological targets as well as ICU admission has been associated with improved
neurological recovery and reduced mortality rates in several observational and quasi-
experimental studies97-101
.
Over the years, a variety of studies have also investigated the use of locally applied (during
surgery) or systemically administered hypothermia, in the acute stages post SCI102
. These studies
are based on promising laboratory data suggesting that cooling reduces tissue energy metabolism
and mitigates secondary injury mechanisms early after injury. To date there is no data to support
the efficacy of local or systemic hypothermia in the context of SCI102
. In the analogous condition
of traumatic brain injury, randomized controlled trials have failed to associate cooling with long-
term clinical benefit103
. Of note however, a recently completed phase I study showed systemic
moderate hypothermia delivered for the first 48 hours after injury to be safe104
. Larger efficacy
trials investigating this therapy are in the planning stages.
1.7.2 Pharmacologic Therapy for SCI
To date, a total of 5 drug therapies for acute SCI have been evaluated in the context of phase III
efficacy trials. Unfortunately, while controversy continues to loom over the use of
methylprednisolone, none of these medications have shown to reliably improve long-term
outcomes for affected individuals.
Of the drugs that have been evaluated, Methylprednisolone Sodium Succinate (MPSS) has been
the most intensively investigated, predominately in the context of the three National Acute
Spinal Cord Injury Studies (NASCIS). In the first NASCIS study (NASCIS I), comparison of a
10 day regimen of medium dose MPSS to low dose MPSS revealed no difference with respect to
neurological recovery at 6 months follow-up72
. In the subsequent NASCIS II, a 24 hour high
15
dose MPSS regimen was compared to placebo with respect to the primary outcome of NASCIS
motor score recovery at 6 months follow-up105
. While the primarily analysis failed to a find a
difference between treatment arms, a pre-specified secondary analysis demonstrated that when
treatment was commenced within 8 hours of injury, those treated with MPSS experienced
significantly greater motor recovery. However, in the MPSS treated group, there were also weak
trends towards increased rates of wound infections and gastrointestinal hemorrhage. In the
subsequent NASCIS III study, a 48 hour infusion of high dose MPSS was compared to the 24
hour infusion from NASCIS II106
. In the primarily analysis, the 48 hour regimen failed to
demonstrated evidence of efficacy and was associated with strong trends towards increased
infectious complications. Collating the results of these and other studies on the topic, consensus
guidelines from the relevant physician led organizations, including the American Association of
Neurological Surgeons as well as the Canadian Spine Society and the Canadian Association of
Emergency Physicians, have recommended that the 24 hour infusion of high dose MPSS is only
a treatment option that should only be administered with the knowledge of the potential for
increased infection related complications107-110
. Throughout the individual analyses that comprise
this thesis, administration of the NASCIS II 24hour regimen of MPSS was permitted to SCI
patients, at the discretion of involved clinicians.
Four other agents including the ganglioside compound GM-1 (Sygen), the opiate receptor
blocker naloxone, the L-type calcium channel blocker nimodipine and the putative
neuroprotective agent Tirilazad mesylate, have been evaluated in the context of phase III trials.
Unfortunately, none of these drugs have demonstrated clear evidence of efficacy105,106,111
.
However, in spite of these failures, there are a number of promising new neuroprotective and
neuroregenerative agents in various stages of clinical development, that appear poised to
overcome the limitations of the discussed drugs112-114
. At present, there is no accepted or
standard drug treatment for SCI.
16
Table 1-3: Current Evidence for the Treatment of SCI.
Therapy Current Status of Evidence
Critical-care setting/Hypothermia -Patients with severe cervical level SCI should
be managed in an Intensive Care Unit setting
with continuous hemodynamic, cardiac and
respiratory monitoring for the first 7-14 days
post injury.
-No established role for systemic or local
hypothermia
Hemodynamics Hypotension should be avoided with Mean
Arterial Pressure maintained between 85 and
90 mmHg for the first 7 days post injury
Pharmacologic Therapy -Administration of methylprednisolone for
either 24 or 48 hours is an option that should
only be undertaken with the knowledge of the
potential for an increased occurrence of
complications
Surgical Decompression -Decompressive surgery within 24 hours after
injury has been shown to be safe and feasible
-Prospective non-randomized studies have
associated early surgery with improved
neurological recovery
Cellular Transplantation -Purely an investigation therapy at present time
Modified from Hadley et al, Neurosurgery, 2002 and Wilson et al, CMAJ, 2012
1.7.3 Surgery for SCI
In the context of traumatic SCI, there are three interrelated goals of surgical care: 1) re-establish
the alignment of the spinal column; 2) re-establish the stability of the spinal column; and, 3)
decompress the injured spinal cord115
. Until recently however, spine surgeons were reticent to
proceed to surgery in the early phases post SCI due to concerns that peri-operative hemodynamic
changes could compromise perfusion of the spinal cord and exacerbate neural injury1. As well,
until the advent of instrumented fusion techniques, surgery meant further destabilization of an
already unstable spine. In recent years this perception has changed in light of several factors.
First, a compelling body of preclinical evidence has arisen to support the fact that ongoing
compression of the spinal cord is a potentially reversible form of secondary injury that can be
mitigated through early surgery115-117
. Second, a growing body of clinical literature has
17
confirmed the safety of early decompressive surgery as well as its potential for positive impact
on long-term clinical outcomes118-121
. The recently published Surgical Timing in Acute Spinal
Cord Injury Study (STASCIS), evaluated the safety and efficacy of early (before 24 hours post
SCI) vs. late (at or after 24 hours post SCI) surgical decompression119
. In the final analysis of
this prospective cohort study, early surgery was found to be safe and was associated with a
significantly higher rate of neurological recovery as defined by a 2 AIS grade improvement at 6
months follow-up. Recent survey studies, gauging international surgical opinion, have
determined that at present, the majority of the spine surgery community considers early
decompressive surgery in the setting of SCI a priority, based on the best available evidence122
.
1.7.4 Cellular based Therapy
Implantation of a variety of stem cells and autologous non-stem cells has been associated with
improved neurobehavioral recovery in animal SCI models123
. Theorized mechanisms of action
vary depending on the cellular subtype considered and include: 1) release of growth promoting
trophic factors; 2) environmental modification; and, 3) cellular replacement. While implantation
of several sub-types has shown to be safe in early phase clinical trials, none have demonstrated
evidence of clinical efficacy and as a result, this therapy is presently investigational.
18
2 Chapter 2: Clinical and Radiological Predictors of Neurological outcome, Functional outcome and Complication development after Traumatic Spinal Cord Injury: A Systematic Review
This chapter is modified from the following:
Wilson JR, Cadotte D and Fehlings MG. Clinical predictors of neurological outcome,
functional status, and survival after traumatic spinal cord injury: a systematic review.
Journal of Neurosurgery: Spine 2012 (1Sup); 17:11-26.124
19
2.1 Abstract
A comprehensive systematic review of the literature was performed to identify clinical and
radiological predictors of outcome in the context of traumatic SCI. The review was performed in
two sections. The first section included studies identifying clinical predictors of: 1) neurologic
outcome; 2) functional outcome; or, 3) complication occurrence. Clinical predictors related to
patient demographics, injury mechanism or neurological exam, were extracted from studies
included and the individual relationship to outcome was defined. The second section included
studies identifying acute imaging related variables predicting the same three outcomes. For this
section the relationship between specific quantitative and qualitative imaging variables and
outcome was defined. Selected articles, in both sections, were classified according to their level
of evidence. Overall, the severity of neurological injury (as measured by admission ASIA
Impairment Scale grade, Frankel grade or injury completeness), level of injury and the presence
of a zone of partial preservation were consistent predictors of neurological outcome. The severity
of neurological injury, level of injury, reflex pattern and age were consistent predictors of
functional outcome. As regards complications, severity of neurological injury, level of injury,
and age were the strongest predictors of both acute and chronic complication events. Finally,
regarding imaging predictors, signal pattern consistent with intramedullary hemorrhage was
consistently associated with poor outcome, whereas absence of any abnormal intramedullary
signal change was associated with excellent long-term outcome. There is also consistency that
signal patterns consistent with edema or contusion reflect injuries of intermediate severity in
which the potential for recovery is variable but generally better than patients with hemorrhage
and worse than those with normal cord appearance. Based on the review of literature, the last
section of this chapter frames the overarching hypothesis and central aims of this thesis.
20
2.2 Introduction
To identify acute variables known to be of predictive importance in determining outcome after
traumatic SCI, a systematic literature review was performed. This review was completed in two
parts, with the first section dedicated to identifying clinical variables of predictive significance
and the second component dedicated to identifying radiological variables of predictive
significance. The results of this literature review were used primarily to identify clinical and
radiological variables that could be incorporated as candidate predictor variables in the
construction of clinical prediction models. As well, this review served as the basis to identify
unanswered questions and opportunities for additional investigation, pertinent to the topic of
clinical prediction in the realm of traumatic SCI.
Throughout this chapter the significance of predictive variables were gauged relative to three
main outcomes: 1) neurological outcome, 2) functional outcome, and 3) complication
development. In the first component, the factors examined for their potential predictive value
included the basic clinical elements which physicians gather at patient presentation and which
often influence therapeutic decision-making. For purposes of standardization, these elements fall
into 3 main categories: 1) neurological exam characteristics 2) demographics, and 3) injury
mechanism/etiology. For the second component the factors examined for their potential
predictive value included the radiological (CT and MRI) variables that are used as diagnostic
adjuncts in the acute clinical setting. For purposes of standardization, these variables fall in 2
main categories: 1) qualitative radiological variables, and 2) quantitative radiological variables.
Contained in the last section of this chapter, and based on the findings of this systematic review,
is the overarching hypothesis and specific aims of this thesis that serve as the central framework
for all analyses presented in subsequent chapters.
2.3 Common Systematic Review Methods
We performed a comprehensive computerized literature review using MEDLINE, PubMed,
EMBASE, CINAHL, and the Cochrane Database of Systematic Reviews. For the first
component focused on clinical variables, search items included the MeSH terms: “spinal cord
injury” “prediction” “prognosis” “neurologic outcome measure” “functional outcome measure”
“complications” “adverse event”. For the second component focused on radiological variables,
21
search items were change to include the MeSH terms: “spinal cord injury” “computerized
tomography scan” “magnetic resonance imaging” “neurologic outcome measure” “functional
outcome measure” “complications” “adverse event”.
The literature search was limited to human and English language studies published between 1975
and 2011 (Table 2-1). Case reports and case series with less than 15 patients were excluded.
Studies involving patients with both blunt and penetrating injury mechanisms were included. We
required included studies to correlate clinical and radiological features within 1 month of initial
injury to clinical outcome status at least 3 months after injury. The exception to this was for
studies examining complication outcomes, for which any duration of follow-up was acceptable.
Studies examining the predictive capacity of electrophysiologic investigations were excluded.
When several different studies used the same dataset at different time points, we utilized the
study which displayed the most complete version of the dataset. Information from textbooks or
expert opinion was not included in this analysis.
Table 2-1: Inclusion and Exclusion Criteria for Systematic Review
Study Component Inclusion Exclusion
Participants -Patients with traumatic SCI
-Injury severity ASIA A-D
-Any level of injury
-Age >18
-Non traumatic SCI
-Injury severity ASIA E
Outcomes - Studies containing:
1) clinical information available up to 1
month post injury
2)Functional and Neurologic outcome
data available at least 3 months post
injury or complication data at any time
point
-Studies evaluating predictive
capacity of electrophysiologic
features
Publication -Published in English peer reviewed
journals after January 1, 1975.
-Meeting abstracts, editorials,
letters and narrative or
systematic reviews
-Articles identified as
preliminary reports when
results are published in later
versions
Study Design -Case series or cohort studies of 15 or
more patients
-Case reports or case series
with < 15 patients
22
Each abstract was reviewed to identify relevance to our study question and to confirm
inclusion/exclusion criteria. Citations that appeared to be appropriate or those that could not be
excluded unequivocally from the title and abstract were identified, and the corresponding full
text reports were reviewed. Articles were then manually cross-referenced with their
corresponding reference lists to find a more complete data set. Articles selected for inclusion
were then classified by level of evidence. The method used for assessing the quality of evidence
of individual studies incorporates aspects of the rating scheme developed by the Oxford Centre
for Evidence-based Medicine125
and modified by Wright et al126
. Appendix 3 contains the
evidence appraisal for each of the studies included.
For both components of this review, articles were stratified into one of three domains depending
on whether the primary focus was clinical prediction of: 1) neurologic outcome: study’s main
outcome was a neurological exam feature (i.e. improvement in muscle power) or a neurological
outcome measure validated for use in SCI (i.e. ASIA Impairment Scale, ASIA motor score,
Frankel Grade); 2) functional outcome: study’s main outcome was a functional goal (i.e.
ambulation) or a functional outcome measure validated for use in SCI (i.e. Functional
Independence Measure); or, 3) complications: study’s outcome was the development of
complications post SCI. When a single study related to more than one of the above outcome
groups it was assigned to both or all three. Within each study we examined the relationship
between acute clinical and radiological variables and future outcome.
After compilation and classification of relevant studies, for the first component of this review
concentrated on clinical predictors, we attempted to answer the following three questions, in
order to provide a focused summary of findings:
1) How does neurological examination at admission relate to outcome after traumatic SCI?
2) How do patient demographics relate to outcome after traumatic SCI?
3) How does injury mechanism or etiology relate to outcome after traumatic SCI?
For the second component of the review, concentrated on radiological predictors, the following
two questions were posed:
1) How do quantitative radiological variables relate to outcome after traumatic SCI?
2) How do qualitative radiological variables relate to outcome after traumatic SCI?
Below, results and discussion are compiled in two separate sections, with the first dedicated to
review of clinical predictors and the second dedicated to review of radiological predictors.
23
2.4 Literature Review Section 1: The impact of clinical variables on outcome after SCI
2.4.1 Results
The initial search resulted in 385 citations. Application of the inclusion and exclusion
criteria reduced this to 85 citations including those obtained after a secondary review of
bibliographies. After review of these, 54 relevant articles that identified the clinical predictors of
neurologic, functional, or complication outcomes were identified and graded by the authors. Of
the relevant articles, 26 provided predictors for neurologic outcome, 23 for functional outcome
and 14 for complications, with several of the articles providing information on more than one
type of outcome. As regards level of evidence, all of the included studies were designated as
either level II, or III.
2.4.1.1 Neurological Outcome
The 26 articles identifying clinical predictors of neurological outcome are displayed in Table 2-2.
Specific outcome measures utilized in these studies included: ASIA impairment scale (AIS)
grade or ASIA Motor/Sensory score (AMS/ASS) in 14 studies16,76,127-139
, muscle power grade in
11 studies134-137,140-146
, Frankel grade in 3 studies16,147,148
, Modified Benzel scale in 1 study149
and National Acute Spinal Cord Injury Study motor/sensory scores in 1 study150
. Nine of the
studies were classified as level II evidence while the remaining 17 studies were classified as
level III.
The most frequent predictor of neurological outcome was baseline neurological exam.
Consistently, the presence of an incomplete injury as compared to a complete injury76,130,144,147-
149 and the presence of a Zone of Partial Preservation (ZPP) in those with a complete
injury133,135,140,142,145,146
, were associated with improved neurological recovery. In addition, three
studies reported a worse prognosis for neurologic recovery for complete SCI patients with a
Neurological Level of Injury (NLI) within in the thoracic region 16,134,149
.
Three studies addressed the impact of injury etiology on neurological prognosis16,127,138
. Marino
et al reported that complete injuries arising from violent injuries were more likely to remain
complete as compared to nonviolent injuries, with 95% of the violent injuries noted to be
penetrating in nature. In contrast, in 2 separate studies, Waters et al failed to find an effect of
24
etiology (penetrating vs. non-penetrating injuries and stab-wound related vs. non-stab wound
related injuries) on long-term neurologic recovery.
Age effects were addressed by 4 studies, with 2 of these suggesting a smaller amount of
neurologic recovery with increased age131,132
and a third finding diminished recovery only
amongst older patients with incomplete SCI147
. In contrast, the remaining study found motor
recovery to occur independently of patients’ age at time of injury and sensory recovery to be
superior amongst older patients (age >65 years)150
.
Only one study examined the influence of gender, with Sipski et al reporting superior AMS
recovery amongst women as compared to men on analysis of over 14,000 patients (11,762 males
and 2671 females) within the Model systems database129
.
25
Table 2-2: Evidentiary Table Summarizing Clinical Predictors of Neurological Outcome
Paper LoE Number of
Patients
Predictive factors
assessed
Primary
Outcome(s)
Follow-
up
Findings
Bravo et al,
Paraplegia,
1996
III 50 SCI with
vertebral
fracture
Age
Frankel score on
admission and discharge
Frankel
grade at
follow-up NR
Incomplete lesions and vertebral displacement <30% were
statistically associated with improved neurologic recovery; p
< .001 for both
Age was a factor only in patients with incomplete lesions:
those who were aged < 30 years did significantly better than
those > 30 years, p < .001
Brown et al,
AMPR, 1991
III 29 with C4-
7 motor
complete
injuries
<24-hour MMT
72-hour MMT
Success=
Grade ≥4/5
muscle
strength
(MMT)
3
months
Short-term functional muscle recovery better predicted by the
72-hour MMT than the <24 hour MMT
o At 24 hours, 12/17 (71%) subjects with grade 3 and 4/12
(33%) with grades 1 or 2 achieved success (p > .05)
o At 72-hours, 11/11 (100%) with grade 3 and 5/18 (28%)
with grades 1 or 2 achieved success (p < .001)
Browne et al,
AMPR, 1993
II 24 with
motor
complete
C4-5
injuries
C5 pin sensation
ECR strength
ECR
Power 1 year
Only 2/9 ( 22%) patients with absent C5 pin sensation
recovered ECR to ≥3/5, whereas 14/15 (93%) patients with ½
or 2/2 C5 pin sensation had ECR motor recovery to >3/5
The presence of 1/5 to 2/5 initial ECR strength or ½ or 2/2 C5
pin sensation were highly significant predictors of ECR motor
recovery to ≥3/5
Cifu et al,
APMR, 1999
III 375 cervical
injuries
Age AMS at
follow-up
4-10
months
The two younger groups of patients had a more significant
improvement than did older patients, as indicated by AMS
Coleman et al,
Spine J, 2004
III 760 with
injuries
rostral to
T10
Age
AIS
NLI
Marked
recovery
(MR)*
Modified
Benzel
scale
26
weeks
Injury severity was strongest predictor of MR with AIS
Groups C and D having more MR (84.0%) than Group B
(46.6%), which recovered more than Group A (12.8%); p <
.0001.
The cervical group showed more MR than the thoracic group,
37.2% vs. 15.9, p < .0001
Within AIS Group A the cervical subgroup had higher MR
than the thoracic one (15.5% vs. 7.0%, p < .02), but MR was
nearly equal between regions in the B and CD groups.
Ditunno et al, III 35 with C4-
Power in biceps and
ECR
Power
(strength)
tests at 8
Initial biceps strength is a reliable indicator of wrist extensor
recovery
Most, if not all, C5 neurologic patients will gain one full
26
AMPR, 1987 6 motor
complete
injuries
follow-up months motor level
Ditunno et al,
AMPR, 1992
III 150 with
C4-6 motor
complete
injuries
Power of key muscles in
U/E
Power of
key U/E
muscles at
follow-up
24
months
More subjects with some motor power initially improved to
grade 3/5 at all intervals earlier than those with no motor
power (p<.005)
For the 67 patients with 1-2.5 power in the ZPP, the
improvement to grade 3/5 in the key muscles was
significantly greater than for those with no initial motor power
(68% to 82% vs. 14% to 36%, p<.001)
Ditunno et al,
AMPR, 2000
II 167 with
cervical
injuries
MMT
Power in key muscles of
U/E
U/E motor
recovery
(≥grade
3/5)
Muscle
power at
follow-up
2 years Predicting U/E recovery at a specific motor level is possible
within the first week of injury
Recovery of the biceps (C4 group) showed 70% of complete
compared with 90% of incomplete injuries recovered (p <
.001)
Extensor carpi radialis (C5 group): 75% complete and 90%
incomplete recovered (p < .002)
Triceps (C6 group): 85% of complete and 90% of incomplete
injuries recovered (p < .16)
Fasset et al, J
Neurosurg
Spine, 2007
III 3481 from
SCI registry
Age
AIS
AMS
Level of injury
AIS at
follow-up
Mortality
1 year Patients age >70 years:
were less likely to have as severe neurologic deficits as
younger patients; AIS grade C and D 63% vs. 40%, p < .001
had higher mortality rates (both during hospitalization and 1-
year follow-up); 27.7% vs. 3.2%, p < .001
o Mortality rates in older population directly correlate with
the severity of neurological injury (1-year mortality rate:
AIS grade A 66% vs. grade D 23%, p < .001)
Furlan et al, J
Neurotrauma,
2009
II 499 injuries
Age
NASCIS scores
Pain
Motor
Sensory
NASCIS
scores at
follow-up
1 year
Age and motor recovery or pain scores were not significantly
correlated
Older age was associated with greater sensory recovery and
with greater disability, as assessed using the FIM
27
Katoh et al,
Paraplegia,
1995
III 21 with
Frankel B
injuries
Pattern of sensory
sparing
AMS
1 year
Patients with spinothalamic sensory preservation between the
NLI and the last sacral dermatomes had larger improvements
in AMS at 1 year as compared to those with dull or no
sensation in the sacral dermatomes; 40.3 ± 10.7 vs. 16.5 ±
16.0, p < .001
Similarly, a greater number of patients with sensory
preservation between the NLI and the last sacral dermatomes
were able to walk at 1 year compared with the other 2 groups
combined; 75% vs. 32.5% , p = .02.
Mange et al,
1992
III 39 with C4-
6 motor
complete
injuries
Frankel grade
Power in key muscles of
U/E
Time to
recovery
(assessed by
muscle
power at
follow-up)
6
months
Initial muscle strength was a significant predictor for time to
recovery to specific motor grades
Subjects whose most rostral key muscle in the ZPP who had a
motor power grade of 1/5 had a median time to recovery to
grade 3/5 of 3 months compared to 0.5 months for subjects
with initial motor power grade of 2/5 (p<.001)
Marino et al,
AMPR, 1999
III 3585
injuries
Frankel grade
AIS
AMS
NLI
AMS at
follow-up
AIS at
follow-up
Frankel at
follow-up
1 year Neurologic recovery is influenced by etiology and severity of
injury, as predicted by AMS and AIS
Motor score improvements were related to severity of injury,
with greater increases for better AIS grades except Grade D,
because of ceiling effects
Motor score improvement also depended on the NLI for
complete patients, with paraplegic patients (thoracic and
lumbar injuries) experiencing reduced AMS recovery as
compared to quadriplegic patients (cervical injuries)
Maynard et
al, J
Neurosurg,
1979
III 123 with
cervical
injuries
Frankel grade at 72 hours
post-injury
Frankel
grade at
final follow-
up
1 year
81% (50/62) with Frankel grade A (complete injuries) were
unchanged at 1 year vs. 39% (7/18) with grade B and 30%
(7/23) with grade C.
Pollard et al,
Spine, 2003
III 412 with
incomplete
cervical
injuries
AMS
ASS
AIS
Neurologic
recovery
(determine
d by AMS
and ASS
recovery)
2 years
Most important prognositc variable is the completeness of the
lesion
Improved neurologic outcomes were noted in incomplete SCI,
and younger patients (P = 0.002), and those with either a
central cord or Brown-Sequard syndrome (P = 0.019)
Neurologic recovery was not related to gender, race, type of
fracture, mechanism of injury, high-dose methylprednisolone
28
administration, early definitive surgery, early anterior
decompression for burst fractures or disc herniations, or
decompression of stenotic canals without fracture
Sipski et al,
AMPR, 2004
III 14,433
injuries
AMS
AIS
Gender
AMS at
follow-up 1 year
When grouped by severity of injury, AMS improvements
were significantly greater for women than men with either
complete (P = .035) or incomplete (P = .031) injuries
Women may have more natural neurologic recovery than
men; however, for a given level and degree of neurological
injury, men tend to do better functionally than women at time
of discharge from rehabilitation
Steeves et al,
Spinal Cord,
2011
III 426 cervical
motor-
sensory
complete
AMS
AIS
ASS
AMS at
follow-up
AIS at
follow-up
UEMS at
follow-up
1 year
AIS grade conversion did not significantly influence motor
level changes
Average spontaneous improvement in AMS was 10 points
Moderate relationship between a change in UEMS and a
change in cervical motor level
Distribution of the UEMS change may be more important
functionally than the total UEMS recovered
van
Middendorp
et al, Spinal
Cord, 2009
III 273 injuries
AIS
TUG test
10MWT
AIS
follow-up
Ability to
walk
1 year AIS conversion outcome measure is poorly related to the ability
to walk
Ratio of patients with both recovery of ambulatory function
and AIS conversion (n=101) differed significantly (P < 0.001)
between the acute phase AIS grade scores
Waters et al,
AMPR, 1992
II 148 with
complete
thoracolumb
ar injuries
NLI
ASS
AMS
AIS
Power in key muscles
Motor
function
AMS at
follow-up
Muscle
power at
follow-up
1 year
No patients with a NLI above T9 regained any LE motor
function; 38% of those with a NLI at or below T9 had some
return of LE motor function
Motor function dependent on initial NLI; sensory independent
on initial NLI
Six (4%) of the 148 patients demonstrated “late” conversion
(more than 4 months after injury) from complete to
incomplete SCI status
Waters et al,
AMPR, 1993
II 61 with
complete
cervical
injuries
AIS
AMS
ASS
Power in key muscles
AMS at
follow-up
Muscle
power at
follow-up
3 years
55/61 (90%) remained complete; 6/51 cases underwent late
conversion to incomplete injury status
97% of muscles at 1/5 or 2/5 strength one month after injury
recovered to ≥ 3/5 strength
Muscles with 0/5 strength one month after injury and located
one level below the most caudal level having motor function
regained ≥ 3/5 strength in only 27% of cases and at two levels
in 1% of cases
29
Waters et al,
AMPR, 1994a
II 54 with
incomplete
paraplegia
NLI
AIS
AMS
Frankel grade
Power in key muscles
LEMS at
follow-up
AMS at
follow-up
Muscle
power at
follow-up
1-2
year
LEMS differed significantly (p < .03) between groups of
patients partitioned by category of initial NLI
Improvement in LEMS independent of NLI
o 74% showed no improvement in NLI
Absence of any motor function 1 month after injury is not an
absolute indicator of poor motor recovery
Waters et al,
AMPR, 1994b
II 50 with
incomplete
tetraplegia
AIS
AMS
ASS
Power in key muscles
AMS at
follow-up
LEMS
Communit
y
ambulation
Muscle
power at
follow-up
1 year
0/5 patients who were motor complete with the presence of
sacral sensation unilaterally recovered any L/E motor function
8 initially motor complete subjects retained presence of
bilateral sacral sharp/ dull sensation, mean LEMS increased to
12.1 7.8
87% (20/23) of patients having a LEMS ≥ 10 at 1 month were
community ambulators using crutches and orthoses at 1 year
follow-up
Waters et al,
AMPR, 1995
II 278 injuries
AMS
Injury etiology
(penetrating vs. non-
penetrating)
AMS at
follow-up 1 year
Although the etiology injury determined the severity of the
initial neurologic deficit, it did not provide additional
predictive power concerning the amount of motor recovery
Waters et al,
Paraplegia,
1995b
II 32 injuries
Injury etiology (stab
wound)
AMS
1 year
Average change in AMS from admission to follow-up was
16.9 overall, 6.0 in complete tetraplegics and 22.7 in
incomplete tetraplegics.
No difference in motor recovery noted between current series
of stab wound patients and those with injuries from other
etiologies.
Wu et al,
AMPR, 1992
III 34 with C4-
7 motor
complete
injuries
Power in key muscles of
U/E
Frankel grade
Power in
U/E
muscles at
follow-up
1 year
Subjects with no initial motor power in the ZPP had a much
better chance of recovery to grade 3/5 strength by one year
post-injury if they developed grade 2/5 strength by 3 months
post-injury (p < .001)
Zariffa et al,
Spinal Cord,
2011
III 399 thoracic
motor-
sensory
complete
AIS grade
AMS
SCIM
Neurological level
Sensory level (ASS)
AIS at
follow-up
AMS at
follow-up
ASS at
follow-up
1 year
Any length of a sensory ZPP at baseline does not predict
either a change in sensory level or a change in LEMS
A baseline sensory ZPP of three or more segments correlated
with an increase in the proportion of subjects who converted
to a grade of AIS-B or greater by 48 weeks
Sensory levels and neurological levels exhibited minor
changes but most subjects remained within one segment of
their initial injury level
Motor recovery occurred predominately in subjects with low
30
AMS: ASIA Motor Score; ASS: ASIA Sensory Score; AIS: ASIA Impairment Severity grade; ECR: extensor carpi radialis; FIM:
Functional independence measure; LEMS: Lower extremity motor score; LoE: level of evidence; MMT: manual muscle test; MPSS:
methylprednisolone sodium succinate; MR: Marked Recovery; primary endpoint of study; NLI: Neurological Level of Injury; SCI: spinal
cord injury; SCIM: Spinal cord independence measure; TUG: timed up and go; UEMS: Upper extremity motor score; ZPP: zone of partial
preservation; 10MWT: 10-meter walk test. *Defined as improvement of at least 2 grades from AIS at baseline to Modified Benzel Scale at
week 26.
thoracic SCI
31
2.4.1.2 Functional Outcome
Twenty-three articles were identified addressing the clinical predictors of functional outcome
after SCI (Table 2-3). A diverse array of outcome measures were used across these studies
including: ambulatory status in 13 studies76,133,134,136,137,148,151-157
, Functional Independence
Measure (FIM) in 7 studies129,132,150,158-161
, Walking Index for Spinal Cord Injury II (WISCI II)
in 1 study162
, 10 Meter Walk Test (10MWT) in 1 study76
, Timed Up and Go test in 1 study76
,
Modified Barthel Index in 1 study163
and the 6 minute walk test in 1 study162
. Six of the studies
were classified as level II evidence while the remaining 17 studies were classified as level III.
As seen with neurological recovery, baseline neurological exam features were the predictors of
functional recovery most frequently observed. Overall, a less severe pattern of neurological
injury76,133,136,137,148,151,155,158,159,162-165
, absence of a delayed plantar response (DPR)154,157
,
presence of hyperactive deep tendon reflex responses152
and a more caudal location for the
NLI134,158,160
were associated with a superior degree of functional recovery after SCI.
One study found no impact of injury etiology (gunshot related vs non-gunshot related) on FIM
scores at rehabilitation discharge161
.
One study addressed the influence of gender, finding greater FIM improvements in males with
motor complete lesions129
, and 3 studies addressed the influence of age, with a consensus finding
that older age is associated with worse functional outcome at long term follow-up132,150,158
.
32
Table 2-3: Evidentiary Table Summarizing Clinical Predictors of Functional Outcome
Paper LoE Number of
Patients
Predictive factors
assessed
Primary
Outcome(s)
Follow-
up
Findings
Burns et al,
AMPR, 1997
III 105 ASIA
C/D cervical
injuries
AIS
Age
Ambulation
NR
91% of ASIA C patients younger than 50 years became
ambulatory by discharge, versus 42% of those older than 50
years
All ASIA D patients became ambulatory by discharge
Calancie et al,
Clinical
Neurophysiol
ogy, 2004
II 229 injuries
Tendon reflex responses Ambulation
1 year
Larger deep tendon response amplitudes and the presence of
the crossed adductor response to patellar taps at the acute
stage were highly predictive of functional motor recovery
following SCI (motor-complete accuracy: 100%; motor-
incomplete accuracy: 91%)
Cifu et al,
APMR, 1999
III 375 cervical
injuries
AMS
FIM
Age
FIM at
follow-up 4-10
months
Younger patients’ scores on the FIM motor subscale improved
significantly more than did the older groups
Crozier et al,
APMR, 1991
III 27 Frankel
B injuries
Muscle strength
(quadriceps)
Ambulation
1 year
Motor incomplete SCI patients who recovered to a >3/5
quadricep strength by 2 months post injury had an excellent
prognosis for subsequent ambulation by 6 months post injury
Furlan et al, J
Neurotrauma,
2009
II 499 injuries
Age
FIM
FIM at
follow-up 1 year
Patients >65 years experienced greater functional deficit
(based on FIM scores) than younger individuals despite
experiencing similar rates of sensorimotor recovery
Katoh et al,
Paraplegia,
1995
III 21 with
Frankel B
injuries
Sensory sparing Ambulation
1 year
Patients with spinothalamic sensory preservation between the
NLI and the last sacral dermatomes had higher rates of
ambulation as compared to those with dull sensation in the
sacral dermatomes
Kay et al,
APMR, 2007
III 343 injuries
AIS
Level of injury
Age
FIM walking
instrument 2
months
At discharge from in-patient rehab, 0.9% of ASIA A or B
patients, 28.3% of ASIA C patients, and 67.2% of ASIA D
patients were ambulatory
Subjects with AIS grades C and D injuries were equally likely
to walk at discharge regardless of injury level
Age 50 years or older had a significant negative effect on
ambulation in ASIA D patients
Ko et al, III 50 injuries
Reflexes (BC, DPR,
CRM, AJ, KJ, NPR)
Ambulation
6-8
14 patients who became functional ambulators on f/u either
had no DPR or a DPR of only one day’s duration
33
Spinal Cord,
1999
weeks 35 subjects had a DPR of 2 days or longer duration and these
subjects were not ambulatory
Lazar et al,
APMR, 1989
III 78 injuries
MIS MBI
NR (≥ 2
months)
For quadriplegics the initial MIS positively correlated strongly
with functional status (MBI) at admission (p = .001), at 60
days, and rehab discharge (p = .001)
For paraplegics, the MBI total score at admission, discharge,
and at 30-day intervals did not correlate with early motor
function
For paraplegics the initial MIS positively correlated
significantly with MBI self-care sub-score at 60 days and at
discharge
Marino et al,
APMR, 2004
III 4338
injuries
U/E and L/E AMS
Total AMS
FIM
1.5
months
Use of separate ASIA upper extremity and lower extremity
motor scores improved prediction of motor FIM scores more
so than that of the total ASIA motor score (R2= .71 vs .59,
respectively)
Maynard et
al, J
Neurosurg,
1979
III 123 with
cervical
injuries
Frankel classification Ambulation
1 year
Of patients with sensory incomplete injuries at 72 hours, 47%
were walking at 1 year
Of patients with motor incomplete injuries at 72 hours, 87%
were walking at 1 year
McKinley et
al, AJPMR,
1999
III 217 injuries
Injury etiology (gunshot
wound)
FIM
Rehab
discharg
e
No significant difference in FIM score at discharge between
gunshot wound (FIM=94.8) and non-gunshot wound groups
(FIM=86.4).
Oleson et al,
AMPR, 2005
III 131 ASIA B
injuries
L/E and sacral pin prick
preservation
Ambulation
1 year
Baseline lower-extremity pinprick preservation (>50% of L/E
L2-S1 dermatomes) and sacral pinprick preservation at 4
weeks post-injury are associated with an improved prognosis
for ambulation at 1 year
Ota et al,
Spinal Cord,
1996
III 100 with
motor
complete
injuries
AMS
FIM
NLI
FIM at
follow-up NR
Sequential increase in mean FIM scores at follow-up with a
more caudal initial level of injury
Sipski et al,
AMPR, 2004
III 14, 433
injuries
AMS
FIM instrument score
Gender
FIM at
follow-up 1 year
Men had higher FIM motor scores at rehab discharge as
compared to women among those with motor complete lesions
except for those with C1-4 and C6 neurologic level
van
Middendorp
III 273 injuries
AIS TUG
10MWT
Ambulation 6 month
The AIS conversion outcome measure is poorly related to the
ability to walk. The ratio of patients with both recovery of
ambulatory function and AIS conversion differed significantly
34
et al, Spinal
Cord, 2009
between the acute phase AIS grade scores
van
Middendorp
et al, Lancet,
2011
III 492 injuries
Age
Motor scores of the
quadriceps femoris (L3)
Motor scores of the
gastrocsoleus (S1)
muscles
Light touch sensation of
dermatomes L3 and S1
Ambulation
1 year
Lower age and higher motor scores of the quadriceps femoris,
gastrocsoleus, and light touch sensation of dermatomes L3 and
S1 were the best predictors of independent ambulation at 1
year
Vasquez et al,
J Forensic
and Legal
Medicine,
2008
III 173 injuries
AIS grade Rate of
reaching
“functional
levels”
5 years
Only 6% (4/68) of AIS A patients showed neurologic
improvement, with no patients reaching functional levels
33% (9/27) of AIS B patients showed neurologic
improvement of whom 33% were functional
74% (38/51) of AIS C patients showed neurologic
improvement all of whom were functional
All ASIA D patients were functional at discharge
Waters et al,
AMPR, 1992
II 148 with
complete
thoracolumb
ar injuries
NLI
Ambulation
1 year
Of the 142 (96%) patients who remained complete injuries at
f/u, none with an intial NLI above T9 regained any L/E motor
function
38% of patients with an initial NLI at or below T9 had some
return of L/E motor function
20% of those with a NLI below T12 were able to achieve
reciprocal ambulation
Waters et al,
AMPR, 1994a
II 54 with
incomplete
paraplegia
NLI
LEMS
Ambulation
1 year
The amount of motor recovery was independent of the initial
NLI
All patients with a 1 month LEMS >10 with hip flexion or
knee extension >2/5 were able to ambulate in the community
Waters et al,
AMPR, 1994b
II 50 with
incomplete
cervical
injuries
LEMS Ambulation
1 year
87% (20/23) of patients having a LEMS ≥ 10 at 1 month were
community ambulators using crutches and orthoses at 1 year
follow-up, compared to 11% (3/27) of patients with a LEMS <
10
Weinstein et
al, J of Spinal
Cord Med,
1997
III 36 SCI
subjects
DPR Ambulation
4
months
DPR associated with poor prognosis for recovery of
ambulation
High correlation of the DPR with motor complete injuries
35
Zorner et al, J
Neurotrauma,
2010
III 90 motor
incomplete
injuries
LEMS
AIS
WISC II
6 minute
walk test 6
months
Initial, higher LEMS was most predictive, alone or in
combination with a higher SSEP or with the AIS, for
ambulatory capacity
AJ: Ankle jerk; AMS: ASIA Motor Score; AIS: AIS grade Impairment Severity grade; BC: bulbo-cavernosis; CRM: cremasteric; DPR:
Delayed plantar response; FIM: Functional independence measure; KJ: Knee jerk; LEMS: Lower extremity motor score; LoE: level of
evidence; MBI: Modified Barthel index; MIS: Motor index score; MMT: manual muscle test; MPSS: methylprednisolone sodium
succinate; MR: Marked Recovery; primary endpoint of study; NLI: Neurological Level of Injury; NPR: Normal plantar response; SCI:
spinal cord injury; SSEP: somatosensory evoked potential; TUG: timed up and go; 10MWT: 10-meter walk test
36
2.4.1.3 Complications
Fourteen articles specifically examined clinical features predicting complication occurrence post
SCI (Table 3-4). Six of these studies were classified as providing level II evidence, and 8 studies
were considered level III. Ten studies examined complication development in the sub-acute or
chronic period while the remaining 4 considered complications during the acute hospital period.
Neurological exam features were consistently important in predicting complication occurrence
with more severe patterns of neurological injury31,166-173
and a more cephalad neurologic level of
injury166,167,169-173
associated with a higher incidence of complications.
One study identified violence related etiologies (either gunshot or stabbing related), with a higher
incidence of complications167
.
Increasing age appeared as a positive predictor of complications across 8 of the studies
examined21,166,167,169-172,174
. One study showed no effect of age on complications175
. As regards
gender, while no study demonstrated significant differences in the overall incidence of
complications between men and women, 2 studies demonstrated gender specific differences in
the profile of complications experienced170,176
. A single study demonstrated no influence of
racial background on complication development177
.
37
Table 2-4: Evidentiary Table Summarizing Clinical Predictors of Complications
Paper LoE Number of
Patients
Predictive factors
assessed
Primary
Outcome(s)
Complication
assessment
Findings
Haisma et al,
J Rehabil
Med, 2007
II 212 injuries
Age
Level of Injury
Severity of Injury
Cause of injury
Complication
occurrence Rehab up to 1
year post
admission
Urinary Tract Infection and Decubitus Ulcers were
most common complications encountered (47% and
36% respectively)
Increased age, traumatic lesions, cervical level of
injury and AIS A injuries were associated with greater
risk of secondary complications
Werhagen et
al, Spinal
Cord, 2004
III 402 injuries
Age
Gender
Level of Injury
Severity of Injury
Neuropathic
Pain
occurrence After Acute
discharge 2
months-
46years
Increased age associated with increased risk of
developing neuropathic pain post injury
McKinley et
al, APMR,
1999
III 6,776
injuries
Age
Gender
Level of Injury
Mechanism of
Injury
Severity of Injury
Complication
occurrence Chronic
period1-20
years post SCI
Decubitus ulcers, AD, pneumonia and atelectasis were
most common complications
Risk factors included AIS A injuries, tetraplegia, older
age, violent injury mechanism and concomitant illness.
Aito et al,
Spinal Cord,
1991
III 588 injuries
Lesion Severity Complication
occurrence Acute period
up to 60 days
AIS A patients had a greater risk of respiratory
complications, decubitus ulcers and heterotopic
ossification.
Injury severity had no impact on urinary complications
or thrombo-embolic complications.
Chen et al,
APMR, 1999
III 1649
injuries
Age
Gender
Injury Severity
Level of Injury
Complication
Occurrence During Acute
Rehab
Admission
Complete injuries associated with increased incidence
of decubitus ulcers
Complete cervical injuries associated with greatest risk
of AD.
Patients less than 20y.o. had reduced rate of DVT (5%)
as compared to those >20y.o.(9.8-15.3%)
Older age, cervical level of injury and complete lesions
associated with an increased risk of respiratory
complication development.
Greater incidence of GI hemorrhage in patients over
age 50.
38
DeVivo et al,
Arch Neurol,
1990
II 886 injuries
Age Complicati
on
occurrence Chronic period
up to 2 years
Patients ≥61 demonstrate 2.1x greater risk of
pneumonia, 2.7x greater risk of GI hemorrhage, 5.6x
greater risk of PE and 16.8x greater risk of
nephrolithiasis as compared to those between the ages
of 16 and 30.
Levi et al,
Paraplegia,
1995
III 353 injuries
Injury Severity
Level of injury
Age
Gender
Complication
occurrence 0-44 years post
injury
Multivariate analysis demonstrated: Lower age at
injury associated with increased risk of spinal
deformity, male sex increase sexual problems, female
sex and fractures/spinal deformity, complete injuries
decubitus ulcers as well as urinary tract infections,
cervical injuries spasticity
Dryden et al,
Spinal Cord,
2004
III 233 injuries
Injury Severity Complication
Occurrence Chronic period
up to 6 years
93.2% of complete patients were treated for
complications as compared to 47.1% of patients with
incomplete injuries.
Krassioukov
et al, J
Neurotrauma,
2003
III 58 injuries
Age Complication
Occurrence Within 2
months of
injury
Incidence of complication for patients ≥60 y.o. 14.3%
vs. 6.7% in patients <60 y.o. (p=0.42)
No difference in profile of complications between age
groups
Furlan et al, J
Neurotrauma,
2005
II 55 injuries
Gender Complication
Occurrence Within 2
months of
injury
Overall incidence of complications 44.7% amongst
males and 52.9% amongst females (p=0.77)
Trend towards higher incidence of DVT and
depression amongst women.
Meade et al,
APMR, 2004
II 628 injuries
Race (African
American and White
Caucasian)
Complication
Occurrence During acute
hospital and
rehab stay
No difference in the incidence and the profile of
complications between African American and White
Caucasian SCI patients.
Hitzig et al,
AJPMR, 2008
II 781 injuries
Age
Years post injury
Injury Severity
Level of Injury
Complication
Occurrence Chronic phase
1-60 years
Multivariate Analysis demonstrated:
Increased age assoc. with increased odds of
cardiorespiratory complications; Increased age
associated with decreased odds of AD, bladder
infections and heterotopic ossification.
Complete injuries assoc. with increased odds of
decubitus ulcers, AD and bladder infection.
Cervical injuries assoc. with increased odds of AD
Vidal et al, III 884 injuries
Age
Gender
Decubitus
Ulcer Chronic
Patients with DU more likely to be frankel A and older
than those without DU.
39
Paraplegia,
1991
Level of Injury
Injury Severity
Education
Psychiatric Co-
morbidity
occurrence unspecified Those with DU more likely to have psychiatric illness
and addiction problems as compared to those without
DU.
No association with gender, education.
Groah et al,
Spinal Cord,
2001
II 545
Injury Severity
Level of Injury
Incidence of
Cardiovascula
r disease Min 20 years
After Age Adjustment Rates of CV disease were 35.3,
29.9 and 21.2 per 1000 in Cervical Frankel A-C,
Thoracic/Lumbar Frankel A-C and all Frankel D
groups respectively.
DU: Decubitus Ulcer; AD: Autonomic Dysreflexia; CV: Cardiovascular; DVT: Deep Venous Thrombosis; AIS grade: ASIA Impairment
Scale Grade;
40
2.4.2 Discussion
For purposes of focusing the conclusions of this review, we have used the results of our analysis
to answer the 3 pre-formulated research questions described in the methods section.
How does neurological examination at admission relate to outcome after traumatic SCI?
2.4.2.1 AIS grade & neurological outcome
Injury severity is historically considered to be the most important predictor of future neurological
outcome after traumatic SCI. This is typically documented as the initial neurological exam in the
form of AIS grade or Frankel grade. In general, the extent of anticipated recovery diminishes as
the injury severity increases. Approximately 10-15% of individuals designated initially as having
complete lesions (AIS grade A) will convert to incomplete status, with only 2% converting from
complete (AIS A) to AIS D16,178
. AIS grade A tetraplegic patients will experience, on average, a
12 point improvement in AMS (of a total of 100 points) over the course of follow-up. For AIS
grade B patients, the mean AMS recovery at one year post injury is approximately 28 points for
tetraplegics, with one third generally remaining motor complete, one third converting to AIS
grade C and one third converting to AIS grade D or E16,178
. For AIS grade C patients, the mean
AMS recovery at one year post injury is approximately 43 points for tetraplegic patients with
approximately 70% converting to AIS grade D or E16,178
. The motor gains and conversion rates
are more limited with AIS grade D injuries. This is the result of a ceiling effect, an inherent
limitation with this grading scale, whereby only 4% of AIS grade D patients convert to AIS
grade E at one year16,178
.
2.4.2.2 Neurological Level of Injury & neurological outcome
In addition to injury severity, initial neurological level of injury is an important component of the
neurological exam that has been shown to be predictive of outcome. From a neurological
outcome perspective, the average improvement in AMS 1 year after AIS grade A SCI is 9.6
points (out of a maximum of 100) for cervical as compared to 2.6 points for thoracic/lumbar
complete lesions16
. This finding is supported by Coleman and Geisler’s analysis of the Sygen
trial data, which revealed that amongst AIS grade A patients, the cervical subgroup had a higher
frequency of marked neurologic recovery as compared to the thoracic one (15.5% vs. 7.0%)149
.
41
The explanation for this difference in recovery has been related to our inability to manually test
for motor recovery between T2 and T12 thus allowing structural repair to occur without clinical
motor change appreciated. As a result, patients with an AIS grade A injury, high thoracic SCI
(between T1 and T9) are unlikely to experience significant improvement in ASIA motor scores
at follow-up, since any small degree of motor recovery that does occur will not be captured by
this outcome measure134
. Similarly, these same patients are unlikely to experience conversion to
AIS grade C or D status at follow-up, since this would require some degree of motor recovery.
Another possible explanation for the reduced recovery is, in light of the mechanical
reinforcement provided by the rib cage in the thoracic spine, the biomechanical forces involved
in thoracic complete SCI are likely to be larger than in cervical SCI, potentially leading to a
greater degree of neurological tissue destruction. Overall, there seems to be consensus across the
literature that patients with complete thoracic injuries experience a reduced potential for
neurologic recovery at follow-up as compared to cervical complete patients. However, the same
cannot be said in comparing incomplete cervical to thoracic patients, who have demonstrated
similar degrees of neurological recovery across the literature. As an example, in the same
analysis of the Sygen data described above, there was no difference in marked neurologic
recovery between the cervical and thoracic groups for patients with incomplete injuries149
.
2.4.2.3 AIS grade & functional outcome
The severity of initial injury pattern impacts the extent of functional recovery in a similar
fashion. For patients with motor complete SCI, the likelihood of future ambulation is limited.
Waters et al reported that 5% of complete paraplegics were walking at 1 year, and that none of
61 complete tetraplegic patients were ambulatory in spite of 10% converting to motor incomplete
status134,135
. This is consistent with the work of Kay et al who found that zero out of 135
complete tetraplegic patients and 2 out of 84 complete paraplegic patients (2.4%) returned to
walking at discharge from rehabilitation158
. For motor incomplete SCI, the prognosis for
ambulation is more optimistic but remains varied throughout the literature. At discharge from
rehabilitation, Kay et al have reported ambulation rates for ASIA C tetraplegic and paraplegic
patients at 27.3% and 31.3% respectively, implicating that future ability to walk is not dependent
on neurological level of injury in the context of ASIA C injuries158
. These rates are less
optimistic than those published by others including Waters et al. who found that 67% of ASIA C
tetraplegics ambulated at discharge from rehabilitation137
. Disparities in rates between individual
42
studies can be explained based on differences in timing of initial assessment, definition of
“walking”, and duration and intensity of inpatient rehabilitation. In general, 1.5 to 6 times as
many ASIA D patients are ambulatory at discharge from rehab as compared to ASIA C
patients151
. Ambulatory rates for ASIA D patients at discharge from rehab range from 67-100%
and appear to be independent of the neurological level of injury158
.
2.4.2.4 Neurological level of injury & functional outcome
From the standpoint of functional outcome, although the level of injury has little impact on the
likelihood of walking for patients with complete injuries, it has shown to significantly relate to
the ability to function independently and to perform activities of daily living. In an attempt to
link motor level to function status, Ota et al found that for patients with complete injuries and an
initial level at C4, 5, 6, 7 or 8, there was a sequential increase in mean FIM scores at follow-up
with a more caudal initial level of injury160
. These incremental improvements in FIM score
translate into meaningful improvements in self-care, eating, dressing, toileting and transferring
to/from a wheelchair160
.
2.4.2.5 Neurological exam & complications
There is consistency in the finding that injury severity influences the overall incidence of
complication development during both the period of acute hospitalization and during the chronic
phases post SCI, with more severe injuries portending a higher likelihood of complication
development. In studies evaluating the impact of acute factors on individual complication
occurrence, complete injuries (AIS A or Frankel A injuries) have routinely been associated with
an increased risk of decubitus ulcer development, urinary tract infections, respiratory
complications, autonomic dysreflexia, venous thromboembolism and heterotopic
ossification31,166-172
. Haisma et al reported that as compared to patients with incomplete injuries,
those with complete injuries were associated with 1.73 and 1.81 times greater odds of decubitus
ulcer and urinary tract infection development respectively166
. Chen et al noted that the incidence
of autonomic dysreflexia decreases sequentially according to injury severity with the incidence
among complete patients 12.8%, sensory incomplete (AIS B) patients 8.7% and motor
incomplete (AIS C) patients 4.3%169
. Regarding respiratory complications, including pneumonia
or atelectasis, the same study reported a 16.3% incidence in complete patients and 7.2%
incidence in patients with incomplete injuries169
.
43
While injury completeness is of significant importance in predicting complication occurrence, in
many instances this relationship seems to be moderated by the neurological level of injury. More
specifically, complete injuries are associated with a greater incidence of complication
development when paired with a cervical neurological level of injury as compared to a thoracic
or lumbar neurological level. As an example, Chen et al demonstrated that 23.1% of patients
with complete paraplegia developed a decubitus ulcer during rehabilitation versus 39.5% of
patients with complete tetraplegia169
. Similarly, with respect to respiratory complications, the
incidence among complete paraplegics at 2 years follow-up was 2.0% versus 9.8% among
complete tetraplegics. The same phenomenon has been shown for cardiac complications
including autonomic dysreflexia171,173
. Overall this finding can be attributed to the progressive
increase in the burden of neurological and physiological impairment that occurs for complete
patients as the lesion level extends progressively cephalad. For patients with less severe,
incomplete injuries, the effect of neurological level on complication development is less
significant; this is likely a reflection of the fact that incomplete patients experience significant
neurological/functional recovery independent of initial injury level and are therefore less
susceptible to complications related to immobility and neurological dysfunction.
2.4.2.6 The special case of Zone of Partial Preservation
Another clinical entity that has been shown repeatedly to influence neurologic recovery in
complete injuries is the degree of motor/sensory preservation in the zone of partial preservation.
The term “zone of partial preservation” applies only to complete injuries and refers to those
myotomes and dermatomes caudal to the neurological level of injury that remain partially
innervated74
. From a motor perspective, for cervical complete lesions, 90% of muscles with
grade 1-2 /5 power in the zone of partial preservation one month post injury improved to 3/5 or
greater power by 1 year post injury135
. Conversely, muscles with 0/5 strength one month after
injury and located one neurologic level below the most caudal level having motor function
regained greater than 3/5 strength in only 27% of cases135
. Similarly, Wu et al, demonstrated that
100% of cervical complete patients by 1 year experienced recovery to at least grade 3/5 power in
the muscle innervated by the cord segment directly rostral to the NLI if power was at least 2/5 in
this muscle in the acute setting post injury142
. From a sensory perspective, in a study that
included C4-5 complete patients, extensor carpi radialis recovery was more likely to occur in
those patients with spared pin prick sensation in the C5 dermatome as compared to those with
44
absent sensation in this region141
. In general, sensory and motor preservation within the zone of
partial preservation positively influences neurological recovery after complete SCI.
2.4.2.7 The special case of Reflex pattern
In addition to loss of motor and sensory function caudal to the injury site, severe SCI also can
result in an immediate and prolonged depression of stretch reflex excitability in spinal segments
lying caudal to the injury152
. Hence the presence or absence of reflexes, as well as the evolution
of reflex recovery over time is important when attempting to predict recovery. Calancie et al
found that hyperactive deep tendon reflexes, which were confirmed by EMG, in addition to the
presence of a crossed-adductor response in the acute period post SCI, portended a favorable
prognosis for functional motor recovery in the long term152
. In contrast, the presence and
persistence of a pathological reflex, called the delayed plantar response, was associated with a
poor prognosis for recovery of ambulation in two studies154,157
. The delayed plantar response
requires an unusually strong stimulus applied by stroking with a blunt instrument upward from
the heel toward the toes along the lateral sole of the foot then continued medially across the
metatarsal heads157
. When the delayed plantar response is present, this stimulus is followed by
toe flexion and relaxation in a delayed sequence.
How do patient demographics relate to outcome after traumatic SCI?
2.4.2.8 Demographic variables & neurological outcome
The effect of age on neurologic recovery remains controversial. Cifu et al, found that patients
aged 18-64 had significantly larger gains in ASIA motor score improvement over time as
compared to injury matched patients older than 65132
. However, based on analysis of the
NASCIS III dataset, the Fehlings group found motor recovery to be independent of age, with
older age predicting superior sensory neurological recovery but worse functional recovery at
follow-up150
. The fact that older SCI patients may experience similar neurologic improvements
as compared to younger patients is mitigated by the common finding that this neurologic
improvement fails to be translated into an equivalent functional gain132,150,158
.
45
In analyzing the Model System data, Sipski et al demonstrated that improvement in ASIA motor
scores from admission to 1 year post injury were significantly greater for women than men,
implicating that women potentially possess a slightly greater natural potential for neurologic
recovery129
. However, in the absence of confirmatory evidence, and without a sound supporting
biological rationale, this finding must be interpreted with caution.
2.4.2.9 Demographic variables & functional outcome
As noted above, age has been found consistently to influence outcome across studies examining
long term functional status. Cifu et al and Furlan et al have shown that patients older than 65
experience smaller improvements in FIM scores over follow-up as compared to younger patients
matched for injury characteristics132,150
. In evaluating ambulation rates amongst AIS grade C
patients with tetraplegia, Burns et al reported that 91% of patients younger than 50 became
ambulatory whereas only 42% of those older than 50 achieved the same outcome. Although the
specific cutoff value varies, there is general agreement between studies that older age at the time
of SCI portends worse functional outcome.
In the same study that demonstrated superior neurological recovery for females, males with
motor complete lesions were found to have significantly greater FIM motor scores compared to
women at rehabilitation discharge129
. Given the paucity of evidence available, no specific
conclusions can be made regarding the effects of gender on outcome after SCI. Future
prospective studies, supported by biological rationale from the basic science realm, are necessary
to answer this question definitively.
2.4.2.10 Demographic variables & complications
Across the literature, older age is associated with an overall increased risk of complication
development21,166,167,169,171,172,174
. One study found that age, dichotomized at 60, was not a
significant factor predicting the overall occurrence of complications175
. However this study did
find a trend towards increased complications in the older group and was likely underpowered to
detect a statistically significant difference. When considering individual complications post SCI,
the impact of age is differential. The incidences of pulmonary infections and cardiovascular
disease have consistently been elevated in older as compared to younger patients. Chen et al
demonstrated the incidence of pneumonia over a five year post injury period to be 7.1% for those
46
older than 60, 4.1% for those between the ages of 40-59 and 2.2% for those younger than 40169
.
Considering age as a continuous variable, Haisma et al found that a per year increase in age at
time of injury was associated with 1.05 times greater odds of both pulmonary infection and
cardiovascular disease166
. The increased incidence of these complications amongst older patients
likely reflects the decreased physiological reserve seen within increasing age. Other
complications reported less frequently but noted to be increased in older age groups include
gastro-intestinal hemorrhage and nephrolithiasis21
. The development of decubitus ulcers was not
consistently related to age across the reviewed articles. In two studies, older age was negatively
associated with the development of autonomic dysreflexia, however the authors of these articles
were unable to provide a biological explanation for these findings166,167
.
No study has shown that the overall incidence of complications differs between men and women.
However, 2 studies have shown gender specific differences in the profile of complications
observed post SCI. Levi et al showed male sex to be associated with increased incidence of
subjective sexual problems and female sex to be associated with an increase in fractures and
spinal deformity at long-term follow-up170
. In addition, Furlan et al showed trends towards an
increased incidence of deep venous thrombosis and depression amongst women176
. However, in
both of these studies it may be that observed gender associations are non-specific to SCI patients
and are a reflection of the general population as a whole.
In a single study specifically examining the impact of race on clinical outcomes, no difference
was observed with respect to the incidence or profile of complications between African
American and Caucasian SCI patients177
.
How does injury mechanism or etiology relate to outcome after SCI?
Since the etiology/mechanism of SCI influences the degree of neural tissue destruction and
injury severity, one would expect this factor to have a significant impact on neurological and
functional recovery, in addition to complications after SCI. However, relatively few studies have
explored these clinical hypotheses in detail.
2.4.2.11 Injury Mechanism & Neurological/Functional Outcome
Using information from the Model Systems database, Marino et al found that individuals
with SCI that resulted from violent trauma were more likely to be complete than those with
47
nonviolent injuries and, if complete, were more likely to remain complete than those with
nonviolent injuries16
. In contrast however, Waters et al performed a similar analysis and
concluded that although injuries more severely disruptive of the spinal canal were more likely to
result in complete SCI, there was no significant difference in motor recovery based on type of
injury (penetrating vs. non-penetrating), type of spinal fracture, or bullet location138
. Exploring
violence related injuries in slightly greater detail Waters et al found that motor recovery after
injuries related to stab wounds was no different than for non-stab wound related injuries127
.
Similarly, McKinley et al showed that gunshot etiology was not a significant predictor of follow-
up functional status161
.
Overall, based on the available evidence, it remains unclear whether specific SCI etiologies (i.e
MVA vs. fall), differentially influence neurological or function recovery.
2.4.2.12 Injury Mechanism & Complications
One study associated violence related injuries with increased complications as compared to non-
violence related injuries167
. However, it is likely that the difference observed is secondary to the
confounding influence of the aforementioned finding that violent injuries were more likely to be
complete and, if complete, were more likely to remain complete than those with nonviolent
injuries.
2.4.3 Summary: The impact of clinical variables on outcome after SCI
Clearly the single most important clinical predictor of the three outcomes considered in this
section is the acute neurological examination, more specifically, the severity of neurological
injury as measured by the AIS or Frankel grade. Consistently patients with more severe injuries
at presentation experienced worse neurological and functional outcomes and a higher incidence
of complications. With respect to neurological level of injury, complete thoracic injuries are
associated with reduced motor recovery as compared to complete cervical injuries predominately
due to our inability to clinically test for motor recovery in the thoracic region. A progressively
more cephalad neurological level of injury is generally associated with reduced functional
outcome and increased complications. Increasing age has shown to be associated with reduced
functional outcome and increased complications but has no clear association with long-term
neurological outcome. Gender has had no demonstrable impact on neurologic or functional
48
outcome or on the overall incidence of complications post injury. However, individual studies
have shown that the profile of complications observed may be different between male and female
patients. There is currently no consistent evidence suggesting that specific injury mechanisms are
predictive of neurologic, functional or complication outcomes after SCI.
2.5 Literature Review Section 2: The impact of radiological variables on outcome after SCI
2.5.1 Results
The initial search resulted in 250 citations. Application of the inclusion and exclusion criteria
reduced this to 190 citations including those obtained after a secondary review of bibliographies.
After review of these, 18 relevant articles that identified radiological predictors of neurologic,
functional, or complication outcomes were identified and graded. Of the relevant articles, 15
provided predictors for neurologic outcome, 2 for functional outcome and 1 for complications.
As regards level of evidence, all of the included studies were designated as providing level I, II,
or III. While several MRI variables were found to be important, no significant CT related
predictors were discovered. The timing of acute radiological evaluation varied across studies,
with 7 reporting MRI completion prior to 72 hours and 11 studies reporting completion outside
of this window.
2.5.1.1 Neurological Outcome
The 15 articles identifying radiological predictors of neurological outcome are displayed in Table
2-5. Specific outcome measures utilized in these studies included Frankel classification in 6
studies179-185
, AIS grade in 2 studies186,187
, ASIA motor score or ASIA sensory score in 4
studies187-189
, “neurological deficit resolution” in 2 studies190,191
, motor index score in 1 study186
,
NASCIS motor and sensory scores in 1 study192
and trauma motor index in 1 study179
. Nine of
the studies were classified as level II evidence while the remaining 6 studies were classified as
level III.
MRI intramedullary signal characteristics were shown to be of importance in predicting
neurological outcome across all 15 of the studies examined. Most frequently these studies
evaluated qualitative evidence of intramedullary hemorrhage, edema or contusion.
Intramedullary hemorrhage was consistently the strongest negative predictor of neurological
49
recovery with the absence of abnormal signal consistently the strongest positive predictor of
neurological recovery. In 3 studies, the cranio-caudal length of intramedullary signal change was
found to significantly predict neurological recovery184,187,193
. Two studies associated focal spinal
cord swelling with reduced neurological recovery at follow-up181,188
Four out of the 5 studies that considered the impact of spinal cord compression on neurological
outcomes found increasing cord compression to be a negative predictor of
recovery180,181,184,186,188
. In only one of these studies was the degree of spinal cord compression
quantified using objective radiological measurement methods188
. Two studies evaluated the
impact of spinal canal size on neurological outcome; neither of these studies found a significant
association180,188
. The presence of a disc herniation or fracture/deformity was not associated with
neurological outcome in individual studies181,188
.
2.5.1.2 Functional Outcome
The 2 articles identifying radiological predictors of functional outcome are also displayed in
Table 2-5. In the respective studies, FIM motor score194
and the Japanese Orthopedic Association
Scale195
were the functional outcome measures assessed. Both studies were classified as
providing level III evidence.
The study by Fladers et al found the qualitative presence of intramedullary hemorrhage and
increased length of intramedullary edema along the longitudinal axis to be associated with worse
functional outcome194
. The second study by Yamazaki et al found the antero-posterior diameter
of the cervical spinal canal to be positively associated with functional outcome195
. In the same
study, the presence of T2 signal change, disc herniation and/or ossification of the posterior
longitudinal ligament had no significant impact of functional outcome.
2.5.1.3 Complications
One study identifying radiological predictors of pulmonary complications is displayed in Table
2-5196
. This study provided level 1 evidence associating increased MRI intramedullary lesion
length with an increased risk of pulmonary complications during the acute and chronic phases
post SCI.
50
Table 2-5: Evidentiary table summarizing radiological predictors of outcome
Paper LoE Number of
Patients
Predictive factors
assessed
Primary
Outcome(s)
F/U Findings
Aarabi et al,
J Neurosurg
Spine, 2012
I 109 -Length of IM Lesion,
MSCC, MCC
-Occurrence of
Respiratory
Complications
-Mean 9.5
months
post SCI
- In multivariate negative binomial model length of IM lesion
>40 mm associated with 2.04 RR of pulmonary complication
development
Bondurant et
al, Spine, 1990
II 37 - IM signal intensity
pattern
-1.5 Tesla MRI within 1
to 8 days of injury
-Frankel
classification
-TMI recovery
Mean 12.1
month
post SCI
-Four patterns of signal intensity were noted which correlated
to neurological outcome:
-Type 1(27% patients): decreased T2 signal consistent with
IM hemorrhage.
-Type 2 (43.2% patients): increased T2 signal consistent
with IM edema
-Type 3(8.1% of patients): Mixed T2 signal consistent with
contusion.
-No signal change (20%)
- Patients with Type 1 injuries showed no Frankel
improvement, whereas all Type 2&3 injuries showed at least
one grade improvement.
- Average TMI recovery: 10.3 for type 1, 21.1for type 2 and
38.4 for type 3.
Bouldin et al,
Spine, 2006
II 29 -Presence of IM
hemorrhage and length
of hematoma
-1.5 Tesla MRI within 5
to 12 days of injury
-Recovery rate
of AMS, ASS
and AIS grade
24-65
months
post SCI
-Presence of hemorrhage associated with decreased sensory
recovery, and trend towards decreased motor recovery.
-In patients with hemorrhage 5/17 improved at least 1 AIS
grade whereas 11/112 patients with no hemorrhage improved
at least 1 AIS grade.
-Length of hemorrhage <4mm positively associated with AIS
grade improvement.
Flanders et al,
AJNR, 1999
III 49 -Presence of IM
hemorrhage and length
of edema
-1.5 Tesla MRI within
72 hours of injury
-FIM Motor
score
Rehab
discharge
-Mean FIM motor score at rehab discharge was 39.8 amongst
patients w/o IM hemorrhage and 21.8 amongst those w/ IM
hemorrhage (p<0.01). Largest difference noted for self- care.
- Increased length of edema and more cephalad location
correlated with worse FIM motor.
Hayashi et al,
Paraplegia,
1995
III 31 cervical
SCI
-Severity of Spinal Cord
Compression, the canal
narrowing, IM signal
intensity changes.
-0.2-1.5 Tesla MRI: 17
patients before 2 weeks,
-Frankel
Classification
Mean
13months
post SCI
-Absence of IM signal intensity changes was associated with
improved outcome.
-Patients with severe spinal cord compression showed poor
neurological recovery
51
14 patients after 2 weeks
post SCI
Mascalchi et
al, Clinical
Radiology,
1993
II 32 injuries -IM signal change, cord
swelling, cord
compression,
fracture/deformity
-0.5 Tesla MRI within 4
days of injury
- Frankel
Classification
(Dichotomize
d Frankel A-
C and D and
E)
Mean 13
months
post SCI
- T2 hypointensity (hemorrhage) seen more frequently in
patients with Frankel A-C at follow-up (10/15) and T2
hyperintensity or no signal seen more frequently in patients
with Frankel D-E (13/13).
- Cord swelling seen in 11/14 Frankel A-C patients at follow-
up and in 4/13 Frankel D-E patients.
-Fractures/deformity seen in 12/14 Frankel A-C patients at
follow-up and in 4/13 Frankel D-E patients.
Miyanji et al,
Radiology,
2007
II 100 cervical
injuries
- MSCC, MCC, IM
edema, IM hemorrhage,
STI, HD, Spinal Cord
Swelling
-1.5 Tesla MRI within 24-
48 hours of injury
-ASIA motor
score
Mean 7.3
months
post SCI
-In multi-variable linear model IM hemorrhage, spinal cord
swelling and MSCC were independent predictors of AMS at
follow-up
-After adjusting for baseline AMS only IM hemorrhage and
cord swelling were independent predictors of f/u AMS.
O’Beirne et
al, Injury,
1993
III 44 injuries -IM signal characteristics
-1.5 Tesla MRI conducted
predominately within first
week post injury
-Frankel
Scale
NR -12/20 patients without Frankel scale improvement had IM
signal change consistent with hemorrhage or transection.
-All 17 patients that improved at least one Frankel grade had
no IM signal change or IM edema.
Ramon et al,
Spinal Cord,
1997
III 55 injuries -IM signal characteristics,
spinal cord compression,
spinal cord transection
-1.5 Tesla MRI completed
on average at 8 days post
SCI.
-AIS grade
-MIS
Mean 17.4
moths post
SCI
- Type 1 patients (IM hemorrhage): 0/14 experienced at least a
1 AIS grade improvement at f/u. Mean MIS improvement 5
points
-Type 2 patients (IM edema):11/15 experienced at least a 1
AIS grade improvement at f/u. Mean MIS improvement 21
points
-Type 3 patients (IM contusion): 6/10 experienced at least a 1
AIS grade improvement at f/u. Mean MIS improvement 52
points.
-Type 4 patients (Spinal Cord Compression):2/8 experienced
at least a 1 AIS grade improvement at f/u. Mean MIS
improvement 8 points.
-Type 5 patients (Transection): 0/2 patients experienced at
least a 1 AIS grade improvement at f/u. No MIS improve.
Sato et al,
Paraplegia,
1994
III 18 cervical
injuries
-IM signal characteristics
-1.5 Tesla MRI completed
within 3 to 48 hours post
injury
-Frankel
Scale
Mean 12
months
post SCI
-Patients with slightly low signal on T1 and low signal on T2
demonstrated the worst prognosis for neurological recovery
(0/6 patients experienced at least 1 Frankel grade
improvement).
Schaefer et al, III 57 cervical -IM signal characteristics -Motor Mean 7.8 -Group 1(IM signal consistent with hemorrhage): Median
52
J Neurosurg,
1992
injuries -1.5 Tesla MRI completed
within 2 weeks of injury
percent
recovery
months
post SCI
percent recovery was 9%
-Group 2(IM signal consistent with edema over ≥1 spinal
segment): Median percent recovery was 41%
-Group 3(IM signal consistent with edema over < 1 spinal
segment: Median percent recovery was 72%.
Selden et al,
Neurosurgery,
1999
II 55 cervical
injuries
-presence and length of
IM hematoma, length of
IM edema, CC and CC
from extra-axial
hematoma
-1.5 Tesla MRI completed
within 22 hours of injury
- Frankel
Scale
Mean 18.5
months
post SCI
-Presence of IM hematoma, CC and CC from extra-axial
hematoma negatively associated with Frankel grade recovery
at F/U.
- All patients with Frankel A at admission and IM hematoma
failed to improve at F/U whereas 32% of patients with IM
hematomas and less severe injuries improved at F/U.
- In multivariate analysis presence of IM hematoma, length of
IM hematoma, length of IM edema, and presence of CC from
extra-axial hematoma were significant negative predictors of
recovery.
Shepard and
Bracken,
Spinal Cord,
1999
II 199 injuries -IM signal characteristics
consistent with
hemorrhage, edema or
contusion.
-MRI within 72 hours of
injury
-NASCIS
motor, light
touch and
pinprick score
6 weeks
post SCI
-With respect to the outcome of motor recovery multi-variable
linear regression revealed trends towards a negative
association with IM hemorrhage (p=0.29), IM contusion
(p=0.87), IM edema (0.06)
Shimada et al,
Spinal Cord,
1999
II 75 cervical
injuries
-IM signal characteristics
-1.5 Tesla MRI performed
within 48 hours of injury
-“Neurologic
deficit
resolution”
6months/1
year post
SCI
-Four patterns identified:
-Type 1 (no signal change): 10/10 patients demonstrated
neurologic recovery
-Type 2 (T2 hyperintensity): 24/25 demonstrated neurologic
recovery
-Type 3(T2 hyperintensity with delayed T1 hypointensity):
16/30 demonstrated neurologic recovery
-Type 4(T2 hypointensity/hyperintensity): 1/10 demonstrated
neurological recovery
Shin, Yonsei
Medical
Journal, 2005
II 30 cervical
injuries
-IM signal characteristics
-1.5 Tesla MRI performed
within 7 days of injury
- Neurological
Level
improvement
-Motor/Sensory
percent
recovery
-6 months
post SCI
- IM edema associated with at least a 1 segment neuro
recovery in 7/15 patients.
- mixed IM signal associated with at least a 1 segment neuro
recovery in 5/14 patients.
-Motor recovery ratio 35.10 in patients with IM edema and
6.35 in patients with mixed IM signal(p<0.05).
-Sensory recovery ratio 50.27 in patients with IM edema and
17.14 in patients with mixed IM signal(p<0.05).
53
Takahashi, J
Ortho Surg,
2002
II 43 cervical
injuries
-IM signal characteristics
-1.5 Tesla MRI performed
within
“Neurological
Deficit
resolution”
1 year post
SCI
Three patterns noted
Type 1 (T2 hyperintensity): 10/10 demonstrated neurologic
recovery
-Type 2(T2 hyperintensity with delayed T1 hypointensity):
22/31 demonstrated neurologic recovery
-Type 3(T2 hypointensity with delayed T1 hypointensity): 0/2
demonstrated neurological recovery
Tewari et al,
Surg Neurol,
2005
III 40 injuries -IM signal changes
-MRI performed within
72 hours of injury
-Frankel Scale -Acute
hospital
discharge
- Patients with normal MRI or IM edema 6/6 recovered useful
neurological function (Frankel D/E)
-Patients with IM hematoma or contusion 13/13 failed to
recover useful neurological function (Frankel D/E)
Yamazaki et
al, Surg
Neurol, 2005
III 47 cervical
“central cord
sydrome”
injuries
-AP diameter of canal,
presence of canal stenosis,
spur formation, OPLL,
HD, Presence of IM
signal change.
- Timing of MRI NR
-Japanese
Orthopedic
Assoc. Scale
Mean 40.6
months
post SCI
-Divided into excellent recovery and poor recovery groups
-AP diameter 10.3mm in excellent recovery and 8.8mm in
poor recovery(p=0.04)
-No difference in outcome related to presence of canal
stenosis, spur formation, OPLL or HD.
MRI: Magnetic Resonance Imaging; IM: Intramedullary; MSCC: Maximum Spinal Cord Compression; MCC: Maximal Canal
Compromise; HD: Herniated Disc; OPLL: Ossification of Posterior Longitudinal Ligament; AP: Antero-posterior; CC: Cord
Compression; AIS: ASIA Impairment Scale; AMS: ASIA motor score;
54
2.5.2 Discussion
How do Qualitative Radiological Variables Relate to Outcome after SCI?
2.5.2.1 MRI Intramedullary Signal Characteristics
Intramedullary signal has been the most frequently studied MRI parameter for purposes of
predicting outcome post SCI. The initial qualitative classification system proposed by Bodurant
and colleagues outlined four major patterns of intramedullary signal change, each reflective of
specific pathological states179
. In this scheme, pattern number 1, defined as decreased signal
intensity on T2 weighted MRI sequences, represented spinal cord hemorrhage. All 10 patients
with this signal pattern were Frankel grade A at admission and failed to improve beyond this at
long-term follow-up. Pattern number 2, defined as increased signal intensity on T2 weighted
MRI, represented spinal cord edema and was associated with a substantially better prognosis; at
least a 1 Frankel grade improvement was seen in 10/11 patients that presented with neurological
deficit and this pattern of signal change. Pattern number 3 appeared as a mixed picture, with
regions of increased and decreased T2 signal change. This pattern was thought to reflect
petechial hemorrhage or spinal cord contusion surrounded by a region of edema. All three
patients with this imaging pattern recovered 2 Frankel grades at follow-up. Finally, 7 patients in
this study had no evidence of intramedullary signal change on the acute MRI. All of these
patients achieved perfect neurological status at follow-up. Overall, this study defined a severity
spectrum of imaging findings that correlates with the potential for neurological recovery post
injury. On one end of the spectrum, absence of intramedullary signal change predicts excellent
neurologic recovery. At the opposite extreme, intramedullary hemorrhage portends negative
neurological outcome. Signal change consistent with edema and/or contusion reflect lesions of
intermediate severity where the predicted potential for recovery falls between that predicted for
those with no signal change and those with intramedullary hemorrhage.
Follow-up studies have confirmed the predictive importance of the qualitative variables
originally studied by Bondurant and colleagues. Shimada et al demonstrated a similar sequential
stepwise reduction in neurological recovery in association with MRI findings of no
intramedullary signal change, signal change consistent with edema and signal change consistent
with hemorrhage190
. Bouldin et al demonstrated that 5/17 patients with MRI evidence of
55
intramedullary hemorrhage improved at least 1 AIS grade at a minimum follow-up of 2 years,
whereas 11/12 patients with no evidence of hemorrhage experienced this degree of recovery187
.
Even after adjusting for baseline neurological exam status, Miyanji and colleagues found that the
presence of intramedullary hemorrhage remained a significant predictor of motor neurological
recovery188
. In opposition to these findings, an analysis of imaging data collected in the NASCIS
III study revealed no significant association between MRI signals consistent with intramedullary
hemorrhage or edema and 6 week motor recovery192
. However, this study has been criticized for
using 6 week follow-up scores, a time point which is inadequate to fully observe individuals’
profile of motor recovery (plateaus at 4-6 months post injury).
Although the majority of studies evaluating the predictive significance of intramedullary signal
characteristics considered neurological outcomes, Flanders et al considered functional outcome
and showed that the mean Functional Independence Measure motor score at rehabilitation
discharge was significantly lower amongst patients with MRI evidence of hemorrhage (21.8) as
compared to those with no evidence of hemorrhage (39.8)194
. No studies that addressed the
impact of qualitative intramedullary signal change on complication outcomes were identified.
It is important to note that the patterns of MRI signal change which we associate with specific
pathological states, such as edema and hemorrhage, have been validated in both animal and
human studies. Preclinical experiments involving rats, cats and dogs have confirmed that during
the acute period following injury, increased signal intensity on T2 weighted images is associated
with the pathological presence of white and grey matter tissue edema/swelling as well as axonal
degeneration197-202
. In these same experiments the appearance of intramedullary blood has been
shown to appear hypo-intense on acute T2 weighted images. This hypo-intense appearance of
blood is thought to be secondary to the oxidative denaturation of deoxyhemoglobin which
commences shortly after hematoma formation. As the denaturation and decomposition of
hemoglobin continues, the appearance of the hemorrhagic lesion on MRI evolves. Several
clinical studies which have correlated acute MRI findings with post-mortem pathological
findings have by in large confirmed the findings of these animal studies203-205
. Although the
validity of these MRI signal patterns have been confirmed in numerous studies their reliability
remains substantially less studied. There is no existing study that has evaluated the intra-rater or
inter-rater reliability of detecting hemorrhage, contusion or edema on spinal MRI at any time
point after SCI. We can only extrapolate from studies evaluating the reliability of MRI in
56
detecting intra-cranial hemorrhage which have revealed high inter and intra-rater agreement206
.
The lack of a well-defined reliability profile for the discussed qualitative variables is one of the
largest limitations surrounding their use in clinical research.
2.5.2.2 Qualitative Spinal cord compression and swelling
From a qualitative standpoint the definition of spinal cord compression employed throughout the
literature is variable. Of four studies investigating this parameter on acute MRI, one study
reported severe and mild compression (severe defined by distortion of the cord diameter by
greater than approximately 2/3rds normal diameter), one study reported only severe compression
(no criteria defined) and 2 studies reported any degree of spinal cord compression. Given the
variability in its classification, it is unsurprising that the impact of spinal cord compression on
outcomes has found to be differential throughout the literature. Mascalchi et al found that the
proportion of patients with good neurological outcomes at follow-up that had spinal cord
compression on acute MRI was no different than the proportion of patients with poor
neurological outcomes who had this acute finding181
. In contrast, Hayashi et al found that as
compared to mild compression, severe compression was a negative predictor of Frankel grade
improvement at follow-up, however this effect was not quantified in the manuscript text180
.
Similarly, Ramon et al associated severe compression with diminished recovery, finding that of 9
patients with SCI and severe compression, 7 were motor sensory complete at final follow-up186
.
Given the heterogeneity in qualitative compression classification, as well as in its predictive
implications, it is apparent that compression is more suitably defined as a quantitative imaging
variable
A total of 2 studies evaluated the impact of the presence or absence of spinal cord swelling on
long-term neurological outcomes. Although methods for determining the presence of swelling
were not explicitly described in either paper, the authors make mention that this refers to the
presence of a focal widening of the cord at the site of injury. In the study by Miyanji et al,
swelling was found to be a statistically significant negative predictor of future motor
neurological outcome after adjusting for baseline motor neurological status and the presence of
intramedullary hemorrhage188
. The second study Mascalchi et al found that at a mean of 13
months post SCI, a higher proportion of patients experiencing unfavorable outcomes (Frankel
grade A-C) had swelling identified on the acute MRI as compared to those with a favorable
57
outcome (Frankel grade D and E) at follow-up181
. Overall swelling seems to be a negative
predictor of future neurological recovery, however relatively few studies have examined this
parameter in detail. Further, standardized methods for determining the presence of swelling do
not appear throughout the literature, a fact that diminishes its appeal for routine use in the
clinical or research settings.
2.5.2.3 Miscellaneous Qualitative Variables
Two studies demonstrated that the presence of a disc herniation on acute MRI had no impact on
neurological outcome, as measured by AMS recovery, or on functional outcome, as measured by
follow-up mJOA scores188,195
. Individual studies have also shown no association between
paravertebral soft tissue injury, qualitative presence of spinal canal stenosis, or the presence of
vertebral fractures/deformity on clinical outcomes post SCI181,188,195
.
How do Quantitative Radiological Variables Relate to Outcome after SCI?
2.5.2.4 Length of Intramedullary Lesion
Instead of commenting on the presence of hemorrhage or edema, several groups have quantified
the burden of this pathology by measuring the length of corresponding MRI signal change along
the longitudinal axis of the spinal cord. Selden et al found that increased length of intramedullary
signal compatible with either hemorrhage or edema was a significant predictor of a worse
Frankel grade at a mean of 18.5 months post injury184
. In attempting to define a potentially
clinically important threshold for length of intramedullary hemorrhage, Boldin et al reported that
lesion length less than 4 millimeters was uniformly associated with a less severe degree of initial
injury and improved AIS grade recovery at long term follow-up187
. However, in the same study,
a multivariate analysis controlling for the baseline neurological exam found that the length of
intramedullary edema was the only significant predictor of outcome with each millimeter
increase in length associated with a 1.15 times increased risk of retaining complete SCI. Schaefer
et al also explored the predictive significance associated with the length of intramedullary signal
consistent with edema193
. In this study the authors found that patients with edema spanning
greater than 1 vertebral body segment experienced a median percent motor recovery of 41% as
compared to 72% amongst patients with edema spanning less than a vertebral segment.
However, both of these groups experienced comparatively favorable outcomes as compared to
58
patients with MRI evidence of intramedullary hemorrhage who collectively experienced a 5%
median motor recovery.
From the perspective of functional outcome, Flanders et al demonstrated that increases in the
length of intramedullary edema signal negatively correlated with FIM motor score at
rehabilitation discharge (Pearson correlation coefficient -0.36, p<0.05)194
. In addition to
considering the raw length of edema related signal present, the investigators also evaluated the
predictive significance of the location of this signal change, finding that extension of edema to a
more cephalad location negatively correlated with functional outcome. Finally, as regards
complication development, Aarabi et al concluded that intramedullary lesion length greater than
40 millimeters was associated with a 2-fold increased risk of experiencing a pulmonary
complication as compared to lesion length less than 20 millimeters196
. However in this article,
the authors did not specify the MRI characteristics of the intramedullary lesion measured. As a
result it is unclear whether hemorrhage, edema or both were considered in these measurements.
This is significant given that the predictive importance of these measures would likely differ
depending on the nature of the underlying pathology.
Theoretically, numerically quantifying the length of intramedullary signal change present on
acute MRI should provide a more precise estimate of the burden of neural tissue damage and
allow for more accuracy in prediction of long-term outcome. However, several limitations must
be noted when discussing these measurements. First, different authors have employed different
methods and used different MRI sequences for performing these measurements. As a result it is
often difficult to draw direct comparisons between study findings. Second, although the number
of studies exploring the predictive significance of quantitative methods is much less than the
number evaluating qualitative methods, based on the available evidence there is no conclusive
evidence that quantitative methods are of greater value in predicting future outcome. As
standardized methods for measuring intramedullary signal length are validated and accepted
throughout the field, the predictive utility of these measures are likely to improve.
2.5.2.5 Quantitative Spinal Cord Compression
Of the studies included in this review, 2 employed quantitative methods to determine the extent
of spinal cord compression. Both of these studies utilized a valid and reliable measurement
strategy developed by Fehlings and colleagues207,208
. This measurement value, called the
59
maximal spinal cord compression (MSCC), compares the antero-posterior spinal cord diameter at
the level of maximal compression with the cord diameter at the nearest normal levels above and
below, using the mid-sagittal MRI. Using this technique, Miyanji et al found that in a
multivariate analysis, including only imaging related predictors, increased MSCC was a
significant negative predictor of motor neurological status at long-term follow-up188
. However,
when considered in a second multivariate model including baseline motor neurological status,
MSCC was no longer found to be a significant predictor of outcome. In the second study, Aarabi
and colleagues found that while there was a trend towards increased pulmonary complications
amongst patients with higher MSCC values this finding did not reach significance196
.
2.5.2.6 Quantitative Spinal Canal Narrowing
In conjunction with developing MSCC to quantify the degree of spinal cord compression,
Fehlings et al also defined and validated the maximum canal compromise (MCC) measure,
which is a standardized method to quantify the extent of spinal canal compromise on acute
MRI207,208
. This is calculated by comparing the AP canal diameter at the level of maximum
injury with the AP canal diameter at nearest normal levels above and below, using the mid-
sagittal MRI. MCC was not found to be of importance in predicting motor neurological status or
pulmonary complications in two of the studies included in this review. A third article which
examined the association between the raw spinal canal diameter an functional outcome in
patients with central cord syndrome found that the mean diameter was significantly higher in
patients with excellent functional recovery (10.3 mm) as compared to patients with poor
functional recovery (8.8mm)195
. However since the later method uses only a single measurement
in isolation it is unclear whether this finding reflects true focal stenosis or simply congenitally
smaller spinal canal amongst patients with worse functional outcomes.
2.5.3 Summary: The impact of radiological variables on clinical outcomes after SCI
Presently, there is no clear consensus surrounding which acute radiological variables are the best
suited to predict long-term outcome after SCI. A 2007 evidence based review on neuroimaging
for SCI performed through the United States National Institute on Disability and Rehabilitation
Research concluded that “MRI findings of parenchymal hemorrhage/contusion, edema and
spinal cord disruption in acute and subacute SCI may contribute to the understanding of severity
60
of injury and prognosis for neurological improvement”60
. While this statement is rather vague in
its direction, it is a reflection of the current state of the evidence on this topic. Overall, however,
it is clear that qualitative intramedullary signal characteristics are the best studied imaging
variables for purposes of outcome prediction. Further, there is consistency across the studies
considered that signal pattern consistent with intramedullary hemorrhage is associated with poor
outcome whereas absence of any abnormal intramedullary signal change is uniformly associated
with excellent long-term outcome. There is also consistency that signal patterns consistent with
edema or contusion reflect injuries of intermediate severity in which the potential for recovery is
variable but generally better than patients with hemorrhage and worse than those with normal
cord appearance.
Apart from intramedullary signal characteristics, the remaining qualitative variables considered
were found to have inconsistent effects or had an insufficient body of evidence available to
support their use as predictive variables.
As regards the quantitative variables considered, the major challenge arises with respect to
standardization of measurement techniques. Techniques for quantifying the extent of
intramedullary signal change to provide a numeric estimate of the spinal cord damage appear
promising. However, none of these measures have been standardized or validated. Further, to
date, none of these measures have demonstrated clear evidence of superior predictive capacity as
compared to the discussed qualitative intramedullary signal characteristics. Of studies that did
use standardized and validated measurement techniques (i.e MSCC and MSC) there is
insufficient evidence that these are of independent importance with respect to long-term outcome
prediction.
2.6 Limitations of systematic review
The quality of evidence on this topic is varied with the majority of studies found to be either
retrospective or prospective cohort studies (levels 1, 2, and 3), with only one case-control study
(level 3) (Appendix 3). However, the consistency of findings across these studies reinforces the
predictive importance of many of the clinical and radiological variables discussed. Another
potential challenge is the lack of uniformity in definitions of outcome across the literature.
Although the International Standard ASIA indices are now accepted as the standard method for
recording neurological outcome, such uniformity does not exist in the reporting of functional
61
outcome measures. While it is true that different functional outcome measurement tools assess
different aspects of functional recovery, the objective was to define the predictors of functional
outcome in general, not simply as measured by a single outcome measure. Similarly, with
respect to complications, there was considerable heterogeneity in the reporting and
categorization of these events across studies. This point underlies the need for the development
of a standardized SCI complications reporting system for implementation in future clinical
studies. As acknowledged above, with respect to the MRI variables considered, one major
limitation is the lack of psychometric information available for these indices. For the qualitative
MRI variables associating specific patterns of intramedullary signal intensity with specific
pathological states (spinal cord hemorrhage, contusion or edema), there is a significant body of
pre-clinical and clinical literature establishing their validity. However there are no studies
available exploring the reliability of these characteristics. With the exception of MCC and
MSCC, the psychometric properties associated with the remaining qualitative and quantitative
MRI variables remain largely unexplored.
2.7 Conclusion
Given the clinical and pathological heterogeneity of SCI, identification of a single predictor that
is capable of accounting for the full extent of the outcome spectrum, is a formidable challenge.
Based on a systematic review of the literature, this chapter summarizes the constellation of
clinical and radiological variables relating to neurological and functional outcome, as well as
complication development. As mentioned in the introduction, defining these relationships will
help with the selection of candidate predictor variables in the creation of clinical prediction
models and has allowed for the identification of variety of new research opportunities, pertinent
to the topic of clinical prediction in the realm of traumatic SCI.
62
2.8 Rationale, Overarching Hypothesis and Specific Aims
It is apparent that the first few days after SCI, from both a diagnostic and therapeutic perspective,
are critical. To standardize terminology within the field, the International Campaign for Cures of
Spinal Cord Injury Paralysis consensus guidelines defined the first 3 days after injury as the
“acute” injury period, as discussed in chapter 1. During this acute period, the initial neurological
exam, as well as diagnostic imaging tests are performed to determine the injury severity and to
characterize anatomical injury characteristics. Based on this information, practitioners make a
number of treatment decisions related to the acute medical stabilization of the injured patient and
formulate a definitive management strategy, which often includes surgical decompression of the
spinal cord. Not surprisingly, this same acute period coincides with the time of greatest anguish
for an injured patient and their family as they face significant prognostic uncertainty. In this
setting, it is the responsibility of the clinicians involved to use acute diagnostic information,
coupled with the available scientific literature, to provide a prognosis and manage expectations
for the future.
From a research perspective, many trials investigating the effectiveness of a particular
pharmaceutical or surgical intervention in patients with SCI are underway46
. However it is
difficult to meaningfully analyze for treatment effects in such a diverse patient population
without defining homogenous subgroups178
. Such subdivision would require a detailed
understanding about how acute injury variables relate to patient outcomes in the long-term, so
that individuals could be classified at the time of hospital admission depending on their expected
outcome.
To achieve these clinical and a research goals, a number of studies have shown several
individual clinical, radiological and demographic variables to be of importance in predicting
clinical outcome after traumatic SCI73,74,130,141,147,149,150,152,153,160,163
. A comprehensive review of
these variables is presented in the first part of this chapter. When considering the studies
presented in the above review, it is apparent that analyses assessing the significance of
combining individual parameters to improve outcome predictions, are sparse in the literature.
Further, the few existing studies that employ a combination of variables to predict long-term
outcome, use information collected outside the initial injury period (greater than 3 days post
SCI) and are therefore less useful as acute clinical prediction tools156,162
Zoerner et al produced a
63
clinical algorithm to predict ambulatory capacity 6 months post SCI, based on 90 patients with
incomplete SCI (79% traumatic, 21% non-traumatic)162
. A more recent study from the European
Multicenter SCI study group generated a clinical prediction rule for future ambulatory outcomes
using age, motor scores from the quadriceps and soleus muscles and light touch sensations in the
L3 and S1 dermatomes, as predictor variables156
. This analysis was based on traumatic SCI
patients with neurological examination variables obtained up to 15 days after injury, with the
vast majority of examinations taking place after 3 days post injury.
These studies provide evidence that the use of a combination of variables may enable a more
reliable prediction of long-term outcome after SCI as compared to the use of single predictors in
isolation. However, a number of key limitations in the existing literature must be recognized.
Previous studies have demonstrated that extraneous patient and injury related variables in the
acute setting can interfere with the accuracy of the initial neurological examination, yielding
results that are discordant with those obtained later post admission148,209
. Hence, previous
prediction rules formed using data outside the acute period are less suitable for use at the time of
patient admission to hospital, a fact which diminishes their clinical and research utility. In the
controlled setting of a hospital ward several weeks after injury, there are few distractions
impeding the accuracy of exam findings; in this setting the complete ISNCSCI neurological
exam is likely sufficient for purposes of outcome prediction. However in the first hours and days
post injury, time constraints as well as factors such as distracting injuries, sedation and patient
fatigue may impede clinicians’ ability to obtain the desired quality of neurological exam. This
fact addresses the need for prediction models that reflect the realities and the challenges of
patient assessment during the acute period. Ideally, such models would incorporate parameters
that can be easily collected during the acute injury setting. As an example, collecting
electrophysiological measurements or obtaining a very precise ASIA motor score value may not
always be realistic. Further, the variables incorporated should be familiar to acute care
physicians and should be part of the routine diagnostic work-up performed to investigate SCI
patients. As an example, radiologic investigations, both CT and MRI alike, are important
diagnostic adjuncts, used by modern day clinicians, to make management decisions and outcome
predictions. None of the existing SCI prediction models incorporate radiological variables as
predictors; these would be important for consideration in future acute models.
64
From an outcome assessment standpoint, the majority of previous studies exploring the
predictors of outcome after SCI, including the studies mentioned above, have utilized walking or
motor recovery as the primary measure of long-term outcome133,151,152,154,156,164
. While these
outcomes are certainly of interest, it is of paramount importance to consider other dimensions of
functional and neurological status at long-term follow-up. In a large survey assessing quality of
life in over 200,000 SCI patients, regaining arm and hand function was the highest priority for
quadriplegics, while regaining bowel/bladder and sexual function were the highest priorities for
paraplegics71
. These results underscore the importance of incorporating assessment tools that
include multi-dimensional assessments of outcome. On the topic of outcome it is also noteworthy
that our review of the literature identified a paucity of studies exploring predictors of
complications occurrence during the acute in-hospital period. Since these complication events
are known to affect both mortality rates and rates of neurological recovery, achieving an adept
understanding of the interplay of factors affecting their occurrence is important for purposes of
quality improvement and optimization of long-term outcomes.
There appears considerable opportunity to contribute to, and improve upon, the existing body of
knowledge pertaining to the acute prediction of outcome after traumatic SCI. To address this
void, this thesis relies on acute care data obtained exclusively within 3 days of injury for all
patients, which is used to generate predictive models that can be used at the time of admission by
acute care practitioners and SCI researchers to estimate short-term and long-term clinical
outcomes. In addition to a wide range of clinical factors, this thesis also explores and
incorporates relevant information obtained from acute diagnostic imaging to predict outcome.
Lastly, this thesis makes use of several outcome measures at follow-up that assess a spectrum of
outcomes important to clinicians, researchers and patients alike.
2.8.1 Overarching Hypothesis
Given the information presented above, it is hypothesized that clinical outcomes can accurately
be predicted after traumatic spinal cord injury based on acute patient, injury and radiological
features, leading to improved communication in the clinical realm and facilitating the
classification of patients within future clinical trials. To fully explore this hypothesis, outcome
has been defined in 3 ways: neurological outcome, functional outcome and complications.
65
2.8.2 Specific Aims
To investigate the overarching hypothesis and to expand upon the existing body of literature, 5
interrelated projects have been completed each with a specific aim related to outcome prediction
after SCI. Each of these projects represents a single chapter within this thesis, with several
supplemental chapters included for purposes of additional description and discussion, as outlined
below:
Specific Aim 1: To perform a comprehensive literature review identifying important clinical
predictors of neurological, functional and complication outcomes. (presented above as Chapter
2)
Specific Aim 2: To develop and internally validate a regression model to predict long-term
functional outcome after SCI based on acute clinical and radiological features (Chapter 4)
Specific Aim 3: To evaluate the radiologic finding of cervical facet dislocation as a predictor of
neurological outcome after SCI (Chapter 5)
Specific Aim 4: To identify the combination of clinical variables that predict acute in-hospital
complications and to develop a model to predict their occurrence (Chapter 6)
Specific Aim 5: To investigate age as a potential effect modifier influencing the relationship
between acute neurological status and long-term functional outcome after SCI. (Chapter 7)
A description of the data sources used (Chapter 3), in addition to a general combined discussion
section (Chapter 8), have been included as separate chapters.
66
3 Chapter 3: Description of Datasets and Approach to Missing Data
67
3.1 Introduction
The analyses included in the ensuing chapters incorporate 2 prospectively accrued multicenter
SCI datasets. Depending on the specific details of the individual analyses, one or a combination
of both these datasets, were utilized. Below is a description of each dataset as well as an outline
of steps taken to combine the data to produce a single harmonized dataset.
3.2 Surgical Timing in Acute Spinal Cord Injury Study (STASCIS) Database
STASCIS was a prospective, multicenter, cohort study investigating the role of surgical timing in
cervical SCI119
. This study involved hospitals at seven institutions throughout North America
during a 7 year enrollment period between August 2002 and September 2009. All of the centers
included in this study were specialized and equipped to manage spinal trauma and acute SCI.
Since this study was primarily concerned with acute surgical intervention, patients arrived at the
study center and were enrolled within 24 hours of injury. At hospital arrival, acute neurological
examination was performed within 72 hours of injury according to the International Standards
for Neurological Classification of SCI (ISNCSCI)49
. The primary outcome point for this study
was 6 months, however data collection continued for up to 1 year for the majority of patients
enrolled. At both 6 month and 1 year follow-up points, ASIA neurological assessment and
functional outcome assessments were performed. In addition, data on acute in-patient
complications was available in this dataset.
3.3 North American Clinical Trials Network for SCI (NACTN) Database
NACTN is a clinical research consortium which consists of 8 North American university
affiliated departments of neurosurgery, each managing a high volume of SCI patients210
. Since
its creation in 2006, one of the principle mandates of NACTN has been to develop and maintain
a large prospective SCI registry to attain an improved understanding of the natural history of
recovery as well as to explore the factors predicting outcome. Similar to the STASCIS dataset,
the NACTN registry contains acute neurological examination data as well as acute imaging
related data. Follow-up neurological and functional outcome data are also available at 6 months
and 1 year after injury. Data on acute in-patient complications was not available in this dataset.
68
3.4 Combined/Harmonized Dataset
In order to maximize the sample size available for specific analyses the NACTN and STASCIS
datasets were combined. This was accomplished by identifying the common data elements
present in both datasets. The specific language used to define these common variables as well as
the coding schemes were modified wherever necessary to ensure harmony between the two
sources. Any data elements that were not common to the two datasets were excluded. With a
total of 426 patients contributed from the STASCIS database and 303 patients contributed by the
NACTN database, the combined harmonized dataset contained data on a total of 729 patients. A
complete list of acute and follow-up data elements included in the dataset can be found in Table
3-1.The combined dataset was used as the basis for analyses presented in Chapter 4 and 7. For
Chapters 5 and 6, only data from STASCIS was used in light of the absence of data fields in the
NACTN dataset pertinent to the study questions posed in those chapters.
69
Table 3-1: Variables included in Combined/Harmonized Dataset. Note, only variables found
in both the STASCIS and NACTN datasets were included in the final combined dataset. (Further information available in Appendix 4)
Panel A: Clinical and Radiological Variables Available from acute Admission
Dataset Gender Age Etiology Energy Admis. ASIA exam
Admis. GCS
MRI IM signal charac.
STASCIS X X X X X X X NACTN X X X X X X X Data present In C/H Dataset (/729)
729
100%
721
99%
714
98%
714
98%
729
100%
711
98%
312
43%
Admis: Admission; ASIA: American Spinal Injury Association; IM: Intramedullary; C/H:
Combined Harmonized; GCS: Glasgow Coma Scale;
Panel B: Treatment related variables
Dataset Steroid
Administration
Surgical
Decompression Surgical
Timing
Surgical
Details
STASCIS X X X X
NACTN X X X X
Data present
In C/H
Dataset (/729)
708
97%
674
92%
674
92%
674
92%
C/H: Combined Harmonized
Panel C: Outcome Variables
Dataset F/u
ASIA exam parameters
(6/12mos)
F/U
FIM
(6/12mos)
STASCIS X X
NACTN X X
Data present
In C/H
Dataset (/729)
458
63%
394
54%
C/H: Combined Harmonized; ASIA: American Spinal Injury Association; FIM: Function
Independence Measure;
70
3.5 Approach to missing data within the combined dataset
Once the combined dataset was created, the percentage of patient data missing for each variable
was quantified throughout. In general the pattern of missing data can be classified in one of three
ways as: missing not at random (MNAR), missing at random (MAR) and missing completely at
random (MCAR)211,212
. Missing not at random implies that the missingness of a given variable
X(1) is related to the specific value of variable X(1). For instance if the baseline neurological
status tends to be missing more frequently in patients with severe baseline injuries (AIS grade A)
this would represent a MNAR for the variable neurological status. Missing at random is the most
common pattern observed and implies that the missingness of a given variable X(1) is not related
to the value of X(1) but to the value of another patient variable X(2). For instance, if the baseline
neurological status tends to be missing more frequently in older patients and in men, this would
represent a MAR situation for the variable neurological status. Finally, in the case of missing
completely at random, missing values are a random sample from the complete population and are
not related to other known or unknown patient variables. Such missingness is rarely encountered
and when it does occur, is typically related to administrative errors where a random portion of a
dataset is deleted or misplaced.
Broadly speaking, there are 2 main approaches to dealing with missing data when considering
model development. The first is complete case analysis, which involves incorporating only
patients with data available for all of the variables of interest213,214
. From both a pragmatic and
methodological perspective, complete case analysis poses two main challenges. The first is that
since a large percentage of patients in most datasets will have at least one variable of interest
missing, this approach becomes very inefficient and often results in a dramatic reduction in the
sample size available for model generation. The second is that this approach may introduce bias
to the model. If the variables included in model have a significant proportion of values missing at
random this leads to biased estimates for model regression coefficients215
. Since MAR is the
most typical pattern of data missingness, and maximizing statistical power through sample size
preservation is a ubiquitous concern, complete case analysis is generally discouraged. The
alternative approach is implementing a statistical procedure to substitute the missing values with
plausible values to preserve sample size and avoid bias. While a variety of techniques have been
71
described for this purpose, each with their own set of strengths and limitations, the most accepted
and statistically robust technique presently available is multiple imputation216-218
.
Throughout a multiple imputations procedure, missing values are approximated estimating the
relationship between the missing and the available data through the generation of an imputation
model219
. This procedure continues in an iterative fashion until the imputation model converges,
generating a final dataset with missing values filled by final estimates. The process is repeated X
number of times generating X imputed versions of the datasets, each with slightly different final
estimates for the missing values. The variability in the estimates of the missing values between
the different imputed versions of the dataset provides an approximate measure of the uncertainty
of the accuracy of the estimated values. The individual imputed versions of the dataset or a
combination of these versions, are then used to generate the model.
In 2 of the ensuing chapters (Chapters 4 and 7) a multiple imputations procedure has been
utilized to estimate missing values amongst the utilized predictor variables to augment the
sample size and protect against bias. In the remaining chapters (5 and 6) a complete case analysis
was performed due to the fact that the proportion of missing data amongst predictor variables in
these analyses was small.
72
4 Chapter 4: A Clinical Prediction Model for Long-Term Functional Outcome after Traumatic Spinal Cord Injury Based on Acute Clinical and Imaging Factors
This chapter is modified from the following:
Wilson JR, Grossman RG, Frankowski RF, Kiss A, Davis AM, Kulkarni AV, Harrop JS,
Aarabi B, Vaccaro A, Tator CH, Dvorak M, Shaffrey CI, Harkema S, Guest JD and
Fehlings MG. A clinical prediction model for long-term functional outcome after
traumatic spinal cord injury based on acute clinical and imaging factors. J Neurotrauma
2012; 29:2263-2271.220
73
4.1 Abstract
To improve clinicians’ ability to predict outcome after spinal cord injury (SCI) and to help
stratify patients within clinical trials, a novel prediction model was created relating acute clinical
and imaging information to functional outcome at one year. Data were obtained from two large
prospective SCI datasets. Functional Independence Measure (FIM) motor score at one year
follow-up was the primary outcome. Functional independence (score ≥6 for each FIM-motor
item) was the secondary outcome. A linear regression model was created with the primary
outcome modeled relative to clinical and imaging predictors obtained within three days of injury.
A logistic model was then created using the dichotomized secondary outcome and the same
predictor variables. Model validation was performed using a bootstrap resampling procedure. Of
729 patients, 376 met the inclusion/exclusion criteria. The mean FIM-motor score at one year
was 62.9(±28.6). Better functional status was predicted by less severe initial ASIA Impairment
Scale grade and by an ASIA Motor Score >50 at admission. In contrast, older age and MRI
signal characteristics consistent with spinal cord edema or hemorrhage predicted worse
functional outcome. The linear model predicting FIM motor score accounted for 52% of variance
in the original dataset and 52%(95%CI:52-53%) across the 200 bootstraps. Functional
independence was achieved by 148 patients(39.4%). For the logistic model, the area under the
curve was 0.93 in the original dataset and 0.92(95%CI:0.92,0.93) across the bootstraps,
indicating excellent predictive discrimination. These models will have important clinical impact
to guide decision making and to help in the counseling of patients and families.
74
4.2 Introduction
Spinal cord injury (SCI) is a devastating and debilitating condition that affects all regions of the
world221
. The preponderance of those affected are young men, sustaining severe cervical injuries,
leading to devastating personal losses and imposing a significant economic burden on society as
a whole9,10,222
. In spite of the immense impact of SCI at a personal and societal level,
comprehensive treatment strategies, aimed at reducing the initial degree of neurologic injury and
improving patients’ functional capacity, are lacking112,223
.
There is a pressing unmet need to accurately predict, early after SCI, patients’ functional
outcome. During the first few days after SCI, definitive management strategies are formulated,
which often include aggressive surgical decompression of the spinal cord. This is also the time of
greatest anguish for an injured patient and their family as they face significant prognostic
uncertainty. From a research perspective, it is difficult to meaningfully analyze the efficacy of
novel interventions, without defining homogenous subgroups.
Unfortunately, physicians and researchers have little to scientifically guide their prediction of
outcome following SCI. The few existing studies often employ variables collected outside of the
initial injury period (greater than three days post SCI) and are therefore less useful as acute
clinical prediction tools156,162
. Further, the existing studies fail to incorporate imaging related
variables, which are known to be an important element of the diagnostic process in the modern
day.
To improve the ability of clinicians to predict long-term outcome in the acute clinical setting and
to aid in the stratification of patients within clinical trials, we aimed to create, and internally
validate, a clinical prediction model that relates clinical and imaging findings to functional
outcome at one year following SCI.
4.3 Methods
4.3.1 Data Source
These analyses are based on individual patient data from the combination of two prospective
datasets: the North American Clinical Trials Network for SCI (NACTN) database and the
75
Surgical Timing in Acute Spinal Cord Injury Study (STASCIS) database. As described in the
previous chapter NACTN is a prospective registry enrolling patients with acute traumatic SCI
from eight North American Centers since 2006. STASCIS was a prospective, multicenter, cohort
study investigating the role of surgical timing in SCI119
. The NACTN and STASCIS datasets
contain similar acute data elements as well as long term functional and neurological outcome
data at 6 and 12 month follow-up points. Patients with severe head injury or poly-trauma,
defined by an inability to participate in a standardized ASIA neurological assessment during
acute admission, were not enrolled in either database. After research ethics board approval at
each participating site was obtained, the datasets were harmonized based on their common data
elements to produce a single database.
For all patients, neurologic examination was performed at patient presentation in concordance
with the recommendations of the International Standards for Neurological Classification of
Spinal Cord Injury by a trained physician, nurse or research assistant224
. Injury characteristics
were then classified according to neurologic level of injury (NLI), American Spinal Injury
Association motor score (AMS), American Spinal Injury Association sensory score (ASS) and
the overall American Spinal Injury Association Impairment Scale (AIS) grade. To identify
neuroanatomical injury characteristics, patients underwent magnetic resonance imaging (MRI) of
the spine shortly after hospital presentation. All patients received appropriate medical support
which included permissive or induced hypertensive therapy (mean BP > 85 mm Hg) in
accordance with the most recent consensus guidelines95
. Methylprednisolone was used at the
discretion of the treating team, according to the recommendations of the second National Acute
Spinal Cord Injury Study (NASCIS II) and based on established guidelines105,108
. Decisions
surrounding surgery, including the timing of decompression, were made by the attending spine
surgeon in each case. Whenever possible, patients with radiological evidence of spinal cord
compression underwent decompressive surgery within 24 hours post SCI. All patients underwent
an individualized rehabilitation protocol, tailored to specific needs and injury characteristics.
Using the combined dataset, the cohort of interest was defined to include patients 16 years of age
and older, presenting to an acute care facility with traumatic SCI and neurological deficit (AIS
grade A-D). Only patients with a documented ASIA neurological examination performed within
three days of injury were included. This criterion was included to ensure that only acute
variables were used as predictors in the development of the prediction model.
76
4.3.2 Predictor Variables
Clinical and radiological predictors were chosen based on three main criteria: 1) author
consensus of predictive importance of given variable; 2) literature support for the predictive
importance of given variable; and 3) the ease with which the given variable could be collected,
interpreted and recorded within the day to day acute injury setting. Based on these criteria, four
variables were selected to form the basis of model construction: 1) acute ASIA Impairment
Severity grade, 2) acute ASIA motor score (dichotomized at a score of 50), 3) patient age at
injury, and 4) intra-medullary signal characteristics on the spinal MRI(Table 4-1). All four of
these predictors have demonstrated predictive significance in relation to long-term functional
outcome after SCI132,150,151,158,159,164,185,190,194,225
. The two examination-related predictor variables,
AIS grade and AMS, were assessed and recorded within three days of injury for all patients. If an
individual patient underwent more than one neurological examination within the first three days,
we utilized the first examination that took place at acute hospital admission. Spinal MRI was also
performed within three days of injury and read by a site specific neuro-radiologist. With respect
to MR image acquisition, a two-dimensional spin echo sequence used for acquiring the T1
images and a two-dimensional fast-recovery fast spin-echo sequence used for acquiring the T2
images. Age was not categorized but rather its full variability was preserved in the form of a
continuous predictor as is the recommended practice211,212
.
77
Table 4-1: Pre-specified Predictor Variables utilized in the development of prediction
models
Predictor Coding
Initial ASIA Impairment
Scale (AIS) Grade
AIS grade A=1
no motor or sensory function is preserved in the
sacral segments
AIS grade B=2
sensory but no motor function is preserved below
the neurological level and includes the sacral
segments
AIS grade C=3
motor function is preserved below the neurological
level, and more than half of key muscles below this
level have a muscle grade less than 3
AIS grade D=4
motor function is preserved below the neurological
level, and more than half of key muscles below this
level have a muscle grade of 3 or more
Initial ASIA Motor Score
(AMS)
AMS ≤ 50 = 0
AMS > 50 = 1
Age Continuous predictor
Spinal MRI Intra-
medullary Signal
Characteristics
No Signal=0
Signal consistent with spinal cord edema=1
Signal consistent with spinal cord hemorrhage= 2
4.3.3 Outcome and Follow-up
The primary outcome variable was FIM motor score at one year post injury(Table 4-2). The FIM
motor score consists of 13 items which assess functioning across four different domains that
include self-care, sphincter control, transfers and locomotion. The performance level for each
item is strictly defined and ranges in value from one to seven, where one indicates complete
dependence in an activity and a score of six or greater indicates that a patient is capable of
performing that activity independently, without supervision or help. The result is an ordinal
outcome variable with a minimum value of 13 and a maximum value of 91, with a larger value
implying superior functional outcome. This outcome measure has been shown valid and reliable
for use in the setting of SCI78,226
. If outcome assessment was not available at one year, six month
FIM motor score was used. Given that the majority of functional recovery occurs in the first six
78
months after injury, this practice is considered valid and has been used in previous studies156,227
.
As a secondary outcome variable, the primary outcome was dichotomized and all of the patients
who achieved a score of at least six for all of the FIM motor score items were identified. These
patients were considered to have achieved functional independence across the 13 items assessed,
as has been the practice in previous stroke related research228
.
Table 4-2: Functional Independence Measure (FIM) motor score utilized as primary
outcome of interest. All patients who achieved a score of 6 or greater for each of these 13 items
were classified as having achieved “functional independence”.
Functional Domain Individual Items
1. Self-Care A. Eating /7
B. Grooming /7
C. Bathing /7
D. Dressing-Upper Body /7
E. Dressing-Lower Body /7
F. Toileting /7
2. Sphincter Control G. Bladder Management /7
H. Bowel Management /7
3. Transfers I. Bed, Chair, Wheelchair /7
J. Toilet /7
K. Tub, Shower /7
4. Locomotion L. Walk/Wheelchair /7
M. Stair /7
Total Score /91
79
4.3.4 Statistical methods
4.3.4.1 Model Creation
All analyses were performed using SAS version 9.3 (Cary, NC, USA) and R statistical software
version 2.14.1. Multi-collinearity was assessed through multivariate linear regression variable
tolerance testing, with a variable tolerance threshold set at 0.4. All missing data amongst the
predictors was assumed to be missing at random and a multiple imputations procedure with ten
imputation iterations was performed using Markov chain Monte Carlo methods to account for
this missing data. This resulted in an augmented dataset, containing ten-times the original sample
size, all with complete data. Such imputation is recommended as being less susceptible to bias
and more efficient than performing a complete case analysis by dropping cases with incomplete
variables211,217
. Of note, this is the preferred method, suggested by the Food and Drug
Administration, for handling missing data in therapeutic trials229
. Using the imputed dataset, with
follow-up FIM motor score as the dependent variable and initial AIS grade, dichotomized AMS,
age and intra-medullary MRI signal characteristics as the pre-specified four covariates, a linear
regression model was generated for each of the ten imputation iterations. A multiple imputation
analysis procedure was then employed, which combines the results of the analyses for each of
the imputation iterations to provide a single set of parameter estimates. R-squared values from
each of the imputation iterations were averaged to quantify the overall model performance.
A binary logistic regression model was created with the dependent variable of functional
independence modeled relative to the same four predictor variables. The model was tested
through the construction of receiver operator characteristic (ROC) curves, for each of the
imputation iterations. The area under the curve (AUC) is a summary measure of the
discriminative ability of the model, with values between 0.90 and 1.00 indicative of excellent
predictive discrimination. AUC values from each of the imputation iterations were averaged to
create a single value quantifying model discrimination.
Model Validation
Internal validation of the linear and logistic prediction models was performed using a bootstrap
re-sampling procedure of the imputed dataset using 200 individual bootstrap replicates (each
with n=376). A “bootstrap” is a group of subjects selected through random sampling with
80
replacement from the original dataset230
. The sampling is continued within each bootstrap
replicate until it contains a number of subjects equal to the total number of subjects in the
original dataset. Once the 200 bootstraps were generated, the pre-specified linear and logistic
models were fitted to each, and a 95% confidence interval was generated about the mean of the
200 replicate parameters estimate values for each covariate. The mean R-squared and AUC
values across the bootstrap replicates were also generated and were assumed to be an unbiased
estimate of the true values.
Finally, the regression coefficients obtained from the internally validated linear and logistic
models were used to estimate FIM motor score and the probability of achieving functional
independence at one year follow-up for hypothetical SCI patients based on their unique clinical
and radiological characteristics at hospital admission.
4.4 Results
4.4.1 Study Population
Prospective data from the NACTN and STASCIS protocols were combined to produce a single
harmonized dataset of 729 patients. From this original cohort, 33 patients did not have a
neurological assessment performed within the first three days after injury and were excluded.
Demographic, injury and imaging characteristics for the remaining 696 patients are listed in
Table 4-3.
81
Table 4-3: Patient Characteristics at hospital admission
Characteristic Patients meeting
Inclusion/Exclusion
criteria
N=696
Cohort
without
adequate
Follow-up
N=320
Cohort with
adequate
follow-up
used in final
analysis
N=376
p-value*
Mean age 44.5 (±17.3) 46·1(±17.5) 43·2 (±16.9) p=0.03
Male gender 531 (76.3%) 237(74.1%) 294(78.2%) p>0.05
Initial ASIA Impairment
Scale Grade:
AIS grade A
AIS grade B
AIS grade C
AIS grade D
258(37.1%)
115(16.5%)
119(17.1%)
204(29.3%)
122(38.1%)
52(16.3%)
61(19.1%)
85(26.6%)
136(36.2%)
63(16.8%)
58(15.4%)
119(31.7%)
p>0.05
Initial ASIA Motor Score
>50:
217(31.2%) 93(29.3%) 124(32.9%)
p>0.05
Neurological Level:
Cervical
Thoracic/Lumbar
633 (90.9)
63 (9.1%)
286(89.4%)
34(10.6%)
347(92.3%)
29(7.7%)
p>0.05
MRI Intra-medullary
signal characteristics:
No signal change
Consistent with edema
Consistent with hemor.
43(14.8%)
149(51.6%)
97(33.6%)
19(15.1%)
65(51·6%)
42(33·3%)
24(14.7%)
84(51·5%)
55(33·7%)
p>0.05
Etiology:
Motor Vehicle Accident
Fall
Sport or Diving Accident
Other
283(40.7%)
247(35.5%)
70(10.1%)
96(13.8%)
137(42.8%)
115(35.9%)
24(7.5%)
44(13.8%)
146(38.8%)
132(35.1%)
46(12.2%)
52(13.8%)
p>0.05
Received Steroids 410(60.7%) 183(59.0%) 227(62.1%) p>0.05
Mean time to
decompressive surgery post
injury (hours)
75.3 (±429.4) 74.5(±429.4) 76.1(±338.9) p>0.05
*Comparing difference between group with and without follow-up
82
Figure 4-1: Flow chart of the study design
.
83
4.4.2 Functional Outcomes
FIM motor score evaluations were available in 310 patients at one year and in 66 patients at six
months, leaving a total of 376 patients on which to base the analysis (Figure 4-1). There were no
significant differences in initial AIS grade, AMS or MRI intra-medullary signal characteristics
between patients included and those excluded due to inadequate follow-up (p>0.05) (Table 4-3).
Of the patients included in the analysis, 227(62.1%) received steroids and 358 (96.5%)
underwent decompressive surgery. There were no significant differences in the proportion of
steroid administration or in the mean time to surgery between patients included and those
excluded due to inadequate follow-up (p>0.05) (Table 4-3). The mean FIM motor score at
follow-up was 62.9(±28.6) with incrementally higher mean scores seen with a progressively less
severe acute admission AIS grade (Table 4-4). One-hundred and forty eight patients (39.4%)
achieved functional independence (a score of at least six on all of the FIM motor score items),
while 228 patients (60.6%) failed to achieve this milestone.
Table 4-4: Mean Follow-up FIM motor score by baseline characteristics
Characteristic Follow-up FIM motor
score
p-value*
Initial ASIA Impairment Scale
Grade:
AIS grade A
AIS grade B
AIS grade C
AIS grade D
41·5
55·7
71·6
87·2
P<0·01
Level:
AMS≤50
AMS>50
51·0
87·7
P<0·01
MRI Intra-medullary signal
characteristics:
No signal change
Consistent with edema
Consistent with hemorrhage
83·3
71·8
49·4
P<0·01
*p-value denotes significance level for the difference in FIM motor score between the different
levels of each predictor
84
4.4.3 Model Development and Validation
Predictor variable parameter estimates, significance values and bootstrapped estimates are found
in Table 4-5, Panel A for the linear model and Panel B for the logistic model. For both models,
the mean parameter estimate values for each covariate across the 200 bootstrap replicates closely
approximated the estimates from the original dataset. Furthermore the 95% confidence intervals
around these mean values are narrow and contain the parameter estimates generated from the
original dataset. For the linear model, performance, judged by the R-squared values averaged
across each of the ten imputation iterations, was 0.52 (min 0.51, max 0.54). The mean R-squared
value across the bootstrap replicates was 0.52 (95% CI: 0.52, 0.53), which is identical to the
model performance in the original dataset. For the logistic model based on the dichotomized
outcome variable, the AUC value averaged across the ten imputation iterations was 0.93 (min
0.91, max 0.93), indicating excellent model predictive discrimination. The mean AUC value
across the bootstrap replicates was 0.92 (95% CI: 0.92, 0.93). Figure 4-2 represents the
internally validated prediction model equations.
Table 4-5: Parameter Estimates for Models predicting a) FIM motor score and b)
Functional Independence, at 1 year follow-up from original sample and bootstrap
replicates
Panel A)
Prognostic
Variable
Parameter
Estimate
P-Value Bootstrap
Parameter Estimate
(95% CI)
Intercept 50.28 <0.01 49.73(49.0,50.5)
Age -0.33 <0.01 -0.33(-0.34,-0.32)
Admission AMS
>50
9.17 <0.01 9.11(8.61,9.61)
Admission AIS
grade
12.47 <0.01 12.54(12.34,12.76)
MRI signal -4.83 0.19 -4.65(-4.91,-4.40)
85
Panel B)
Prognostic
Variable
Parameter
Estimate
Odds Ratio P-Value Bootstrap
Parameter
Estimate (95% CI)
Intercept -2.93 - <0.01 -2.99(-3.11,-2.87)
Age -0.03 0.97 <0.01 -0.03(-0.03,-0.03)
Admission
AMS >50
1.35 3.86 <0.01 1.34(1.27,1.41)
Admission AIS
grade
1.36 3.90 <0.01 1.39(1.27,1.41)
MRI signal -0.29 0.75 0.54 -0.30(-0.34,-0.25)
Lastly, using the regression coefficients from the internally validated models, we estimated the
FIM motor score and the probability of achieving functional independence at one year follow-up
for a variety of hypothetical SCI patient scenarios (Table 4-6).
4.5 Discussion
The current analysis has led to the production of the first prediction models using acute clinical
and radiological data, obtained within the first three days after injury, to predict long term
functional outcome after traumatic spinal cord injury. Four pre-specified predictor variables,
ASIA impairment severity grade, dichotomized ASIA motor score, age and MRI intra-medullary
signal characteristics, were modeled relative to Functional Independence Measure motor score
for a large prospective cohort of SCI patients. The results of this modeling indicate that better
functional status is predicted by a sequentially less severe initial AIS grade and an AMS greater
than 50 at hospital admission. In contrast worse functional outcomes are predicted by older age
and by MRI intra-medullary signal characteristics consistent with spinal cord edema or
hemorrhage. The parameter estimates generated for these predictors were validated using a
bootstrap re-sampling procedure of observations from the original dataset. This research has
generated a simple equation that can be used to estimate long-term functional outcome by
substituting the applicable variables into the equation for each patient at acute hospital
admission. Next, by dichotomizing the primary outcome measure and generating a secondary
86
logistic model, the same four predictor variables demonstrated an excellent ability to
discriminate an individual who will achieve functional independence, from one who will not
achieve this threshold.
Figure 4-2: Predictive Model Equations
87
Table 4-6: Predicted Estimates for Hypothetical Patients
Characteristic Patient 1 Patient 2 Patient 3 Patient 4 Patient 5 Patient 6
Age 75 75 55 55 25 25
AIS A YES NO YES NO NO NO
AIS B NO NO NO NO NO YES
AIS C NO NO NO NO YES NO
AIS D NO YES NO YES NO NO
AMS >50 NO YES NO YES YES NO
MRI hemor. YES NO YES NO NO NO
MRI edema NO NO NO NO YES YES
Predicted
FIM-motor
28 85 44 91 84 62
Prob. of
Functional
Independence
0.01 0.83 0.08 0.90 0.72 0.22
As described in chapter 2, a number of studies have investigated the predictive significance of
individual clinical, radiological and demographic factors in relation to long-term functional
outcome after traumatic SCI74,130,141,149,150,152,153,163
. However, in the real world setting, clinicians
do not consider individual diagnostic elements in isolation, but rather synthesize treatment plans
and make outcome predictions by considering and weighing all pieces of information
simultaneously. Further, by employing statistical analyses which consider only one variable in
relation to outcome, it is impossible to discern, in quantitative terms, what the independent
effects of that variable are on outcome without adjusting for additional covariates known to be of
predictive importance. As a result, multivariable techniques are the preferred method for
investigating and modeling the impact of specific variables on long-term health-related
outcomes. To date, as described earlier in this thesis, a paucity of studies have assessed the
significance of combining individual parameters to improve functional outcome predictions after
SCI. Using clinical exam and electrophysiological parameters obtained from the subacute period
in 90 patients with incomplete SCI (79% traumatic, 21% non-traumatic), Zoerner et al produced
88
an algorithm to predict ambulatory capacity six months follow-up162
. In a more recent study from
Van Middendorp and colleagues, the European Multicenter SCI dataset was used to generate a
clinical prediction rule forecasting future ambulatory capacity using a combination of age, motor
scores from the quadriceps and soleus muscles and light touch sensations in the L3 and S1
dermatomes as predictor variables156
. However, it is noteworthy that the vast majority of
examinations used to construct this model took place after the acute 3 day time window, up to 15
days post SCI.
The studies described above have established the feasibility and potential benefits of using a
combination of variables to obtain more accurate estimates surrounding long-term functional
outcome after SCI. However, several factors limit the potential utility of existing models as
clinical prediction tools for use in the acute clinical setting. All of the existing studies employing
a combination of variables have relied largely on measurements obtained outside the critical first
three days after the injury: the period during which the majority of acute therapeutic decisions
are made and a time during which there is significant demand for the delivery of accurate
prognostic information. The existing literature suggests that during this acute injury window, a
variety of distracting patient and injury related factors, which are not germane during the sub-
acute period, can impede the process of obtaining an accurate neurologic exam148,209
. Previous
prediction rules were conceived without considering the presence of such distracting features and
as result they have incorporated predictors, such as electrophysiology and precise motor/sensory
examinations, which may be impractical or cumbersome for collection during the acute injury
window. Given our reliance on acute care data obtained exclusively within three days of injury
for all patients, the current predictive model can be used at the time of admission by acute care
practitioners to predict long-term functional status. The effects of inaccuracies surrounding the
acute neurological assessment of SCI patients were mitigated in the current prediction model in
two key ways. Firstly, we chose to use AIS grade and dichotomized AMS (above or below a
score of 50) as the main neurological exam predictor variables. Intuitively, these variables are
more easily obtained and categorized in the acute care period as compared to other neurological
parameters, such as specific AMS or ASS values. Secondly, as additional predictors, we included
age and MRI intramedullary signal characteristics which are objectively assessable regardless of
patients’ clinical status and are impervious to the same acute care biases that can affect
examination accuracy.
89
Previous prediction models and analyses exploring the predictors of future functional status have
predominately considered walking as the primary measure of long-term outcome151,152,156,164
.
While ambulatory capacity is of interest, in accordance with the findings of a large survey which
explored the main underpinnings of quality of life in SCI patients, it is also important to study
items such as arm and hand function as well as bowel/bladder and sexual function, in addition to
ambulation 71
. Moreover, only a small fraction of patients recover ambulatory capacity after SCI,
with a maximum of 5% of patients with complete SCI becoming ambulatory at one year,
irrespective of the initial neurological level of injury134,135
. This information reinforces the fact
that ambulation is an unrealistic endpoint for a large proportion of injured patients and
underscores the importance incorporating outcome tools that include multi-dimensional
assessments of functional outcome. In the current study we utilized the FIM motor score which,
in addition to mobility status, assesses items related to self-care, sphincter control and transfers,
areas of functioning that have great relevance to the majority of SCI patients on a daily basis.
One potential drawback of utilizing this outcome measure is its lack of interpretability. We know
that FIM motor values range from 13 (completely dependent) to 91(functionally independent)
and that lower values are associated with poorer functional status and higher values are
associated with better functional outcome. Previous studies have also demonstrated that higher
scores reflect fewer nursing care hours after patient discharge from rehabilitation231
. However,
for a given patient with a specific predicted value, it is difficult to precisely articulate exact
functional expectations within each of the four outcome domains based on the cumulative score.
We have attempted to mitigate this concern by producing a second model which predicts
functional independence: a more interpretable recovery threshold based on dichotomized FIM
motor score.
4.5.1 Study Limitations
While applying the pre-specified models to the bootstrap replicates confirms their validity within
our own dataset, establishing true external validity of this predictive model will require an
evaluation of its performance within a separate group of SCI patients. Before such an evaluation
takes place, the generalizablity of this model, as well as its suitability for use in the clinical
realm, remains unknown. Related to this, it is acknowledged that the majority of patients
included had a cervical neurological level, with a relatively low proportion of patients with
thoracic or lumbar SCI. As a result, these findings will need to be validated in datasets
90
containing a greater proportion of thoracic and lumbar SCI patients. We also recognize that a
large number of patients failed to be included in the analysis due to inadequate follow-up.
However, we have demonstrated that, with the exception of age, no significant differences
existed in the baseline characteristics between those included in the analysis and those excluded
due to inadequate follow-up. The difference in age (46.1 in patients with missing follow-up and
43.2 in patients with follow-up), although statistically significant, is not significant from a
clinical standpoint. We also recognize that treatment related factors not included in our model,
such as time to decompressive surgery or steroid administration, may have influenced patient
outcomes in this study. However this model was intended to incorporate only non-treatment
predictors related to the natural history of recovery for SCI patients. Lastly, while a standardized
MRI protocol was used for the acute evaluation of patients within this study, we acknowledge
the existence of other MR sequences, such as gradient echo, that may have provided a more
sensitive measure of intramedullary hemorrhage.
4.6 Conclusion
In this study, the first prediction models for functional outcome at one year, based on ASIA
Impairment Scale grade, dichotomized ASIA motor score, age and MRI intra-medullary signal
characteristics, with all predictor variables obtained within the first three days after injury for all
patients, have been created and internally validated. After establishing the performance of this
model within a separate cohort of patients, we look forward to its implementation as a predictive
tool in the realm of acute clinical care as well as a stratification tool in future clinical trials to
help produce balanced treatment groups. As additional future steps, we also look forward to
investigating the predictive capacity of different imaging modalities and MR imaging sequences,
as well as to producing models specifically focused on predicting outcome for important SCI
patient subgroups, such as those with incomplete injuries or central cord syndrome.
91
5 Chapter 5: The Impact of Facet Dislocation on Neurological Outcome after Cervical Spinal Cord Injury
This chapter is modified from the following:
Wilson JR, Vaccaro A, Harrop JS, Aarabi B, Shaffrey C, Dvorak M, Fisher C, Arnold P,
Massicotte EM, Lewis S, Rampersaud R, Okonkwo DO and Fehlings MG. The Impact of
Facet Dislocation on Clinical Outcomes after Cervical Spinal Cord Injury: Results of a
Multicenter North American Prospective Cohort Study. Spine (Phila Pa 1976)
2013;38:97-103 .232
92
5.1 Abstract
Reports of dramatic neurological improvement in patients with facet dislocation (FD) and
cervical spinal cord injury (SCI), treated with rapid reduction, have led to the hypothesis that this
represents a subgroup of SCI patients with significant potential for recovery. However, in the
absence of a large systematic analysis, evaluating outcomes in patients with FD and SCI as
compared to SCI patients without FD, this hypothesis remains untested. In the context of the
Surgical Timing in Acute Spinal Cord Injury Study (STASCIS), a multicenter prospective cohort
study, to define differences in baseline characteristics and neurological outcomes between
cervical spinal cord injury (SCI) patients with and without facet dislocation (FD), was
undertaken. Patients were classified into FD and non-FD groups depending on the results of
imaging investigations performed at admission and then followed prospectively over time. The
primary outcome measure was the change in ASIA motor score (AMS) at 1 year follow-up.
Secondary outcome measures included ASIA Impairment Scale (AIS) grade conversion as well
as length of acute hospital stay. Of 421 patients enrolled, 135(32.1%) had a FD and 286(67.9%)
had no FD. Patients in the FD group had significantly worse presenting AIS grade as compared
to those in the non-FD group (p<0.01). Forty of the 48 FD patients who were placed in traction,
successfully achieved closed reduction. There were no differences in baseline gender or age
between the 2 groups (P>0.05). Patients in the FD group more frequently experienced higher
energy injury mechanisms as compared to those in the non-FD group(p<0.01). The mean length
of acute hospital stay was 41.2 days amongst patients with FD and 30.0 amongst patients without
FD (p=0.04). At 1-year follow-up, FD patients experienced a mean AMS improvement of 18.0
points, whereas non-FD patients experienced a significantly larger improvement of 27.9
points(p<0.01). After adjusting for baseline neurological status using linear regression, patients
with FD continued to demonstrate a smaller degree of AMS recovery as compared to the non-FD
patients(p=0.04). Overall, as compared to those without FD, cervical SCI patients with FD
tended to present with a more severe degree of initial injury and displayed less potential for
motor recovery at one year follow-up.
93
5.2 Introduction
Cervical facet dislocation (FD) is an important and prevalent cause of cervical spinal cord injury
(SCI)233-236
. FD typically occurs as a result of a hyper-flexion injury to the neck, causing one, or
both, of the inferior facets of the upper dislocating vertebrae to slide over, and anteriorly to, the
superior facet(s) of the lower dislocating vertebrae237,238
. The final result is a narrowing of the
spinal canal that can lead to spinal cord compression and neurologic injury. While FD is
relatively rare amongst patients with cervical spine trauma in general, accounting for about 5-
10% of all cases seen, the consequences of these injuries are often devastating239
. Approximately
37% of unilateral FD patients, and upwards of 90% of patients with bilateral FD, are diagnosed
with SCI at presentation to hospital240,241
.
In spite of the high incidence of SCI seen with cervical FD, there are reports in the literature of
dramatic neurological recovery with rapid realignment of the spinal column and decompression
of the spinal cord242-247
. Contradicting these findings is the general clinical observation that FDs
are typically associated with high energy biomechanical forces causing significant tissue
disruption and a severe degree of primary neurologic injury248
. In the setting of such a severe
primary SCI, potential for significant neurologic recovery, even with rapid reduction, would
ostensibly be limited. However, in the absence of a large longitudinal analysis, the profile of
recovery for patients with FD and cervical SCI as compared to cervical SCI patients without FD,
remains largely unknown.
In the literature review presented in chapter 2 of this thesis, a paucity of CT or X-ray based
imaging findings were found to be predictive of long-term outcome. However FD, which is most
commonly diagnosed by these imaging modalities, has yet to be investigated for its potential
value in predicting outcome post cervical SCI.
To obtain an improved understanding of the natural history of patients with FD and SCI we have
completed a prospective cohort study comparing baseline characteristics and long-term
neurological outcomes between SCI patients with and without FD.
94
5.3 Methods
5.3.1 Patient Population
Within the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS) a multicenter
prospective cohort study comparing baseline characteristics and long-term outcomes between
patients with and without FD was undertaken. During the 7 year study period between 2002 and
2009 patients were enrolled at 7 institutions throughout North America: 1) University of
Toronto, Toronto, Ontario, Canada; 2) Thomas Jefferson University, Philadelphia, PN, USA; 3)
University of Virginia, Charlottesville, VA, USA; 4) University of Maryland, Baltimore, MD,
USA; 5) University of British Columbia, Vancouver, British Columbia, Canada; 6) University of
Kansas, Kansas City, KS, USA; 7) University of Pittsburgh, Pittsburgh, PA, USA. During the
study period, any patient presenting to one of the participating centers, with traumatic cervical
SCI and radiologic evidence of spinal cord compression, was enrolled. Patients younger than 16
years of age and those with penetrating injuries were excluded. A standardized neurological
examination, performed in accordance with the International Standards for the Neurological
Classification of Spinal Cord Injury, was completed within 72 hours of injury224
. Also at
admission, demographic and injury related variables such as age, gender, mechanism of injury,
injury severity score (ISS) and Glasgow Coma Scale Score (GCS), were also collected. Motor
vehicle accidents were considered high energy injury etiologies, whereas falls from standing,
diving accidents or injuries resulting from blunt violence, were considered low energy etiologies.
At presentation, all patients underwent a cervical X-Ray and/or a CT scan. If unilateral or
bilateral FD was diagnosed, patients underwent either closed reduction, through application of
skeletal traction, or open reduction. If a trial of traction failed, with failure defined as a new or
worsening neurological deficit, medical instability, radiologic evidence of over distraction or
patient/equipment unable to tolerate additional weight, the treating team proceeded with open
reduction on an emergent basis. After a successful trial of closed reduction, a cervical magnetic
resonance imaging (MRI) study was obtained in the majority of cases, to document the degree of
decompression of the spinal cord achieved. If the post reduction MRI demonstrated incomplete
decompression of the spinal cord, then the patient underwent open reduction. Cervical SCI
patients without evidence of FD on initial X-Ray or CT underwent a cervical MRI, to document
the degree of spinal cord compression, followed by surgical decompression. Radiological
95
evidence of spinal cord compression was defined by the method previously described by the
Fehlings group208
. Decision of surgical timing was dependent on patients’ pre-hospital transport
time, the time to obtain diagnostic imaging, the time required for medical stabilization and the
discretion of the attending spinal surgeon.
All patients received standardized acute medical therapy which included permissive or induced
hypertensive therapy (mean BP > 85 mm Hg)95
. Acute administration of intravenous
methylprednisolone sodium succinate was permitted for use at the discretion of the treating team,
according to the recommendations of the second National Acute Spinal Cord Injury Study
(NASCIS-2) and based on established guidelines105,108
. All technical decisions surrounding
surgery were made by the attending spine surgeon in each case. Finally, all patients underwent a
customized course of inpatient spinal cord rehabilitation after acute care discharge.
5.3.2 Outcome Variables
Based on the presence or absence of FD on admission imaging, patients were divided into a FD
cohort and a non-FD cohort and followed prospectively over time. The primary outcome of
interest was the change in ASIA motor score (AMS) at 1 year follow-up, obtained by subtracting
the baseline AMS, obtained at acute hospital admission, from the AMS at 1 year follow-up, for
each patient. Change in ASIA Impairment Scale (AIS) grade was considered as a secondary
neurologic outcome measure. Finally, acute hospital length of stay was compared between the
FD and non-FD groups.
As a secondary analysis baseline characteristics, as well as clinical outcomes, were compared
between patients with bilateral FD and those with unilateral FD.
5.3.3 Statistical Analysis
All analyses were carried out using SAS version 9.2 (Cary, NC, USA). Descriptive statistics
were used to characterize patient demographics and injury characteristics at hospital admission.
Continuous outcome measures were compared between the cohorts using the Student’s t-test and
categorical outcomes were compared using the chi-squared or Fisher’s exact test. An adjusted
analysis was also completed, to account for the effects of confounding variables, using linear
regression. For this analysis the outcome variable was defined as the change in AMS from pre-op
to 1 year follow-up, while the predictor variable of interest was defined as the presence or
96
absence of a FD. Within this regression equation we also included admission AIS grade as a
predictor variable, thereby adjusting for the impact of initial injury severity, the variable known
to have the strongest impact on long-term motor recovery as discussed in Chapter 2.
Table 5-1: Baseline Patient Characteristics
Panel A: Facet Dislocation Cohort vs. Non Facet Dislocation Cohort
Characteristic Overall
(n=421)
FD (n=135) Non-FD
(n=286)
p-value
Presenting AIS
grade:
AIS grade A
AIS grade B
AIS grade C
AIS grade D
149(35.6%)
70(16.6%)
86(20.4%)
116(27.6)
69(51.1%)
18(13.3%)
18(13.3%)
30(22.2%)
80(28.0%)
52(18.2%)
68(23.8%)
86(30.1%)
P<0.01
Mean AMS 37.5(±31.2) 38.5(±34.8) 35.5(±31.3) P=0.37
Male gender 315(74.8%) 97(71.9%) 218(76.2%) p=0.53
Mean Age 48.2(±17.0) 46.4(±15.9) 49.0(±17.5) P=0.15
Mean Injury
Severity Score
12.4(±10.5) 12.8(±9.4) 12.2(±11.0) P=0.70
High energy 212(51.1%) 81(60.5%) 131(46.6%) P<0.01
Mean GCS 13.8(±3.4) 13.7(±3.5) 13.8(±3.3) P=0.70
Panel B: Unilateral Facet Dislocation patients vs. Bilateral Facet Dislocation patients
Characteristic Uni-FD=42 Bilat-FD=93 p-value
Presenting AIS
grade:
AIS grade A
AIS grade B
AIS grade C
AIS grade D
14(33.3%)
6(14.3%)
3(7.1%)
19(45.2%)
55(59.1%)
12(12.9%)
15(16.1%)
11(11.8%)
P<0.01
Mean AMS 51.6(±37.4) 28.1(±24.9) P<0.01
Mean age 45.7(±14.9) 46.8(±16.5) P=0.70
Male gender 28(66.7%) 69(74.2%) P=0.37
High Energy 24(58.5%) 57(61.3%) P=0.76
Mean Injury
Severity Score
11.7(±8.7%) 13.4(±9.8) P=0.51
Mean GCS 13.9(±3.0) 13.6(±3.6) P=0.64
97
5.4 Results
5.4.1 Patient Population
A total of 421 cervical SCI patients were enrolled during the study period, of which 135(32.1%)
had a FD and 286(67.9%) had no FD. Baseline characteristics of the FD and non-FD groups are
presented in Table 5-1, panel A. Patients in the FD group had a significantly worse presenting
AIS grade and higher energy injury mechanisms as compared to those in the non-FD
group(p<0.01). Of the 135 patients with FD, 42 patients (31.1%) had a unilateral FD and 93
patients (68.9%) had bilateral FD (Table 5-1, panel B). The neurological status on admission was
significantly different between the unilateral and bilateral FD groups, with AIS grade A injuries
most common amongst bilateral FD patients and AIS grade D injuries more common amongst
the unilateral FD patients. In addition, the mean AMS at admission was significantly higher
amongst patients with unilateral FD (50.4±37.1) as compared to patients with bilateral FD
(28.1±24.9). The vertebral level of injury for both unilateral and bilateral FD patients is
presented in Figure 5-1.
Figure 5-1: Number of Facet Dislocations per disc space level
98
5.4.2 Treatment
Of the 135 patients with FD, 48(35.6%) initially underwent skeletal traction and 87(64.4%)
initially underwent open reduction (Figure 5-2). Forty (83.3%) of the patients, who initially
received traction, were successfully reduced, while the remaining 8 patients (16.7%) failed, and
underwent surgical spinal cord decompression. There were no neurological complications during
the application of traction. The mean time to decompression in the FD group was 25.1(±34.8)
hours and 41.3(±44.6) hours in the non FD-group (p<0.01). Amongst FD patients, the mean time
to spinal cord decompression in the group initially treated with traction was 11.5(±10.0) hours
and 32.2(±40.7) hours amongst those who initially underwent open reduction (p<0.01). There
was no difference in the proportion of patients that received steroids between the FD and non-FD
groups (p=0.57).
Figure 5-2: Treatment Flow for patients with Facet Dislocation
99
5.4.3 Outcome
At 1 year post injury, follow-up neurological outcome data was available for 211 patients: 67 FD
patients (21 unilateral and 46 bilateral) and 144 non-FD patients. There were no differences in
the baseline characteristics of those with 1 year follow-up as compared to those without 1-year
follow-up (p>0.05 for all characteristics). Table 5-2 provides a comparison in outcomes between
the FD and non-FD group. Patients in the non-FD group had significantly larger improvement in
AMS at follow-up (p<0.01). There was a significant difference in the mean length of acute
hospital stay, with FD patients staying on average 41.2 days as compared to 30.0 days in the non-
FD group (p=0.04). Table 5-3 compares outcomes between unilateral and bilateral FD patients.
While there was a trend towards a shorter length of stay amongst unilateral FD patients, this
difference failed to achieve statistical significance.
Table 5-2: Clinical Outcomes at 1 year Follow-up, FD vs. Non-FD Cohort
Outcome Overall
(n=211)
FD
(n=67)
Non-FD
(n=144)
P-value
Change in AMS at 1
year F/U
24.6(±25.4) 18.0(±21.4) 27.9(±25.1) P<0.01
AIS grade
improvement
≥1grades
≥2grades
122(57.8%)
35(16.6%)
39(58.2%)
11(16.4%)
83(57.4)
24(16.7%)
P=0.93
P=0.96
Acute Length of Stay* 33.6(±53.4) 41.2(±75.2) 30.0(±38.5) P=0.04
*Based on the duration of stay of 421 patients during acute hospital admission
100
Table 5-3: Clinical Outcomes at Follow-up, Unilateral FD vs. Bilateral FD patients
Outcome FD-uni
(N=21)
FD-bi
(N=46)
p-value
Change in AMS at 1-
year F/U
18.8(±20.1) 19.2(±22.5) P=0.95
AIS grade
improvement
≥1grades
≥2grades
13(61.9%)
5(23.8%)
26(56.5%)
6(13.0%)
P=0.68
P=0.27
Acute Length of Stay* 23.9(±41.1) 49.7(±87.4) P=0.07
*Based on the duration of stay of 135 patients during acute hospital admission
For the multivariable analysis, with change in AMS at one year follow-up as the outcome
variable, FD (the predictor of interest) and admission AIS grade appeared as predictor variables
in the final regression equation. Parameter estimates and their corresponding p-values are found
in table 5-4. After adjusting for admission AIS grade, the presence of FD remained a significant
negative predictor of motor recovery at 1-year follow-up (p=0.04).
Table 5-4: Results of Multivariable Analysis (Outcome variable: Change in ASIA motor score
at 1-year follow-up)
Predictor Parameter Estimate p-value
FD vs. Non-FD -6.6 P=0.04
Admission AIS grade:
AIS grade A
AIS grade B
AIS grade C
AIS grade D (ref)
-0.5
23.3
30.2
-
P<0.01
5.5 Discussion
In summary, this represents the largest analysis evaluating the impact of facet dislocation on
outcomes after cervical SCI. In this patient population, we have shown that FD is more
frequently associated with severe neurological deficit at patient presentation and with high
energy injury mechanisms, as compared to non-FD related injuries. At follow-up, based on the
results of the unadjusted analyses, FD patients experienced a smaller amount of motor recovery,
101
as measured by the change in AMS from admission to 1 year, and had a significantly longer
length of acute hospital stay. In the adjusted analysis, after controlling for admission AIS grade,
the presence of FD was a significant predictor of inferior motor recovery 1 year after injury, with
a parameter estimate of -6.6. Stated plainly, after adjusting for admission neurological status,
patients with FD and SCI will experience approximately 7 fewer points in motor recovery at 1
year follow-up, as compared to those without FD and SCI.
In comparing patients with unilateral versus bilateral FD, patients with bilateral FD had a more
severe degree of neurological deficit at presentation. Although there were no statistically
significant differences in outcomes between these groups, the bilateral FD group had a longer
duration of acute hospital stay, with a difference that approached statistical significance.
In evaluating the available literature, there is a paucity of high quality studies which examine the
clinical outcomes of patients with cervical FD and SCI. Several studies have reported
considerable improvement in neurological status after rapid decompression of the spinal cord
through closed reduction243-247
. Although most of these studies are case reports or small case
series, this has led to the assertion that if treated promptly, patients with FD represent a subgroup
with considerable potential for recovery. However, the findings of the present study are contrary
to this observation. In spite of the fact that spinal cord decompression was achieved more rapidly
in the FD patients, whether through open or closed reduction, they exhibited a diminished
potential for motor recovery at long term follow-up as compared to non-FD patients. We
speculate that this finding is related to the observation that FD injuries are associated with higher
energy injury mechanisms causing a greater degree of primary neurologic injury, ultimately
resulting in a diminished potential for motor recovery at follow-up. Presently, it remains unclear
whether or not recovery for an individual patient with FD and SCI, may be optimized by prompt
reduction and decompression of the spinal cord. As a result, the findings of this study should not
deter clinicians from prioritizing the prompt decompression of patients with FD and SCI. We are
currently conducting a separate analysis to specifically determine how time to reduction in
patients with FD and SCI influences outcomes. Until the results of this study are available, we
continue to advocate for early intervention based on compelling biological rationale coupled with
the results of the recent STASCIS study119
.
102
Due to the increased use of sedation, intubation and ventilation in severely injured trauma
patients, a large percentage of those with SCI may not be examinable in the acute clinical setting.
In light of this observation, the identification of objective radiologic findings that help to classify
injury severity and prognosticate long-term outcome, is an important goal188
. Without any
knowledge of an individual trauma patient’s clinical history or physical exam, the current study
demonstrates that the presence of FD on admission X-ray or CT scan should inform the clinician
that the injury is likely to have resulted from a high energy injury mechanism and raise suspicion
of a potentially severe underlying neurological injury. This is significant in light of the pre-
existing absence of CT or X-ray based variables known to be predictive of long-term outcome, as
discussed in the literature review presented in Chapter 2. In terms of predicting long-term
outcome, understanding the negative impact of FD on motor recovery after SCI, will aid
clinicians in conveying accurate prognostic information to patients and their families. Moreover,
this prognostic information can be used to help stratify cervical SCI patients within clinical trials
evaluating the impact of novel therapeutic agents on neurological recovery.
5.5.1 Study Strengths and Limitations
As compared to a single center study, the multicenter nature of patient data accumulated in this
study increases the external validity of the results obtained. As with most observational
comparative studies, there are differences in patient characteristics between the cohorts that
could potentially impact observed outcomes. These imbalances were dealt with in the best way
possible, by performing a multivariate regression analysis, adjusting for relevant confounders. In
addition, a large number of patients were missing to follow-up at 1 year post injury. However the
challenges of obtaining long term follow-up in the trauma population, particularly among those
with severe mobility issues related to spinal cord injury, as compared to patients with electively
treated neurological diseases, are well recognized72,156
. In addition, we demonstrated that there
were no baseline differences in the characteristics of those with 1 year follow-up as compared to
those without available follow-up.
5.6 Conclusion
This analysis demonstrated that, in comparison to those without FD, patients with SCI in
association with cervical FD presented with a greater severity of neurological deficit and
experienced a smaller potential for motor recovery at long-term follow-up, even after adjusting
103
for baseline discrepancies in neurological status. From the standpoint of resource planning, this
analysis also demonstrated that patients with FD and SCI have longer periods of acute hospital
admission. These results have important implications for counseling patients and families
regarding prognosis and also for the stratification of patients within future SCI clinical trials. In
spite of the findings presented in this study, early spinal cord decompression for patients with FD
and SCI is recommended, based on the best available preclinical and clinical evidence115
.
104
6 Chapter 6 Predicting Inpatient Complications after Traumatic Cervical Spinal Cord Injury through the use of Acute Clinical Variables
This chapter is modified from the following:
Wilson JR, Arnold PM, Singh A, Kalsi-Ryan S and Fehlings MG. Clinical prediction
model for acute inpatient complications after traumatic cervical spinal cord injury: a sub-
analysis from the Surgical Timing in Acute Spinal Cord Injury Study. J Neurosurg Spine
2012; 17(Sup):46-51.249
105
6.1 Abstract
While the majority of existing reports focus on complications sustained during the sub-acute and
chronic stages after traumatic spinal cord injury (SCI), the objective of the current study was to
characterize and quantify acute in-patient complications as well as to identify the combination of
clinical variables predicting their occurrence. Based on these analyses, a prediction model using
clinical variables collected at hospital admission to predict complication development in cervical
SCI patients during the acute hospital stay was created. All analyses were based on data from the
Surgical Timing in Acute Spinal Cord Injury Study (STASCIS) database. Complete complication
information was available for 411 patients at acute care discharge. One hundred and sixty
patients (38.9%) experienced 240 complications. The mean age amongst patients who
experienced at least one complication was 45.9 as compared to 43.5 in those who did not
experience a complication (p=0.18). In the univariable analysis, patients who experienced
complications were found to have a lower frequency of steroid administration at admission
(p=0.01), a greater severity of neurological injury as measured by ASIA impairment scale (AIS)
grade at presentation (p<0.01) and a higher frequency of significant comorbid illness (p=0.04).
In the creation of the multivariable logistic regression model, severe initial AIS grade (p<0.01),
high energy injury mechanisms (p=0.07), older age (p=0.047), absence of steroid administration
(p=0.02) and presence of comorbid illness (p=0.02) were associated with a higher odds of
complication development during the period of acute hospitalization. The AUC value for the full
model was 0.75, indicating acceptable predictive discrimination. These results will help
clinicians to identify those cervical SCI patients at greatest risk for complication development,
and based on this recognition, allow for the institution of aggressive complication preventive
measures.
106
6.2 Introduction
Although the incidence of traumatic spinal cord injury (SCI) is low (approximately 15-58 per
million per year in the general population), for the individual affected, this is a life changing
event which often results in significant disability, loss of independence and an increased risk of
mortality4,27,221,250
. In the absence of a proven regenerative or reparative therapy, the current
treatment approach is centered on optimizing long-term clinical outcomes by mitigating
secondary causes of injury and avoiding acute medical complications112,223
.
Given the extent of physiological disruption that can occur after SCI, patients are susceptible to a
myriad of secondary complications in the acute period following injury30,251
. Complications have
been reported to occur after SCI with incidence rates ranging from 20-77%
worldwide7,9,10,14,15,31,252,253
. However, comparisons of complication rates across the literature are
inherently challenging given the heterogeneity in complication categorization and specific time
period considered after injury. With the majority of existing reports having focused on sub-acute
and chronic complications sustained during the rehabilitation phase post SCI, a relative paucity
of studies have examined complications occurring during the period of acute hospitalization.
Regardless of the time point at which they occur, complications have been shown to lead to an
increased length of hospital stay, higher rates of mortality and diminished functional recovery at
long-term follow-up254,255
. As a result, a high degree of vigilance is required on the part of the
treating medical team to recognize their occurrence and promptly institute appropriate treatment.
However, such a process depends on medical practitioners having an understanding of the patient
and injury related factors that are the most predictive of complication development, with high
risk patients subsequently becoming the targets of aggressive complication prevention strategies.
As discussed in Chapter 2, while a variety of studies have identified individual clinical variables
of importance in predicting sub-acute/chronic complications post SCI, few studies have
evaluated the importance of such variables in predicting acute complications.
Given this background, our goal was to characterize and quantify complications occurring during
the acute hospital admission after cervical SCI, using a large prospective cervical SCI dataset.
Second, we planned to identify the combination of clinical variables that best predict acute
complication occurrence by developing a complication prediction model. Finally, based on this
107
analysis, we aimed to estimate the risk of acute complication development for a series of
hypothetical cervical SCI patients, based on their specific presenting characteristics.
6.3 Methods
6.3.1 Study Population
A multi-center, North American prospective cohort study of early (<24hrs) versus late (≥24hrs)
spinal cord decompression following cervical SCI (STASCIS) was performed at 7 North
American centers256
. During the 7 year study period, between 2002 and 2009, any SCI patient
greater than 16 years old, presenting to one of these institutions with an ASIA Impairment Scale
(AIS) grade of A-D, a cervical neurological level of injury and radiological evidence of spinal
cord compression, was considered for enrolment. Spinal cord compression was determined
according to the radiologic methods previously described by the Fehlings group208,257
. For the
current analysis, only patients with a documented neurological examination performed within 72
hours of injury and follow-up information available at acute care discharge were included.
All patients included in this analysis received appropriate medical support according to the
recommendations of the 2002 American Association of Neurological Surgeons cervical SCI
guidelines95,96
. In each case, decisions surrounding surgery and methylprednisolone
administration were made at the discretion of the spinal surgeon and the treating team involved.
When methylprednisolone was given, the 24-hour NASCIS-II protocol was administered
invariably105
.
6.3.2 Candidate Complication Predictor Variables
At patient presentation, neurological examination was completed according to the International
Standards for the Neurological Classification of SCI, with ASIA impairment scale (AIS) grade
defined as the primary neurological predictor variable of interest for the purpose of this study224
.
In addition to neurological exam parameters, a variety of clinical variables were recorded at
admission including age, gender, energy injury mechanism, injury severity score (ISS) and
Charlson Co-morbidity index. For purposes of this analysis, all patients with an admission
Charlson Co-morbidity score ≥ 1 were considered to have co-morbid illness whereas those with a
score of 0 were considered to have no co-morbid illness. Motor vehicle accidents, including
motorcycle and all-terrain vehicle accidents, were considered to be high energy injury
108
mechanisms, whereas falls from standing, sporting related accidents and blunt violence were
considered low energy injury mechanisms.
6.3.3 Complication Categorization and primary outcome
During the period of acute hospitalization, complication information was collected prospectively
for each patient. For purposes of classification, complications were prospectively divided into 5
predefined categories: 1) cardio-pulmonary complications (e.g. pneumonia, acute respiratory
distress syndrome (ARDS), acute myocardial infarction, dysrhythmia); 2) surgical complications
(e.g. wound infection, dehiscence, construct failure, post-operative neurological deficit); 3)
thrombotic complications (e.g. deep venous thrombosis or pulmonary embolism); 4) infectious
complications (urosepsis, gastrointestinal infection, meningitis, other systemic infection); and 5)
decubitus ulcer development. We have not differentiated between major and minor
complications as all of the predefined categories were intended to capture only major
complication events with potential to impact clinical outcomes.
For this analysis, the primary complication outcome was binary, defined by the absence of a
complication event occurrence versus the occurrence of one or more complication events during
the acute inpatient hospital period.
6.3.4 Statistical Analysis
All analyses were completed using SAS version 9.2 (SAS Institute Inc. Cary, NC). An overall
complication incidence rate and category specific incidence rates were calculated. Univariable
statistical analyses were performed to investigate the relationship between complication
occurrence and individual demographic, injury and treatment predictor variables using the chi-
square test for categorical variables and the Student’s t-test for continuous variables.
Multivariable logistic regression was subsequently performed to identify the combination of
variables that predict the dichotomized primary complication outcome. Individual predictor
variables were considered for inclusion in the multivariate model if there was author consensus
about the predictive importance of the variable. Subsequently, these predictors were sequentially
eliminated in a backwards fashion if their corresponding p-value was greater than 0.10. The
predictive capacity of the final model in predicting complication occurrence was tested through
the construction of a receiver operator characteristic (ROC) curve. The area under the ROC
109
curve (AUC) is a summary measure of the discriminative ability of the model, with values
greater than 0.70 indicative of acceptable predictive discrimination.
Finally, the regression coefficients obtained from the full logistic model were used to estimate
the probability of acute complication development for hypothetical cervical SCI patients,
depending on the unique combination of clinical variables that are routinely recorded at hospital
admission.
6.4 Results
Four hundred and eleven patients had acute neurological exam data and prospective complication
information available from the period of acute hospitalization within the STASCIS dataset. The
mean time from admission to acute care discharge was 34.3(±54.6) days, with a significantly
longer mean length of stay noted for patients with a complication (51.5 days) as compared to
those who did not experience a complication (23.5 days) (p<0.01). One hundred and sixty
patients (38.9%) experienced 240 complications, with 59 of these patients experiencing at least 2
complications during the acute hospital admission. Of all 240 complications seen, 94 were
cardio-pulmonary (39.2%), 87 were infectious (36.3%), 28 were surgical (11.3%), 21 were
thrombotic (8.8%) and 11 involved decubitus ulcer development (4.6%) (Figure 6-1). There were
a total of 16 deaths within the study group, with 12 involving patients who experienced a
complication (7.5% mortality rate) and 4 involving patients who did not experience a
complication (1.6% mortality rate) (p<0.01). Within the infectious category, there were 14 post-
operative wound infections in total, 5 amongst patients receiving steroids (2.1%) and 9 amongst
those who did not receive steroids (5.6%) (p=0.07).
110
Figure 6-1: Number of Events per Complication Category
The mean age amongst patients who experienced at least one complication was 45.9 (±18.0) as
compared to 43.5(±16.4) amongst those who did not experience a complication (p=0.18). In the
univariable analysis (Table 6-1), patients who experienced at least 1 complication were found to
have a lower frequency of steroid administration at admission (p=0.02), and had a greater
severity of neurological injury as measured by AIS grade at presentation (p<0.01). In addition,
amongst those with a complication, 36 patients (22.5%) had a diagnosed comorbid illness at
hospital presentation as compared to 37 patients (14.7%) amongst those who did not experience
a complication (p=0.04).
111
Table 6-1: Results of univariable analysis comparing patients who experienced at least one
complication to those who experienced no complications
Predictor Variable Patients with
no
complications
N=251
Patients with at
least 1
complication
N=160
p-value
Mean Age 43.5(±16.4) 45.9(±18.0) p=0.18
Admission AIS grade:
AIS grade A
AIS grade B
AIS grade C
AIS grade D (ref)
57(22.7%)
38(15.1%)
59(23.5%)
97(38.7%)
87(54.4%)
28(17.5%)
27(16.9%)
18(11.3%)
p<0.01
Steroid
administration
155(63.8%) 78(52.0%) p=0.01
High energy injury
mechanism
119(47.6%) 90(57.0%) p=0.07
Male gender 186(74.1%) 122(76.3%) p=0.62
Mean Injury Severity
Score
12.1(±10.8) 13.2(±10.1) p=0.52
Co-Morbid Illness
No
Yes
214(85.3%)
37(14.7%)
124(77.5%)
36(22.5%)
p=0.04
Completion of the multivariable logistic regression analysis revealed 5 independent predictors of
complication development with a p-value<0.10: 1) Age (p=0.047); 2) admission AIS grade
(p<0.01); 3) energy level of injury mechanism (p=0.07); 4) steroid administration (p=0.02); and
5) presence of comorbid illness at presentation (p=0.02). Of note, ISS, felt to be a clinically
important predictor of complication development, was included in the model initially, but was
subsequently eliminated due to an adjusted p-value =0.60, greater than the 0.10 cutoff. Results
of the multivariable analysis with odds ratios and 95% confidence intervals are available in Table
6-2. The AUC value for the five predictor model was 0.75, indicating acceptable predictive
discrimination (Figure 6-2).
112
Table 6-2: Results from the multivariable logistic regression model predicting complication
development.
Predictor Variable Odds Ratio 95% Confidence
Interval
p-value
Age 1.02 1.00,1.03 p=0.047
Admission AIS grade:
AIS grade A
AIS grade B
AIS grade C
AIS grade D (ref)
9.44
4.76
2.00
1.00
4.90,18.16
2.22,10.18
0.95,4.19
-
p<0.01
Steroid administration 0.56 0.35,0.90 p=0.02
Co-morbid illness
No
Yes (ref)
0.46
1.00
0.24,0.91
-
p=0.02
High vs. Low energy
injury mechanism
1.57 0.96, 2.56 P=0.07
Finally, using the regression coefficients from the 5 predictor logistic model, we estimated the
probability of acute complication occurrence for a variety of cervical SCI patient scenarios, each
with different clinical characteristics at hospital admission (Table 6-3).
113
Table 6-3: Predicted Probability estimates for hypothetical SCI patients
Predictor Patient#1 Patient#2 Patient#3 Patient#4 Patient#5 Patient#6
Age 20 20 50 50 75 75
AIS A YES NO YES NO YES NO
AIS B NO NO NO YES NO NO
AIS C NO YES NO NO NO YES
High Energy YES NO YES NO YES NO
Steroid
Administration
YES NO YES YES NO NO
Co-morbid
illness
NO NO YES YES YES YES
Predicted
Probability of
Complication:
48% 17% 69% 47% 81% 42%
6.5 Discussion
In summary, this represents the largest analysis investigating acute in-hospital complications
after cervical SCI performed to date. In addition, this represents the first prediction model using
clinical data collected at hospital admission to predict complication development during the
acute in-hospital period. The results of this modeling indicated that a more severe initial AIS
grade, high energy injury mechanisms, older age, absence of steroid administration and presence
of comorbid illness to be associated with a greater odds of complication development during the
period of acute hospitalization. The combination of these 5 predictors demonstrated acceptable
predictive discrimination in predicting complication occurrence. Finally, we have demonstrated
that it is possible to practically apply the prediction model in the clinical realm, to estimate the
probability that an individual patient will experience an acute complication, based on their
presenting clinical characteristics.
114
Previous studies in the context of traumatic SCI have attempted to identify important patient risk
factors that are associated with the development of secondary complications23,253,254,258-261
. The
results of these studies indicate that increased age, increased body mass index, cervical lesions
and motor complete lesions are associated with an increased frequency of complications253,254,261
.
However, the vast majority of these studies have focused on data derived from the sub-acute or
chronic setting, with relatively few investigating acute in-patient complications. This is relevant
since the profile of complications post SCI is known to evolve as the time from injury
increases27,31
. As an example, clinical experience dictates that decubitus ulcer development and
psychiatric affective disorders are common in the chronic phases, but comparatively less
frequent during the acute phases post SCI262
. Hence, results from analyses based on data
obtained during the sub-acute and chronic stages are not necessarily valid in the acute stages.
Given that the acute hospital admission is the time during which patients are the most vulnerable
to medical instability and life threatening complications, recognizing and preventing
complication occurrence during this period is likely to have a substantial impact on improving
long-term clinical outcomes.
The associations found between the severity of neurological deficit, as well as injury energy
level, and complication development, are unsurprising given that these variables are surrogates
for the extent of neurological tissue destruction and hence the extent of motor weakness and
physiological derangement. Similarly, findings that increasing age was a positive predictor of
complication occurrence, and the absence of comorbid illness was a negative predictor of
complications, are expected in light of the decreased physiologic reserve seen amongst elderly
individuals and those with significant pre-existent disease.
The finding that steroid administration was associated with a reduced risk of complication
development is likely related to several observations. First, when administered, only the 24-hour
low dosage methylprednisolone regimen given in the NASCIS II study was utilized105
. This
dosing regimen has not been associated with a significantly greater incidence of complications at
any time point after injury. Second, this study considered only patients with cervical SCI, which
is in contrast to the NASCIS studies which considered patients with thoracic/lumbar injuries in
addition to cervical injuries. In NASCIS II the rate of wound infection in the steroid group was
7.1% in comparison to 3.6% in the placebo group, however the rate of wound infection in the
current study in the steroid group was 2.1%. This is explained by the fact that while surgery for
115
thoracic and lumbar SCI often involves relatively large posterior incisions, cervical injuries are
often treated with anterior approaches using small incisions and minimal soft tissue disruption.
As a result, cervical SCI patients would intuitively be at a lower risk for wound related infection
events irrespective of steroid use. Third, because we considered a broad range of complications,
and not just infection related events, it is possible that steroid administration mitigated
complication development in certain situations. For instance steroid administration has
previously shown beneficial, in select studies, in the setting of acute myocardial infarction, acute
respiratory distress syndrome (ARDS) and hypotensive shock, all of which were complications
tracked in this study263-265
. Finally, because this study was not intended as a primary efficacy
analysis to compare outcomes amongst steroid and non-steroid treated patients, it is possible that
the steroid effect reflects residual confounding unaccounted for by our modeling procedure.
6.5.1 Study Limitations
Although the complication prediction model was constructed using a multicenter cervical SCI
database containing over 400 patients, the generalizability of this model, as well as its suitability
for use in the clinical realm, remain unproven. Hence, an independent evaluation of model
performance in a separate group of cervical SCI patients will be necessary to establish its
external validity. As well, we recognize that our method of complication categorization may not
capture all complication events that occurred during the acute hospital period. However, this
categorization scheme was developed to preferentially capture the major complication events
that would have the potential to impact long-term clinical outcomes. Finally, the current analysis
provides little granularity surrounding the predictors of specific complications such as
pneumonia. This analysis was primarily intended to look at relevant complications overall and
future analyses will focus on individual complications in the acute SCI period.
6.6 Conclusion
Complications occur with a high frequency after SCI with potentially significant long-term
sequelae. This analysis has identified a combination of factors that are predictive of complication
development in the acute setting post cervical injury. After external validation, these results will
likely help clinicians to identify those patients at greatest risk for complications, with these
116
patients becoming the targets for the administration of aggressive complication prevention
strategies.
117
7 Chapter 7: Defining Age Related Differences in Functional Outcome after Traumatic Spinal Cord Injury
This chapter is modified from the following:
Wilson JR, Davis AM, Kulkarni AV, Kiss A, Frankowski RF, Grossman RG, Fehlings
MG. Defining Age related Differences in Functional Outcome after Traumatic Spinal
Cord Injury. Currently under review at: The Spine Journal.
118
7.1 Abstract
The existing evidence suggests that while older spinal cord injury (SCI) patients experience a
similar degree of neurological recovery as younger patients, older patients experience diminished
functional outcomes at follow-up. However, all studies have assumed that the impact of age on
functional outcome is the same across the spectrum of injury severity. To test this assumption,
we evaluated age as a potential effect modifier governing the relationship between acute
neurological status and long-term functional outcome. Using a combination of the NACTN and
STASCIS datasets, age was primarily dichotomized at a threshold of 65 years. Multivariable
linear regression was used to investigate the modifying effects of age on the relationship between
acute ASIA Impairment Scale (AIS) grade and follow-up FIM motor score. An interaction plot
was generated to understand how the effect of age on long-term functional outcome changes
depending on the acute AIS grade. A second linear regression model investigating the modifying
effects of age was produced adjusting for additional relevant confounding variables. Of 729
patients in the combined dataset, 376 met the eligibility criteria. The mean age was 43.2 (±16.9),
with a total of 41 patients (10.9%) older than 65. In the univariable analysis there was no
significant age related difference in motor recovery at follow-up, however there was a
significantly lower mean FIM motor score observed amongst the older group at one year
(p<0.05). In the multivariable analysis, age was found to have a significant modifying effect on
the relationship between acute AIS grade and future functional outcome (p<0.05). The
interaction plot revealed that while older patients had decreased follow-up FIM motor scores
overall, this effect was largest for AIS B and C patients and smaller for AIS A and D patients.
Adjustment for additional covariates in the second linear model confirmed these results. These
findings will help to facilitate enhanced clinical communication as well as potentially aid in the
development of customized treatment and rehabilitation protocols.
119
7.2 Introduction
Increases in average life expectancy are leading to a shift in societal demographics, with the
proportion of the US population aged 65 and older expected to double over the next 40 years266
.
This demographic restructuring is likely to lead to changes in the epidemiology of traumatic
spinal cord injury (SCI) and result in a greater proportion of injuries amongst the elderly6,221
. As
a result, there is a need to achieve an enhanced understanding of how increasing age impacts
clinical outcomes after SCI. Such an understanding will facilitate the counseling of patients and
families within the acute clinical realm, as well as the design of patient-specific rehabilitation
programs within the sub-acute and chronic phases post SCI.
The existing evidence suggests that while older SCI patients experience a similar degree of
motor recovery as younger patients, older patients experience diminished functional outcomes at
long-term follow-up150,267-269
. However, all of these studies have assumed that the impact of age
on functional outcomes is the same across the spectrum of injury severity. For example, this
assumption implies that the effect of age on function is the same for an ASIA (American Spinal
Injury Association) impairment scale (AIS) grade A patient as it is for an AIS grade B, C or D
patient. In order to test this assumption it is necessary to evaluate age as a potential effect
modifier governing the relationship between acute neurological status and future functional
status (Figure 7-1). An effect modifier is a variable “M” whereby a given predictor variable and
outcome variable have a different relationship between each other at the various levels of “M”270
.
Given this background, we hypothesize that the impact of age on functional outcome is
differential depending on baseline injury severity. To explore this hypothesis, the primary
objective of the current study was to investigate the role of age as an effect modifier in the
relationship between acute neurological status after traumatic SCI and long-term functional
outcome. This was performed by applying well defined statistical procedures to a combined
prospective multicenter SCI dataset.
120
Figure 7-1: Diagram depicting study goal: To understand how the relationship between
baseline neurological severity and long-term functional status varies depending on age.
7.3 Methods
7.3.1 Study Population
Two prospective SCI datasets were combined and used as the basis for all analyses. The North
American Clinical Trials Network for SCI (NACTN) database is a prospective registry that was
created in 2006 and enrolls patients from eight North American centers210
. The Surgical Timing
in Acute SCI Study (STASCIS) was a comparative effectiveness study of surgical timing that
enrolled patients from seven North American centers during a 7 year period beginning in
2002119
. These datasets contain common acute variables as well as outcome data at 6 and 12
month follow-up points. There was no overlap in patient enrollment between these datasets.
After research ethics board approval at each participating site was obtained, the databases were
harmonized and combined to produce a single dataset. Within this combined dataset, only adult
SCI patients (age>16) with an acute neurological exam performed within the first 3 days after
injury were considered in this analysis.
121
In all cases, neurologic assessment was completed at acute care admission according to the
International Standards for Neurological Classification of Spinal Cord Injury by a trained
physician, nurse or research assistant48
. Injury characteristics were classified according to
neurologic level of injury (NLI), ASIA motor score (AMS), ASIA sensory score (ASS) and the
overall ASIA Impairment Scale (AIS) grade. All patients received medical support according to
the recommendations of the 2002 American Association of Neurological Surgeons guidelines for
the treatment of SCI95,96
. When methylprednisolone was administered, a 24-hour high dose
infusion was given, according to the recommendations of the NASCIS-2 study105,108
. Whenever
possible, patients underwent surgical decompression of the spinal cord within 24 hours of injury.
Remaining surgical decisions were made on an individual basis by the attending spine surgeon
involved. After acute care discharge, patients underwent a customized course of rehabilitation
therapy.
7.3.2 Age as a predictor variable
To explore the impact of age as an effect modifier, it was transformed from a continuous variable
into a binary variable. Given that there is no standard definition of what constitutes an ‘old’ vs.
‘young’ SCI patient, age was dichotomized at a threshold of 65 years old, which is the cutoff
utilized by the United States Census Bureau to define the elderly population271
. For purposes of
performing a sensitivity analysis, evaluating a different threshold, all analyses were repeated
using age dichotomized at 60 years old.
7.3.3 Outcome and follow-up
The functional outcome of interest was Functional Independence Measure (FIM) motor score at
1 year follow-up. This functional outcome measure was created for generic use in the clinical
neurosciences and has been found valid and reliable for use in the setting of traumatic SCI78,226
.
Completion of the FIM motor score results in a discrete value between 13 and 91, with a higher
score indicating superior functional outcome. If 1 year follow-up data was not available for a
given patient, 6 month follow-up data was used instead, a practice which has been used and
shown appropriate in previous studies156
.
122
7.3.4 Statistical Methods
All analyses were performed using SAS version 9.3 (Cary, NC, USA). Descriptive analyses were
performed to characterize the demographics and injury characteristics of the patient population.
Missing data amongst predictor variables was assumed to be missing at random and was
accounted for using a multiple imputations procedure (Markov chain Monte Carlo methods) with
10 imputation samples. The first imputed sample was used as the basis for all subsequent
analyses. Such imputation is recommended as being more statistically robust, and less biased,
than removing cases with missing data points211,217
. The modifying effects of age on the
relationship between admission neurological status and follow-up functional outcome was
determined using regression procedures as described by Baron and Kenny272
. In the first linear
regression model we included age dichotomized at 65 and admission AIS grade, as well as their
interaction effect, as covariates modeled relative to the outcome variable of follow-up FIM motor
score. Effect modification exists if the interaction term explained a statistically significant
amount of the variance of the outcome variable, which equates to a p-value <0.05 for the
interaction term. This model was supplemented by graphing an interaction plot for dichotomized
age and admission AIS grade relative to follow-up FIM motor score. In the second linear
regression model, in addition to age, admission AIS grade and their interaction term, we also
included other covariates that could potentially influence long term functional status including
gender, admission Glasgow Coma Scale (GCS) score, neurological level of injury, admission
AMS and steroid administration, modeled relative to the outcome variable of follow-up FIM
motor score. As a sensitivity analysis, a second set of models was produced using age 60 as the
cut-off.
7.4 Results
7.4.1 Patient Population
Of the 729 patients within the combined dataset, 376 patients met eligibility criteria with a
baseline neurological exam available within the first three days after injury and with a FIM
motor score available at 6 months or 1 year follow-up. Amongst these patients, the overall mean
age was 43.2 (±16.9) years old, with a total of 70 patients (18.6%) older than 60 years old and a
total of 41 patients (10.9%) older than 65 years old. Table 7-1 compares baseline characteristics
between older and younger patients using both age cutoff points. There were no differences
123
between the young and old groups with respect to neurological status at admission as measured
by AIS grade, AMS or neurological level of injury (p>0.05). The older groups had a higher
proportion of female patients and a higher proportion of injuries secondary to falls, as compared
to the younger groups (p<0.05).
Table 7-1: Patient characteristics at acute hospital admission
Panel A: Age 65 used to distinguish Older vs. Younger
Characteristic Younger
n=335
Older
n=41
p-value
Age 39.7(±14.2) 72.5(±5.6) p<0.01
Male Gender 267(79.7%) 27(65.9%) p=0.04
Admission AIS
grade
AIS A
AIS B
AIS C
AIS D
126(37.6%)
53(15.8%)
49(14.6%)
107(31.9%)
10(24.4%)
10(24.4%)
9(22.0%)
12(29.3%)
P=0.19
Mean Admission
ASIA Motor Score
41.4(±31.7) 40.7(±30.2) P=0.91
Neurological Level
Cervical
Thoracic or
Lumbar
309(92.2%)
26(7.8%)
38(92.7%)
3(7.3%)
P=0.92
Mean Glasgow
Coma Scale Score
13.9(±3.0) 14.1(±2.6) P=0.77
Injury Etiology
MVA
Fall
Other
136(41.0%)
106(31.9%)
90(27.1%)
10(26.3%)
26(68.4%)
2(5.3%)
P<0.01
124
Panel B: Age 60 used to distinguish Older vs. Younger
Characteristic Younger
n=306
Older
n=70
p-value
Age 37.5(±12.9) 68.3(±6.7) P<0.01
Gender 248(81.1%) 46(65.7%) P<0.01
Admission AIS
grade
AIS A
AIS B
AIS C
AIS D
120(39.2%)
48(15.7%)
43(14.1%)
95(31.1%)
16(22.9%)
15(21.4%)
15(21.4%)
24(34.3%)
P=0.06
Mean Admission
ASIA Motor Score
40.7(±31.6) 44.0(±31.4) P=0.43
Neurological Level
Cervical
Thoracic or
Lumbar
280(91.5%)
26(8.5%)
67(95.7%)
3(4.3%)
P=0.23
Mean Glasgow
Coma Scale Score
14.0(±2.8) 13.7(±3.3) P=0.48
Injury Etiology
MVA
Fall
Other
132(43.6%)
87(28.7%)
84(27.7%)
14(20.9%)
45(67.1%)
8(11.9%)
P<0.01
7.4.2 Univariable analysis
Table 7-2 compares functional and neurological outcomes between the younger and older groups
using both age cutoff points. Using the age 65 cutoff there was a significantly lower mean FIM
motor score observed amongst the older group as compared to the younger group (p=0.03).
There was a similar trend observed using the age 60 cutoff, however this difference failed to
reach significance (p=0.08). There were no significant differences in motor neurological
recovery, as measured by AMS improvement, between the younger and older groups, regardless
of the cutoff considered (p>0.05).
125
Table 7-2: Functional and Neurological Outcomes at Follow-up
Panel A: Age 65 used to distinguish Older vs. Younger
Outcome Younger
N=335
Older
N=41
p-value
FIM motor score 64.2(±28.0) 54.0(±31.7) P=0.03
Change in ASIA
motor score
24.1(±23.4) 20.7(±17.2) P=0.37
Panel B: Age 60 used to distinguish Older vs. Younger
Outcome Younger
N=306
Older
N=70
p-value
FIM motor score 64.3(±28.1) 57.6(±30.2) P=0.08
Change in ASIA
motor score
23.9(±23.7) 23.1(±18.9) P=0.80
7.4.3 Multivariable analysis
Table 7-3 represents the results of the regression analyses investigating the modifying effect of
age on the relationship between admission neurological status and future functional status.
Overall, age was found to be a significant effect modifier, regardless of whether the age 65 or
age 60 cut-off was considered; older patients demonstrated decreased FIM motor scores at
follow-up however the relative impact of this age effect varied depending on the acute AIS
grade. As visualized in the interaction plot (Figure 7-2), this effect was particularly marked for
AIS B and AIS C patients and less prominent for AIS A and AIS D patients. In the multivariable
analysis adjusting for relevant confounding variables, the modifying effects of age on the
relationship between admission AIS grade and follow-up FIM motor score remained significant
(Table 7-4). Further, after covariate adjustment, the impact of the age effect continued to be
largest for AIS B and AIS C patients.
126
Table 7-3: Results of linear regression analysis relative to outcome variable FIM motor
score
Panel A: Age used to distinguish Older vs. Younger
Covariate Parameter estimate p-value
Admission AIS grade
AIS A
AIS B
AIS C
AIS D (ref)
-51.5
-56.0
-26.6
-
P<0.01 (overall)
P<0.01
P<0.01
P<0.01
Age
<65 years old
≥65 years old (ref)
1.8
-
P<0.01
Interaction (Age * AIS grade)
AIS A ≥65y.o. vs AIS A <65y.o.
AIS B ≥65y.o. vs AIS B <65y.o.
AIS C ≥65y.o. vs AIS C <65y.o.
AIS D ≥65y.o. vs AIS D <65y.o. (ref)
-5.7
-28.8
-12.7
-
P=0.02 (overall)
P=0.53
P=<0.01
P=0.18
-
Panel B: Age 60 used to distinguish Older vs. Younger
Covariate Parameter estimate p-value
Admission AIS grade
AIS A
AIS B
AIS C
AIS D (ref)
-53.6
-52.5
-23.4
-
P<0.01 (overall)
<0.01
<0.01
<0.01
-
Age
<60 years old
≥60 years old (ref)
1.8
-
P<0.01
Interaction (Age * AIS grade)
AIS A ≥60y.o. vs AIS A <60y.o.
AIS B ≥60y.o. vs AIS B <60y.o.
AIS C ≥60y.o. vs AIS C <60y.o.
AIS D ≥60y.o. vs AIS D <60y.o. (ref)
-8.3 -27.1
-10.2
-
P=0.01 (overall)
P=0.24 P<0.01
P=0.18
-
127
Table 7-4: Results of multivariable analysis adjusted for additional variables
Panel A: Age 65 used to distinguish Older vs. Younger
Panel B: Age 60 used to distinguish Older vs. Younger
Covariate Parameter estimate p-value
Admission AIS grade
AIS A
AIS B
AIS C
AIS D (ref)
-25.2
-25.7
-11.5
-
P<0.01(overall)
P<0.01
P<0.01
P=0.16
-
Age
<65 years old
≥65 years old (ref)
0.3
-
P<0.01
Male Gender -0.3 P=0.88
Steroid Administration 1.2 P=0.54
Acute Glasgow Coma Score 0.4 P=0.30
Neurological Level of Injury
Cervical (ref)
Thoracolumbar
-
8.6
P=0.03
Acute ASIA Motor Score 0.5 P<0.01
Interaction (Age * AIS grade)
AIS A ≥65y.o. vs AIS A <65y.o.
AIS B ≥65y.o. vs AIS B <65y.o.
AIS C ≥65y.o. vs AIS C <65y.o.
AIS D ≥65y.o. vs AIS D <65y.o. (ref)
-8.4
-24.0
-19.5
-
P=0.02(overall)
P=0.31
P=0.01
P=0.02
-
Covariate Parameter estimate p-value
Admission AIS grade
AIS A
AIS B
AIS C
AIS D (ref)
-25.4
-21.8
-5.1
-
P<0.01(overall)
P<0.01
P<0.01
P=0.42
-
Age
<60 years old
≥60 years old (ref)
1.7
-
P<0.01
Male Gender -0.94 P=0.68
Steroid Administration 0.2 P=0.92
Acute Glasgow Coma Score 0.3 P=0.42
Neurological Level of Injury
Cervical (ref)
Thoracolumbar
-
8.0
P=0.04
Acute ASIA Motor Score 0.5 P<0.01
Interaction (Age * AIS grade)
AIS A ≥60y.o. vs AIS A <60y.o.
AIS B ≥60y.o. vs AIS B <60y.o.
AIS C ≥60y.o. vs AIS C <60y.o.
AIS D ≥60y.o. vs AIS D <60y.o. (ref)
-8.8
-21.7
-13.6
-
P=0.01(overall)
P=0.17
P=0.01
P=0.04
-
128
Figure 7-2: Interaction plot demonstrating the modification effect of age on the relationship
between acute injury severity and functional outcome
129
7.5 Discussion
This is the first study to examine the modification effects of patient age on the relationship
between admission neurological status and long-term functional outcome after traumatic SCI.
The results of our analyses indicate that age is a significant effect modifier governing the
relationship between admission AIS grade and FIM motor score at follow-up. Stated plainly, it is
clear that while older patients will experience, on average, reduced future functional outcome as
compared to younger patients, this age related effect changes depending on patients’ AIS grade
at admission. The largest difference in functional outcome between the older and younger
groups was seen for those that were AIS grade B and C at admission. In comparison, the
magnitude of this difference was smaller for AIS A and AIS D patients, regardless of the age
cutoff considered.
To elaborate on the differential impact of age across the spectrum of injury severity, it is felt that
the reason for the diminished impact of age in the context of AIS A injuries is likely secondary to
the severe degree of primary neural tissue disruption that is often associated with these injuries.
As a result, functional outcomes are generally poor and additional variables, such as age, may
play less of a role in determining future functional status in this particular group. While AIS
grade B and C patients still harbor severe lesions, these individuals have been documented to
possess a greater potential for neurological recovery as compared to AIS A patients16
. In this
setting, other variables, such as older age, may prevent the translation of these neurological gains
into functionally meaningful recovery. On the opposite end of the severity spectrum, as regards
AIS D injuries, functional recovery at long term follow-up is usually excellent, as a result of a
comparatively reduced extent of spinal cord damage. In light of this favorable natural course of
recovery, the independent effects of additional variables, such as age, may be diminished. In
summary, it is felt that the increased age related effect seen within the AIS B and C groups
reflects the fact that these are lesions of intermediate severity where the potential for functional
recovery may be impacted by additional factors. This is in contrast to the very severe AIS A
injuries and the comparatively less serve AIS D injuries, where the potential for recovery is more
defined and less susceptible to the influence of outside factors.
Several studies have attempted to investigate the independent effects of age on clinical outcomes
after SCI. In a re-analysis of data from the NASCIS III study, Furlan et al demonstrated no
130
association between age and FIM scores at 6 months and 1 year follow-up in an unadjusted
analysis150
. However, after adjusting for relevant variables, such as severity of injury (complete
vs. incomplete), a linear increase in age was associated with decreased FIM scores at the same
follow-up points. In the same study, there was no association between age and motor recovery at
1 year follow-up. Cifu et al examined the impact of age on functional and neurological outcomes
in the rehabilitation setting post SCI267,268
. After accounting for the level and severity of injury,
increased age was found to be associated with reduced functional recovery as measured by FIM
change scores from rehabilitation admission to discharge for quadriplegics and paraplegics alike.
In the same studies, older tetraplegic patients were noted to experience reduced AMS
improvement from acute hospital admission to rehabilitation discharge, however there were no
age related differences in motor recovery observed for paraplegic patients. Overall, while there
continues to be controversy surrounding the effect of age on neurological recovery, the existing
evidence seems to suggest that older age is associated with worse functional status at long-term
follow-up124,177,273
. The current study has added to this body of evidence by demonstrating that
the effect of age on long-term functional status is differential, depending on patients’ admission
AIS grade.
From a clinical perspective, the results of the current study may allow for improved
communication and patient counseling within the acute clinical realm by permitting more
accurate outcome predictions based on acute clinical information. From the perspective of sub-
acute and chronic care, this research may allow for the development and institution of patient-
specific rehabilitation programs. For instance, since older AIS grade B patients have been shown
to experience significantly worse functional outcomes as compared to younger AIS B patients, in
spite of similar of similar neurological recovery, perhaps older AIS B patients should be targeted
with more intensive rehabilitation strategies to help optimize long-term functional status. From a
research perspective, the results of this study may be used to help stratify patients within the
context of future clinical trials and clinical therapeutic protocols.
7.5.1 Study Limitations
It should be noted that in recent years, in addition to FIM, other functional outcome measures,
such as the Spinal Cord Independence Measure (SCIM), have been used to assess functional
outcome after SCI. However in the datasets used to complete this analysis, FIM was the primary
131
functional outcome assessment tool incorporated. It is speculated that since many of the items
contained in the SCIM questionnaire are similar to those in the FIM, that the results of this
analysis would not have differed substantially if SCIM had been used as the functional outcome
measure instead of FIM motor score. Finally, although we performed our analyses using a large
multicenter dataset, these findings must be validated using an external dataset before applied in
the clinical and research realms.
7.6 Conclusion
While increasing age has previously been associated with reduced functional outcome after SCI,
in the current study, we have attempted to provide further granularity on the topic by exploring
age as a potential effect modifier governing the relationship between acute neurological status
and functional outcome. This analysis has demonstrated that the negative impact of older age on
functional outcome varies across the spectrum of injury severity, with the largest effect noted for
AIS B and C patients and a comparatively smaller effect seen for AIS A and D patients. In the
future these findings will likely serve to assist in the counseling of SCI patients and their
families, and may help to direct the formation of customized treatment and rehabilitation
protocols, tailored to patient specific characteristics.
132
8 Chapter 8: Summary of Findings, General Discussion, Thesis Limitations and Future Directions
133
8.1 Summary of Findings and Unifying Discussion
As introduced at the outset of this thesis, there exists a strong imperative to provide clinicians
and researchers alike with an evidence based means of predicting future outcomes for patients
with traumatic SCI during the acute period after injury. Achieving progress on this topic serves
not only to aid communication in the clinical realm, but also permits enhanced study of this
patient population from the research perspective. The preceding analyses have each represented
separate aims engineered to probe different elements of the overall hypothesis that neurologic,
functional and complication related outcomes can be accurately predicted using acute patient,
injury and radiological features. Our use of a large combined prospective dataset permitted the
adequate sample size and the relevant data fields necessary to fully address each of these key
aims. Guided by the overarching hypothesis and specific aims, analysis of this dataset allowed
for: 1) the development and internal validation of a clinico-radiographic model predicting
functional outcome, 2) the exploration of facet dislocation (FD) as a radiological predictor of
neurological outcome, 3) the identification of the combination of variables predicting acute in-
hospital complication development after cervical SCI and the subsequent construction of a model
predicting complication occurrence, and 4) the exploration and clarification of the effect of age
on functional outcome. Figure 8-1 provides an overview of the main relationships explored
between acute predictor variables and outcome variables across the multivariable analyses
presented in each chapter of this thesis. In this section, these relationships, and the key findings
emanating from this body of work, are further contextualized and explained for each of the acute
predictor variables and outcomes that were considered.
134
Figure 8-1: Summary of Main Relationships between Acute Predictors and Outcomes
Explored in Multivariable Analyses throughout Thesis
AIS: ASIA Impairment Scale; AMS: ASIA motor score; IM: Intramedullary; FIM: Functional
Independence Measure;
Solid line: p<0.05 in multivariable analyses; Dotted line: p>0.05 in multivariable analyses
8.1.1 Clinical Variables
In each of the four analyses presented in chapters 4-7, while the predictive importance of many
different variables were assessed, neurological exam features were consistently important
predictors, regardless of the outcome considered. In particular, lesion severity, as measured by
the acute AIS grade, was a highly significant predictor of all three outcomes considered. As
discussed in chapter 2, previous studies have raised concern surrounding the accuracy of the
neurological exam when performed within 3 days of injury, thereby bringing into question the
utility of acute exam related variables for purposes of outcome prediction143
. In spite of these
concerns, the goal of the current body of work was to base predictions entirely on information
135
available to clinicians at the time of patients’ acute hospital admission. As a result, all exam
related data used to perform the preceding analyses were obtained within the first 3 days after
hospital admission for all patients. Although it was not possible to compare the predictive
capacity of these acute variables to those obtained from the sub-acute period, based on the
analyses presented above, this acute exam data proved to be a robust predictor of all outcomes
considered. Whenever possible, the effects of potential inaccuracies surrounding the acute
neurological assessment were mitigated in the current analyses through selection of neurological
exam predictors that were felt to be less susceptible to acute biases. As an example, in all four
analyses, AIS grade was chosen as one of the main neurologic exam related predictors of
interest. As opposed to AMS, which requires precise quantification of individual muscle strength
in 20 myotomes, AIS grade is more easily obtained in the acute care period, requiring
classification into only one of 5 broad categories. While acute care distractions such as patient
fatigue, pain or concomitant injuries are likely to impact on the accuracy of AMS values148,274
, it
was felt that the accuracy of AIS grade assessments were less susceptible to the influence of
these same factors. When AMS was used as a predictor of interest in the clinico-radiographic
model described in chapter 4, the dichotomized value (above or below a score of 50) was
considered. This approach was felt to be more tolerant to any inaccuracy present in the raw
ordinal scores, thereby limiting the impact of the aforementioned acute distractions on model
validity. ASS was not incorporated in any of the preceding analyses as the incremental predictive
value contributed by this parameter, over and above AIS grade, was felt to be minimal.
Apart from neurological exam parameters, the only other injury related variable considered in
this thesis was injury etiology. The literature review presented in chapter 2 did not provide
consistent support for the predictive importance of any injury mechanism in relationship to any
of the considered outcomes. However, the main challenge when examining this topic is the lack
of homogeneity between studies with respect to the classification of injury etiology. In analyzing
the combination of factors associated with acute complication development, high energy injury
mechanisms were associated with a trend towards an increased incidence of these events in both
the univariable and multivariable analyses. Overall, further future analyses employing a
standardized template for etiology categorization will be necessary to clarify the impact of this
variable on clinical outcomes post SCI.
136
8.1.2 Demographic Variables
A sizeable body of literature has accumulated over the years to support the conclusion that while
neurological recovery is unaffected by increasing age, long-term functional outcome is decreased
amongst elderly patients150,275
. However, the existing studies on this topic have suggested that
this detrimental impact of age is uniformly applicable to older SCI patients and do not consider
how additional patient and injury characteristics might contribute to this effect. The analysis
presented in chapter 7 has contributed new understanding to this topic by demonstrating that
while older patients do experience diminished functional outcomes in the long-term, the negative
effect of increased age is differential across the spectrum of injury severity. The most substantive
difference in functional outcome between older patients and younger patients was discovered for
those who presented acutely with AIS grade B or C injuries. In contrast, there was very little
difference in functional outcomes observed between the age groups for patients who presented at
the extremes of injury severity with either AIS grade A or AIS grade D. It is felt that these
findings reflect the fact that AIS B and C injuries are lesions of intermediate severity where the
potential for functional recovery may be impacted by additional variables such as age. In
contrast, for AIS A and D injuries, the profile for recovery is more defined and less susceptible
to the influence of extraneous variables. Overall, while this analysis has confirmed the negative
impact of increasing age on function, it has also served to refine and improve upon our
understanding of how this demographic variable affects outcome depending on additional injury
related characteristics.
In addition to considering the impact of age in relation to long-term functional outcome, the
effect of increasing age was also considered in relation to the occurrence of acute in-hospital
complications. While age has consistently shown to be of predictive importance with respect to
the development of such complication events, few studies have evaluated the independent effect
of age by adjusting for other relevant confounding variables, such as medical comorbidities. The
analysis presented in chapter 6 showed increasing age to significantly predict acute complication
occurrence, even after adjusting for covariates such as comorbid illness, acute AIS grade, injury
etiology and steroid administration.
While the predictive capacity of additional demographic features such as gender and race were
reviewed in Chapter 2, these were not explored in substantial detail in subsequent chapters. At
137
present time, there is lack of evidence supporting the predictive importance of these variables in
relation to both short-term and long-term outcomes post SCI.
8.1.3 Radiological Variables
When considering the current analyses in the setting of previous work evaluating the role of
radiological variables in predicting outcome after SCI, several points are pertinent. First, of the
quantitative and qualitative MRI variables that have been considered for their potential
importance in predicting outcome, qualitative intramedullary signal characteristics are the best
studied and have demonstrated value in this capacity, as discussed in Chapter 260,179,187
.
However, by in large, existing studies evaluated the importance of these MRI variables in
isolation, without considering them in the context of a broader array of diagnostic and clinical
variables that also have established value in predicting outcome. The goal of the current research
was not to develop or investigate the use of new MRI variables for purposes of outcome
prediction, but rather to use the existing variables known to be of importance in combination
with other non-radiological variables. Theoretically this approach optimizes the accuracy of
outcome predictions by permitting consideration of the holistic package of information that
clinicians use in the acute clinical realm to forecast predictions for the future. In chapter 4 we
showed that acute MRI intramedullary signal characteristics (no signal, signal consistent with
edema and signal consistent with hemorrhage) could be used in combination with demographic
and exam features to augment predictions of long-term functional outcome. More than anything
this analysis has demonstrated the feasibility, and potential value, in using radiological variables
as a supplement to provide an understanding of anatomic injury severity. Such an understanding
is especially of value in the acute injury setting when external influences can interfere with the
accuracy of neurological examination. In contrast to examination predictors, radiological
variables are immune to the described acute care biases and as a result, their accuracy and
predictive importance is unlikely to vary according to the time period. In the future, as the
accuracy of MRI in quantifying injury to neural tissue improves, the incremental predictive
utility contributed by these imaging variables is likely to increase in a commensurate fashion.
In the combined NACTN/STASCIS dataset, a large proportion of patients did not undergo MRI
within the first 3 days after injury, bringing into question the generalizability of a prediction
model which includes MRI variables. However the missing imaging data can be explained along
138
several lines. First, because the subjects contained in these datasets were enrolled up to 10 years
ago, several of the participating institutions, in the early years of enrolment, may not have
considered MRI as a standard part of the diagnostic work-up of an SCI patient. In the interim,
published guidelines have led many clinicians involved in the acute management of SCI to
prioritize early MRI as an essential element of the diagnostic work-up60
. Second, since STASCIS
focused on cervical injuries, a significant proportion of injured patients included in this dataset
had facet dislocations (FD). There is some remnant controversy about the role of acute MRI in
the setting of FD; while some clinicians forego MRI to expedite closed or open reduction of the
spinal column, others continue to rely on the anatomic information provided by MRI prior to
proceeding with reduction247
. As a result, a proportion of patients with cervical SCI secondary to
FD did not undergo MRI as a consequence of the preferences of the treating physician on this
subject. All perspectives considered, MRI has become the most useful imaging modality to direct
diagnosis, treatment and outcome predictions after SCI, justifying the use of MRI variables in the
clinico-radiographic model described in Chapter 460
.
In addition to compiling the existing data on MRI predictors of outcome, Chapter 2 also
underlined the absence of literature identifying CT or X-ray related variables of proven
predictive importance. Although CT and X-ray are poorly suited for visualizing the spinal cord
and neural structures, they are highly suited for diagnosing injury to the bony spinal column.
Further, at least one of either CT or X-ray is performed at admission in virtually every suspected
case of SCI. This is in contrast to MRI which, as acknowledged above, may not be performed in
certain patients with SCI for reasons including clinician preference, presence of implanted
metallic hardware or, in certain regions, inadequate patient access to the technology. In addition,
while MRI can often take several hours to organize and perform, X-ray and CT are typically
performed within minutes of hospital arrival55
. In chapter 5, the identification of FD as
significant predictor of motor recovery at 1 year follow-up, established the feasibility of using
patterns of bony injury, detected on acute X-ray or CT, to predict outcome. While these findings
will require confirmation in future studies, FD appears to be the first CT or X-ray based finding
to demonstrate predictive significance in the context of traumatic SCI.
139
8.1.4 Outcome variables
Long-term functional outcome, as measured by Functional Independence Measure (FIM) motor
score, was the outcome most often considered throughout this thesis. As opposed to neurological
outcome which refers to the direct assessment of the integrity and health of ones’ nervous
system, assessment of function allows us to understand how one interacts with their environment
and completes useful daily tasks. As a result, while neurological outcome is influenced
predominately by the severity of neurological injury, functional outcome is also influenced by
non-injury related factors, such as age, that affect the translation of neurologic function into
useful activity. The fact that a variety of different factors have the potential to underlie functional
outcome post SCI, explains our decision to often consider this outcome throughout the preceding
analyses. The work presented in this thesis has expanded our understanding of how such factors
can be used to improve functional outcome predictions after SCI. The analysis presented in
chapter 4 demonstrated, for the first time, the feasibility of using a combination of acute
variables, collected within the first 3 days after injury, to predict future functional status.
Furthermore, this work established that acute radiological, demographic and exam related
variables can be combined to accurately predict functional independence. Next, although
functional outcome has previously shown to be negatively affected by increasing age, the
analyses presented in chapter 7 brought further clarity to this topic by demonstrating this
negative age related effect to vary across the spectrum of injury severity, as explained above.
In chapter 5 neurological outcome, as defined by change in AMS from admission to 1 year
follow-up, was considered as the primary outcome of interest. Change in AMS, as reflected by
its selection as the primary outcome in previous SCI therapeutic trials, is considered by many
within the SCI research community to be the preferred measure of neurological recovery. As
identified in the preceding systematic review (Chapter 2), numerous patient, injury and MRI
related variables have been considered as potential predictors of this outcome, with injury
severity (AIS grade) generally having shown the greatest utility in this regard. However, little
investigation has probed the use of CT or X-ray related predictors of neurological recovery. The
analysis in chapter 5 showed that presence of FD on admission X-ray or CT was a significant
predictor of motor neurological recovery, even after adjusting for the baseline degree of
neurologic severity. Overall, these findings provide clinicians and researchers alike with a new
140
imaging predictor of future neurological status that can be obtained independent of a patients’
capacity to cooperate with neurological exam.
Acute complication development was the third outcome considered in relation to a variety of
clinical predictor variables within this thesis. Since the nature and severity of complication
events differ depending on the time elapsed since injury, it is essential to understand the specific
factors which portend a higher risk of complication development at each stage of patient care.
During the chronic stages after injury, the period considered in most of the topical studies,
complications are dominated by events related to prolonged immobilization and chronic
disability such as decubitus ulcer formation, urinary tract infections and the development of
psychiatric affective disorders. In contrast, during the initial in-hospital period, complications
events are more often severe and life-threatening in nature, reflecting the increased acuity of this
stage of patient care. In the literature review presented in chapter 2, of 14 articles identifying
clinical predictors of complication development, only 4 pertained to complications encountered
during the acute in-hospital period, with none of these employing multivariable techniques to
identify the combination of factors predicting this outcome. To overcome this void, in chapter 5,
a core set of five patient, injury and treatment related variables were identified that predicted,
with acceptable predictive discrimination, acute in-hospital complication occurrence after
cervical SCI. These variables form the basis of a new predictive model that can be used to
estimate individuals’ probability of acute complication at admission. Ideally this model will be
used to identify high risk patients that can subsequently be monitored in an intensive fashion to
prevent complication development and the attendant morbidity of such events.
8.2 Thesis Limitations
First, the presence of missing data within the combined dataset, both amongst acute predictor
variables and outcome variables, is a limitation that must be acknowledged for the analyses
presented in Chapters 4-7. This is particularly true in relation to some of the imaging related
data, for reasons enumerated in the discussion section above. For predictor variables, imputation
methods were used to minimize the introduction of bias and the negative impact of missingness
on sample size. For outcome variables, in each of the chapters, there did not appear to be
clinically significant differences in the characteristics of patients included and those excluded
due to inadequate follow-up. In spite of attempts to mitigate the effect of missing data, it is
141
acknowledged that complete datasets would have been preferable for completion of these
analyses. Second, with respect to the functional outcome measure used in these analyses, it
should be acknowledged that although FIM is among the best studied of any functional outcome
measure in the context of SCI, SCIM overcomes some of the responsiveness issues identified
with FIM. However, because SCIM is a relative newcomer, the databases employed in this study
did not routinely collect this information. It is speculated that since many of the items contained
in the SCIM questionnaire are similar to those in the FIM, that the results of this analysis would
not have differed substantially if SCIM had been used as the functional outcome measure instead
of FIM motor score. Third, the acute injury window was defined across these analyses as the first
3 days after injury, in accordance with the definitions put forward by the 2007 International
Campaign for Cures of spinal cord injury Paralysis70
. It is acknowledged however that this
definition is somewhat arbitrary and may need revisiting in the future, especially with emerging
evidence supporting early surgical intervention within the first 24 hours after injury119,120
. Fourth,
while one year was considered to represent long-term follow-up for both neurological and
function outcome in this thesis, it has been shown that patients can experience incremental
improvements beyond that time point75
. However, given that only a small proportion of patients
experience such recovery after one year, this was felt to be a reasonable assumption for the
purposes of this thesis. Finally, in chapters 4 and 7, analyses incorporated cervical, thoracic and
lumbar SCI patients. However, it is acknowledged that in both of these analyses, the majority of
patients studied had a cervical neurological level, with a relatively low proportion of patients
with thoracic or lumbar SCI. As a result, prior to clinical use, these findings will need to be
validated in external datasets, containing a greater proportion of thoracic and lumbar SCI
patients.
142
8.3 Future Directions
With the completion of this work, the next immediate step is to begin the process of establishing
external validity for the findings presented here within. As mentioned earlier, until such validity
is established, these results remain preliminary. Exploring the question of external validity will
require the use of new SCI datasets with data fields similar to the ones considered in the NACTN
and STASCIS datasets. In order to maximize generalizability, the new data should emanate from
centers that were not involved in enrolling subjects within NACTN or STASCIS, ideally from
outside North America. This would allow for confirmation of findings in a system or country
where assessment and treatment standards are subtly different from those observed in the North
American healthcare systems. Once such data is procured the next step would be to evaluate the
performance of the statistical models that were conceived in this thesis within the context of the
new datasets. As an example, for the clinico-radiographic prediction models described in chapter
4, the models themselves would be applied to the new data resulting in a new set of indices for
performance including R-squared and AUC values. These newly generated values would then be
compared to those obtained from the original data, to arrive at bias corrected indices. Assessment
of validity for such statistical models is an ongoing and iterative process that never leads to a
final conclusion of “valid” or “not valid”. Rather, through evaluation of model performance in
several different datasets, the generalizability of the model, and hence its appropriateness for use
in the clinical and research settings, becomes increasingly apparent. At present, efforts are
underway to obtain additional datasets for purposes of initiating such validity studies.
Once the generalizability of the preceding analyses are more completely understood, efforts will
be directed to translating these findings into more practical tools that can be used by clinicians
and researchers alike to help optimize clinical communication and facilitate improved
investigation. To this end, one potential strategy would be to simplify the predictive relationships
observed above into a straightforward predictive classification or score analogous to the Glasgow
Coma Score for traumatic brain injury. A simplified predictive classification system is more
facile for use is the often chaotic setting of acute clinical care as compared to the mathematical
equations presented above. Based on the analyses within this thesis, classifications predictive of
functional outcome and acute complications are currently under construction.
143
In chapter 7, age was found to significantly modify the relationship between acute injury severity
and long-term functional outcome. This raises the possibility of investigating the impact of
similar “third variable” effects on other predictor/outcome relationships in the context of SCI. In
addition to effect modification, it would also be of interest to investigate the mediating effects of
different variables on such relationships. To this end, future work will focus on developing a
theoretical framework for the relationship between predictor variables and outcomes,
incorporating such modification and mediation effects.
The current work has also highlighted the need for the development of improved imaging
variables that reflect injury severity and are predictive of outcome. Admittedly, the quantitative
presence of signal consistent with edema or hemorrhage within the spinal cord is a rather crude
method for this purpose as it does not consider the integrity of the individual anatomic tracts and
structures within the spinal cord itself. Fortunately imaging technology is evolving to a point
where high anatomic resolution will permit assessment of the structural integrity of individual
tracts. Further, the capacity of functional imaging modalities are progressing and may soon serve
to augment the structural information provided by traditional MRI by offering an imaging
representation of neurological function. When such technology becomes practical for use in the
acute clinical setting, assessing the added contributions of these imaging advances in predicting
outcome will be of definite interest.
Finally, outcomes considered in this thesis were limited to long-term functional outcome and
neurological recovery as well as acute in-hospital complications. However, the importance of
other outcomes not considered here, such as health related quality of life, pain and disability,
cannot be overemphasized. Future analyses investigating the underpinnings and predictors of
these and other outcomes are also planned with the overall goal of identifying specific factors
that can be targeted with intervention to improve the long-term health of SCI patients.
When considering these and other avenues of research stemming from the analyses presented
above, the opportunities for further work and investigation are truly unlimited.
144
References
1. Wilson JR, Forgione N, Fehlings MG. Emerging therapies for acute traumatic spinal cord
injury. CMAJ. Dec 10 2012.
2. Chin L, Mesfin F, Dawodu S. Spinal Cord Injuries.
http://emedicine.medscape.com/article/793582-overview#a0101. Accessed December 12,
2012, 2012.
3. Couris C, Guilcher S, Munce S, et al. Characteristics of adults with incident traumatic
spinal cord injury in Ontario Canada. Spinal Cord. 2010;48:40-44.
4. Pickett G, Campos-Benitez M, Keller J, Duggal N. Epidemiology of Traumatic Spinal
Cord Injury in Canada. Spine. 2006;31:799-805.
5. Noonan V, Fingas M, Farry A, et al. The incidence and prevalence of spinal cord injury
in Canada: A National Perspective. Neuroepidemiology. 2011;38:219-226.
6. Farry A, Baxter D. The Incidence and Prevalence of Spinal Cord Injury in Canada
Overview and estimates based on current evidence. 2010;
http://www.urbanfutures.com/reports/Report%2080.pdf. Accessed June 15, 2011.
7. Dryden D, Saunders L, Rowe B, et al. The Epidemiology of Traumatic Spinal Cord
Injury in Alberta, Canada. Canadian Journal of Neurological Sciences. 2003;30:113-121.
8. Pirouzmand F. Epidemiological trends of spine and spinal cord injuries in the largest
Canadian adult trauma center from 1986 to 2006. Journal of Neurosurgery Spine.
2010;12:131-140.
9. Ackery A, Tator C, Krassioukov A. A global perspective on spinal cord injury
epidemiology. J Neurotrauma. 2004;21:1355-1370.
10. Sekhon L, Fehlings M. Epidemiology,demographics, and pathophysiology of acute spinal
cord injury. Spine. 2001;26:S1-12.
11. Kraus J, Franti C, Riggins R, Richards D, Borhani N. Incidence of traumatic spinal cord
lesions. J Chronic Dis. 1975;28:471-492.
12. Martins F, Freitas F, Martins L, Dartigues J, Barat M. Spinal cord injuries epidemiology
in Portugal's central region. Spinal Cord. 1998;36:574-578.
13. O'Connor P. Incidence and patterns of spinal cord injury in Australia. Accid Anal Prev.
2002;34:405-415.
14. Levy L, Makarowo S, Madzivire D, Bhebhe E, Verbeek N, Parry O. Problems, struggles
and some successes with spinal cord injury in Zimbabwe. Spinal Cord. 1998;36:213-218.
145
15. Hart C, Williams E. Epidemiology of spinal cord injuries: a reflection of changes in
South African society. Paraplegia. 1994;32:709-714.
16. Marino R, Dittuno J, Donovan W, Maynard F. Neurologic Recovery after Traumatic
Spinal Cord Injury: Data from the Model Spinal Cord Injury Systems. Arch Phys Med
Rehab. 1999;80:1391-1396.
17. Catz A, Thaleisnik M, Fishel B, et al. Survival following spinal cord injury in Israel.
Spinal Cord. 2002;40:595-598.
18. Silberstein B, Rabinovich S. Epidemiology of spinal cord injuries in Novosibirsk, Russia.
Paraplegia. 1995;33:183-188.
19. Levi R, Hultling C, Nash MS, Seiger A. The Stockholm spinal cord injury study: 1.
Medical problems in a regional SCI population. Paraplegia. Jun 1995;33(6):308-315.
20. Nobunaga AI, Go BK, Karunas RB. Recent demographic and injury trends in people
served by the Model Spinal Cord Injury Care Systems. Arch Phys Med Rehabil. Nov
1999;80(11):1372-1382.
21. DeVivo MJ, Kartus PL, Rutt RD, Stover SL, Fine PR. The influence of age at time of
spinal cord injury on rehabilitation outcome. Arch Neurol. Jun 1990;47(6):687-691.
22. StatisticsCanada. Canada's Aging Population. 2004;
http://www.sustreport.org/signals/canpop_age.html. Accessed June 15, 2011.
23. Krassioukov A, Furlan J, Fehlings M. Medical Co-Morbidities, Secondary
Complications, and Mortality in Elderly with Acute Spinal Cord Injury. J Neurotrauma.
2003;20:391-399.
24. Geisler F, Coleman W, Grieco G, Poonian D. The Sygen multicenter acute spinal cord
injury study. Spine. 2001;26:87-98.
25. Hoque MF, Grangeon C, Reed K. Spinal cord lesions in Bangladesh: an epidemiological
study 1994 - 1995. Spinal Cord. Dec 1999;37(12):858-861.
26. Kraus J. Injury to the head and spinal cord: The epidemiological relevance of the medical
literature published from 1960 to 1978. J Neurosurgery. 1980:S3-10.
27. Krause J, Sternberb M, Lottes S, Maides J. Mortality after Spinal Cord Injury: An 11-
year prospective study. Arch Phys Med Rehab. 1997;78:815-821.
28. Krause J, Carter R, Pickelsimer E, Wilson D. A prospective study of health and risk of
mortality after spinal cord injury. Arch Phys Med Rehab. 2008;89:1482-1491.
29. O'Connor P. Survival after spinal cord injury in Australia. Arch Phys Med Rehab.
2005;86:37-47.
30. DeVivo M, Stover S. Long term survival and causes of death. Gaithersburg1995.
146
31. Dryden D, Saunders L, Rowe B, et al. Utilization of health services following spinal cord
injury: a 6-year follow-up study. Spinal Cord. 2004;42:513-525.
32. Winslow C, Bode R, Felton D, Chen D, Meyer R. Impact of respiratory complications on
length of stay and hospital costs in acute cervical spine injury. Chest. 2002;121:1548-
1554.
33. Dryden D, Saunders L, Jacobs P, et al. Direct health care costs after traumatic spinal cord
injury. J Trauma. 2005;59:443-449.
34. Rick-Hansen-Institute. High Cost of Healthcare. Information about SCI 2012;
http://www.rickhanseninstitute.org/en/information-about-sci/high-cost-of-healthcare.
Accessed December 12, 2012.
35. Amar A, Levy M. Pathogenesis and pharmacological strategies for mitigating secondary
damage in acute spinal cord injury. Neurosurgery. 1999;44:1027-1039.
36. Fehlings M, Sekhon L, eds. Cellular, ionic and biomolecular mechanisms of the injury
process. Chicago: American Association of Neurologic Surgeons; 2000. E B, CH T, eds.
Contemporary Management of Spinal Cord Injury: From Impact to Rehabilitation.
37. Tator C, Fehlings M. Review of the secondary injury theory of acute spinal cord trauma
with emphasis on vascular mechanisms. J Nerosurg. 1991;75:15-26.
38. Sandler A, Tator C. Review of the effect of spinal cord trauma on the vessels and blood
flow in the spinal cord. J Nerosurg. 1976;45:638-646.
39. Tator C, Koyanagi I. Secondary injury mechanisms of spinal cord trauma: a novel
therapeutic approach for the management of secondary pathophysiology with the sodium
channel blocker riluzole. J Neurosurg. 1997;86:483-492.
40. Tator CH, Koyanagi I. Vascular mechanisms in the pathophysiology of human spinal
cord injury. J Neurosurg. Mar 1997;86(3):483-492.
41. Sandler A, Tator C. Effects of acute spinal cord compression injury on regional spinal
cord blood flow in primates. J Neurosurg. 1976;45:660-676.
42. Jorgensen M, Diemer N. Selective neuron loss after cerebral ischemia in the rat: possible
role of transmitter glutamate. Acta neurologica Scandinavica. 1982;66:536-546.
43. Braughler J, Duncan L, Chase R. Interaction of lipid peroxidation and calcium in the
pathogenesis of neuronal injury. Cent Nerv Syst Trauma. 1985;2:269-283.
44. Schwartz G, Fehlings M. Secondary injury mechanisms of spinal cord trauma: a novel
therapeutic approach for the management of secondary pathophysiology with the sodium
channel blocker riluzole. Prog Brain Res. 2002;137:177-190.
45. Rothman S, Olney J. Glutamate and the pathophysiology of hypoxic-ischemic brain
damage. Ann Neurology. 1986;19:105-111.
147
46. Baptiste D, Fehlings M. Emerging drugs for Spinal Cord Injury. Expert Opin Emerg
Drugs. 2008;13:63-80.
47. ASIA, ed Standards for Neurological Classification of Spinal Injured Patients. Chicago:
American Spinal Injury Association; 1982.
48. Kirshblum SC, Waring W, Biering-Sorensen F, et al. Reference for the 2011 revision of
the International Standards for Neurological Classification of Spinal Cord Injury. J Spinal
Cord Med. Nov 2011;34(6):547-554.
49. Kirshblum SC, Burns SP, Biering-Sorensen F, et al. International standards for
neurological classification of spinal cord injury (revised 2011). J Spinal Cord Med. Nov
2011;34(6):535-546.
50. Marino RJ, Jones L, Kirshblum S, Tal J, Dasgupta A. Reliability and repeatability of the
motor and sensory examination of the international standards for neurological
classification of spinal cord injury. J Spinal Cord Med. 2008;31(2):166-170.
51. Mulcahey MJ, Gaughan J, Betz RR, Johansen KJ. The International Standards for
Neurological Classification of Spinal Cord Injury: reliability of data when applied to
children and youths. Spinal Cord. Jun 2007;45(6):452-459.
52. Marino R, Jones L, Kirshblum S, Tal J. Reliability of the ASIA motor and sensory
examination. J Spinal Cord Med. 2004;27:194.
53. Savic G, Bergstrom EM, Frankel HL, Jamous MA, Jones PW. Inter-rater reliability of
motor and sensory examinations performed according to American Spinal Injury
Association standards. Spinal Cord. Jun 2007;45(6):444-451.
54. Furlan JC, Fehlings MG, Tator CH, Davis AM. Motor and sensory assessment of patients
in clinical trials for pharmacological therapy of acute spinal cord injury: psychometric
properties of the ASIA Standards. J Neurotrauma. Nov 2008;25(11):1273-1301.
55. ACPS. Advanced trauma life support system program for physicians. Chicago: American
College of Physicians and Surgeons; 1993.
56. Mirvis S, Diaconis J, Chirico P. Protocol-driven radiologic evaluation of suspected
cervical spine injury: efficacy study. Radiology. 1989;170:831-834.
57. Acheson M, Livingston R, Richardson M, Stimac G. High-resolution CT scanning in the
evaluation of cervical spine fractures:Comparison with plain film examinations. AJR.
1987;148:1179-1185.
58. Woodring J, Lee C. Limitations of cervical radiography in the evaluation of acute
cervical trauma. J Trauma. 1993;34:32-39.
59. Vaccaro A, Hurlbert J, Patel A, et al. The Subaxial Cervical Spine Injury Classification
System. Spine. 2007;32:2365-2374.
148
60. Lammertse D, Dungan D, Dreisbach J, et al. Neuroimaging in traumatic spinal cord
injury: an evidence-based review for clinical practice and research. J Spinal Cord Med.
2007;30(3):205-214.
61. Krzyzak AT, Jasinski A, Weglarz WP, Adamek D, Sagnowskil P, Baj M. Visualisation of
the extent of damage in a rat spinal cord injury model using MR microsopy of the water
diffusion tensor. Acta Neurobiol Exp (Wars). 2005;65(3):255-264.
62. Schwartz ED, Yezierski RP, Pattany PM, Quencer RM, Weaver RG. Diffusion-weighted
MR imaging in a rat model of syringomyelia after excitotoxic spinal cord injury. AJNR
Am J Neuroradiol. Sep 1999;20(8):1422-1428.
63. Bammer R, Fazekas F. Diffusion imaging of the human spinal cord and the vertebral
column. Top Magn Reson Imaging. Dec 2003;14(6):461-476.
64. Pattany PM, Yezierski RP, Widerstrom-Noga EG, et al. Proton magnetic resonance
spectroscopy of the thalamus in patients with chronic neuropathic pain after spinal cord
injury. AJNR Am J Neuroradiol. Jun-Jul 2002;23(6):901-905.
65. Winchester P, McColl R, Querry R, et al. Changes in supraspinal activation patterns
following robotic locomotor therapy in motor-incomplete spinal cord injury.
Neurorehabil Neural Repair. Dec 2005;19(4):313-324.
66. Cadotte DW, Stroman PW, Mikulis D, Fehlings MG. A systematic review of spinal fMRI
research: outlining the elements of experimental design. J Neurosurg Spine. Sep
2012;17(1 Suppl):102-118.
67. Cadotte DW, Bosma R, Mikulis D, et al. Plasticity of the injured human spinal cord:
insights revealed by spinal cord functional MRI. PLoS One. 2012;7(9):e45560.
68. Stroman PW, Kornelsen J, Bergman A, et al. Noninvasive assessment of the injured
human spinal cord by means of functional magnetic resonance imaging. Spinal Cord. Feb
2004;42(2):59-66.
69. World Health Organization International Classification of Functioning, Disabilities and
Health (ICF). 2001. http://www.who.int/classifications/icf/en/.
70. Steeves J, Lammertse D, Curt A, et al. Guidelines for conduct of clinical trials for spinal
cord injury as developed by the ICCP panel: clinical trial outcome measures. Spinal
Cord. 2007;45:206-221.
71. Anderson K. Targeting recovery: priorities of the spinal cord injured population. J
Neurotrauma. 2004;10:1371-1383.
72. Bracken M, Collins W, Freeman D, et al. Efficacy of methylprednisolone in acute spinal
cord injury. JAMA. 1984;251:45-52.
149
73. Frankel H, Hancock D, Hyslop G, et al. The value of postural reduction in initial
management of closed injuries of the spine with paraplegia and tetraplegia. Paraplegia.
1969;7:179-192.
74. Burns A, Ditunno J. Establishing Prognosis and Maximizing Functional Outcomes after
Spinal Cord Injury. Spine. 2001;26:S137-S145.
75. Kirshblum S, Millis S, McKinley W, Tulsky D. Late neurologic recovery after traumatic
spinal cord injury. Arch Phys Med Rehabil. Nov 2004;85(11):1811-1817.
76. Van Middendorp J, Hosman A, Pouw M. ASIA Impairment scale conversion in traumatic
SCI: is it related with the ability to walk? . Spinal Cord. 2009;47:555-560.
77. Hall KM, Cohen ME, Wright J, Call M, Werner P. Characteristics of the Functional
Independence Measure in traumatic spinal cord injury. Arch Phys Med Rehabil. Nov
1999;80(11):1471-1476.
78. Hamilton B, Laughlin J, Fiedler R, Granger C. Interrater reliability of the 7-level
functional independence measure (FIM). Scand J Rehab Med. 1994;26:115-119.
79. Ottenbacher KJ, Hsu Y, Granger CV, Fiedler RC. The reliability of the functional
independence measure: a quantitative review. Arch Phys Med Rehabil. Dec
1996;77(12):1226-1232.
80. Dodds TA, Martin DP, Stolov WC, Deyo RA. A validation of the functional
independence measurement and its performance among rehabilitation inpatients. Arch
Phys Med Rehabil. May 1993;74(5):531-536.
81. Anderson K, Aito S, Atkins M, et al. Functional recovery measures for spinal cord injury:
an evidence-based review for clinical practice and research. J Spinal Cord Med.
2008;31(2):133-144.
82. Davidoff GN, Roth EJ, Haughton JS, Ardner MS. Cognitive dysfunction in spinal cord
injury patients: sensitivity of the Functional Independence Measure subscales vs
neuropsychologic assessment. Arch Phys Med Rehabil. Apr 1990;71(5):326-329.
83. Catz A, Itzkovich M, Agranov E, Ring H, Tamir A. SCIM--spinal cord independence
measure: a new disability scale for patients with spinal cord lesions. Spinal Cord. Dec
1997;35(12):850-856.
84. Catz A, Itzkovich M, Tesio L, et al. A multicenter international study on the Spinal Cord
Independence Measure, version III: Rasch psychometric validation. Spinal Cord. Apr
2007;45(4):275-291.
85. Itzkovich M, Gelernter I, Biering-Sorensen F, et al. The Spinal Cord Independence
Measure (SCIM) version III: reliability and validity in a multi-center international study.
Disabil Rehabil. Dec 30 2007;29(24):1926-1933.
150
86. Catz A, Itzkovich M, Agranov E, Ring H, Tamir A. The spinal cord independence
measure (SCIM): sensitivity to functional changes in subgroups of spinal cord lesion
patients. Spinal Cord. Feb 2001;39(2):97-100.
87. Catz A, Itzkovich M, Steinberg F, et al. The Catz-Itzkovich SCIM: a revised version of
the Spinal Cord Independence Measure. Disabil Rehabil. Apr 15 2001;23(6):263-268.
88. Yavuz N, Tezyurek M, Akyuz M. A comparison of two functional tests in quadriplegia:
the quadriplegia index of function and the functional independence measure. Spinal
Cord. Dec 1998;36(12):832-837.
89. van Hedel HJ, Dietz V. Walking during daily life can be validly and responsively
assessed in subjects with a spinal cord injury. Neurorehabil Neural Repair. Feb
2009;23(2):117-124.
90. Dittuno PL, Ditunno JF, Jr. Walking index for spinal cord injury (WISCI II): scale
revision. Spinal Cord. Dec 2001;39(12):654-656.
91. Kucukdeveci AA, Yavuzer G, Tennant A, Suldur N, Sonel B, Arasil T. Adaptation of the
modified Barthel Index for use in physical medicine and rehabilitation in Turkey. Scand J
Rehabil Med. Jun 2000;32(2):87-92.
92. Failli V, Kopp MA, Gericke C, et al. Functional neurological recovery after spinal cord
injury is impaired in patients with infections. Brain. Nov 2012;135(Pt 11):3238-3250.
93. Chesnut R, Marshall L, Klauber M, et al. The role of secondary brain injury in
determining outcome from severe head injury. J Trauma. 1993;34:216-222.
94. Ploumis A, Yadlapalli N, Fehlings MG, Kwon BK, Vaccaro AR. A systematic review of
the evidence supporting a role for vasopressor support in acute SCI. Spinal Cord. May
2010;48(5):356-362.
95. Hadley M, Walters B, Grabb P, et al. Blood pressure management after acute spinal cord
injury. Neurosurgery. 2002;50:S58-62.
96. Hadley M, Walters B, Grabb P, et al. Management of Acute Spinal Cord Injuries in an
Intestive Care Unit or Other Monitored Setting. Neurosurgery. 2002;S51-57.
97. Vale F, Burns J, Jackson A, Hadley M. Combined medical and surgical treatment after
acute spinal cord injury: Results of a prospective pilot study to assess the merits of
aggressive medical resuscitation and blood pressure management. J Neurosurg.
1997;87:239-246.
98. Tator C, Duncan E, Edmonds V, Lapczak L, Andrews D. Changes in epidemiology of
acute spinal cord injury from 1947 to 1981. Surg Neurol. 1993;40:207-215.
99. Tator C, Rowed D, Schwartz M, et al. Management of acute spinal cord injuries.
Canadian journal of surgery. Journal canadien de chirurgie. 1984;27:289-294.
151
100. Levi L, Wolf A, Belzber H. Hemodynamic parameters in patients with acute cervical
cord trauma: Description, intervention, and prediction of outcome. Neurosurgery.
1993;33:1007-1017.
101. Gschaedler R, Dollfus P, Mole J, Mole L, Loeb J. Reflections on the intesive care of
acute cervical spinal cord injuries in a general traumatology centre. Paraplegia.
1979;17:58-61.
102. Kwon BK, Mann C, Sohn HM, et al. Hypothermia for spinal cord injury. Spine J. Nov-
Dec 2008;8(6):859-874.
103. Sydenham E, Roberts I, Alderson P. Hypothermia for traumatic head injury. Cochrane
Database Syst Rev. 2009(2):CD001048.
104. Levi AD, Casella G, Green BA, et al. Clinical outcomes using modest intravascular
hypothermia after acute cervical spinal cord injury. Neurosurgery. Apr 2010;66(4):670-
677.
105. Bracken M, Shepard M, Collins W, et al. A randomized, controlled trial of
methylprednisolone or naloxone in the treatment of acute spinal-cord injury. Results of
the Second National Acute Spinal Cord Injury Study. New Engl J Med. 1990;17:1405-
1411.
106. Bracken M, Shepard M, Holford T, et al. Administration of methylprednisolone for 24 or
48 hours or tirilazad mesylate for 48 hours in the treatment of acute spinal cord injury.
Results of the Third National Acute Spinal Cord Injury Randomized Controlled Trial.
National Acute Spinal Cord Injury Study. JAMA. 1997;277:1597-1604.
107. Hugenholtz H, Cass D, Dvorak M. High dose methylprednisilone for acute closed spinal
cord injury—only a treatment option. Canadian Journal of Neurological Sciences.
2002(29):227-235.
108. Hadley M, Walters B, Grabb P, et al. Pharmacological Therapy after Acute Cervical
Spinal Cord Injury. Neurosurgery. 2002;50:563-572.
109. Bracken MB. Steroids for acute spinal cord injury. Cochrane Database Syst Rev.
2012;1:CD001046.
110. Hurlbert R. Methylprednisolone for acute spinal cord injury: an inappropriate standard of
care. J Nerosurg Spine. 2000;93:1-7.
111. Petijean M, Pointillart V, Dixmerias F, et al. Traitment medicamenteux de la lesion
medullaire traumatique au stade aigu. Annales fracaises d'anesthesie et de reanimation.
1998;17:115-122.
112. Rowland J, Hawryluk G, Kwon B, Fehlings M. Current status of acute spinal cord injury
pathophysiology and emerging therapies: promise on the horizon. Neurosurgical Focus.
2008;25(5):E2.
152
113. Fehlings MG, Wilson J, Frankowski R, et al. Riluzole for the Treatment of Acute
Traumatic Spinal Cord Injury: Rationale for and Design of the NACTN Phase I Clinical
Trial. J Neurosurgery: Spine. 2012;17:151-156.
114. Fehlings M, Theodore N, Harrop J, et al. A phase I/IIa clinical trial of a recombinant Rho
protein antagonist in acute spinal cord injury. J Neurotrauma. 2011;28:787-796.
115. Wilson J, Fehlings M. Emerging Approaches to the surgical management of acute
traumatic Spinal Cord Injury. Neurotherapeutics : the journal of the American Society for
Experimental NeuroTherapeutics. 2011;8:197-194.
116. Dolan E, Tator C, Endrenyi L. The value of decompression for acute experimental spinal
cord compression injury. J Neurosurgery. 1980;53:749-755.
117. Guha A, Tator C, Endrenyi L. Decompression of the spinal cord improves recovery after
acute experimental spinal cord compression injury. Paraplegia. 1987;25:324-339.
118. Furlan J, Noonan V, Cadotte D, Fehlings M. Timing of Decompressive Surgery of Spinal
Cord after Traumatic Spinal Cord Injury: An Evidence-Based Examination of Pre-
Clinical and Clinical Studies. J Neurotrauma. 2009;27:1-29.
119. Fehlings MG, Vaccaro A, Wilson J, et al. Early versus Delayed Decompression for
Traumatic Cervical Spinal Cord Injury: Results of the Surgical Timing in Acute Spinal
Cord Injury Study (STASCIS). PLoS One. 2012;7(2):1-8.
120. Wilson JR, Singh A, Craven C, et al. Early versus late surgery for traumatic spinal cord
injury: the results of a prospective Canadian cohort study. Spinal Cord. May 8 2012.
121. Fehlings MG, Perrin RG. The timing of surgical intervention in the treatment of spinal
cord injury: a systematic review of recent clinical evidence. Spine (Phila Pa 1976). May
15 2006;31(11 Suppl):S28-35; discussion S36.
122. Fehlings M, Rabin D, Sears W, Cadotte D, Aarabi B. Current practice in the timing of
surgical intervention in spinal cord injury. Spine. 2010;35:166-173.
123. Tetzlaff W, Okon EB, Karimi-Abdolrezaee S, et al. A systematic review of cellular
transplantation therapies for spinal cord injury. J Neurotrauma. Aug 2011;28(8):1611-
1682.
124. Wilson JR, Cadotte D, Fehlings MG. Clinical predictors of neurological outcome,
functional status, and survival after traumatic spinal cord injury: a systematic review. J
Neurosurg Spine. 2012;17(1 Sup):11-26.
125. Phillips B, Ball C, Sackett D, et al. Oxford Center for Evidence-Based Medicine:Levels
of evidence 2003; http://www.cebm.net/index.aspx?o=1025. Accessed April 20, 2011,
2011.
126. Wright JG, Swiontkowski MF, Heckman JD. Introducing levels of evidence to the
journal. J Bone Joint Surg Am. Jan 2003;85-A(1):1-3.
153
127. Waters RL, Sie I, Adkins RH, Yakura JS. Motor recovery following spinal cord injury
caused by stab wounds: a multicenter study. Paraplegia. Feb 1995;33(2):98-101.
128. Steeves J, Kramer J, Fawcett J, et al. Extent of spontaneous motor recovery after
traumatic cervical sensorimotor complete cord injury. Spinal Cord. 2011;49:257-265.
129. Sipski M, Jackson A, Gomez-Martin O, Estores I, Stein A. Effects of Gender on
Neurologic and Functional Recovery after Spinal Cord Injury. Arch Phys Med Rehab.
2004;85:1826-1836.
130. Pollard M, Apple D. Factors associated with improved neurologic outcomes in patients
with incomplete tetraplegia. Spine. 2003;28:33-39.
131. Fassett D, Harrop J, Maltenfort M, et al. Mortality rates in geriatric patients with spinal
cord injuries. J Neurosurgery Spine. 2007;7:277-281.
132. Cifu D, Seel R, Kreutzer J, McKinley W. A multicenter investigation of age-related
differences in lengths of stay, hospitalization charges, and outcomes for a matched
tetraplegic sample. Arch Phys Med Rehab. 1999;80:733-740.
133. Katoh S, Masry WE. Motor recovery of patients presenting with motor paralysis and
sensory sparing following cervical spinal cord injuries. Paraplegia. 1995;33:506-509.
134. Waters R, Yakura J, Adkins R, Sie I. Recovery Following Complete Paraplegia. Arch
Phys Med Rehab. 1992;73:784-789.
135. Waters R, Adkins R, Yakura J, Sie I. Motor and Sensory recovery following complete
tetraplegia. Arch Phys Med Rehab. 1993;74:242-247.
136. Waters R, Adkins R, Yakura J, Sie I. Motor and Sensory Recovery Following Incomplete
Paraplegia. Arch Phys Med Rehab. 1994;75:67-72.
137. Waters R, Adkins R, Yakura J, Sie I. Motor and sensory recovery following incomplete
tetraplegia. Arch Phys Med Rehab. 1994;80:733-740.
138. Waters R, Sie I, Adkins R, Yakura J. Injury pattern effect on motor recovery after
traumatic spinal cord injury. Arch Phys Med Rehab. 1995;76:440-443.
139. Zariffa J, Kramer J, Fawcett J, et al. Characterization of neurological recovery following
traumatic sensorimotor complete thoracic spinal cord injury. Spinal Cord. 2011;49:463-
471.
140. Dittuno J, Stover S, Freed M, Ahn J. Motor Recovery of the Upper Extremities in
Traumatic Quadriplegia: A Multicenter Study. Arch Phys Med Rehab. 1992;73:431-436.
141. Browne B, Jacobs S, Herbison G, Ditunno J. Pin Sensation as a Predictor of Extensor
Carpi Radialis Recovery in Spinal Cord Injury. Arch Phys Med Rehab. 1993;74:14-18.
154
142. Wu L, Marino R, Herbison G, Dittuno J. Recovery of zero-grade muscles in the zone of
partial preservation in motor complete quadriplegia. Arch Phys Med Rehab. 1992;73:40-
43.
143. Brown P, Marino R, Herbison G, Dittuno J. The 72-hour examination as a predictor of
recovery in motor complete quadriplegia. Arch Phys Med Rehab. 1991;72:546-548.
144. Ditunno J, Cohen M, Hauck W, Jackson A, Sipski M. Recovery of upper-extremity
strength in complete and incomplete tetraplegia: a multicenter study. Arch Phys Med
Rehab. 2000;81:389-393.
145. Mange K, Marino R, Gregory P, Herbison G, Dittuno J. Course of motor recovery in the
Zone of Partial Preservation in spinal cord injury. Arch Phys Med Rehab. 1992;73:437-
441.
146. Dittuno J, Sipski M, Posuniak E, Chen Y, Staas W, Herbison G. Wrist extensor recovery
in traumatic quadriplegia. Arch Phys Med Rehab. 1987;68:287-290.
147. Bravo P, Labarta C, Alcaraz M, Mendoza J, Verdu A. An assessment of factors affecting
neurologic recovery after spinal cord injury with vertebral fracture. Paraplegia.
1996;34:164-166.
148. Maynard F, Reynolds G, Fountain S, Wilmot C, Hamilton R. Neurologic prognosis after
traumatic quadriplegia. J Neurosurgery. 1979;50:611-616.
149. Coleman W, Geisler F. Injury severity as primary predictor of outcome in acute spinal
cord injury: retrospective results from a large multicenter clinical trial. Spine Journal.
2004;4:373-378.
150. Furlan J, Fehlings M. The impact of Age on Mortality, Impairment, and Disability among
Adults with Acute Traumatic Spinal Cord Injury. J Neurotrauma. 2009;26:1707-1717.
151. Burns S, Golding D, Rolle W, Graziani V, Dittuno J. Recovery of Ambulation in Motor-
Incomplete Tetraplegia. Arch Phys Med Rehab. 1997;78:1169-1172.
152. Calancie B, Molano M, Broton J. Tendon reflexes for prediction movement recovery
after acute spinal cord injury in humans. Clinical Neurophysiology. 2004;115:2350-2363.
153. Crozier K, Graziani V, Dittuno J, Herbison G. Spinal Cord Injury: Prognosis for
Ambulation Based on Sensory Examination in Patients Who are Initially Motor
Complete. Arch Phys Med Rehab. 1991;72:118-121S.
154. Ko H, Dittuno J, Graziani V, Little J. The pattern of reflex recovery during spinal shock.
Spinal Cord. 1999;37:402-409.
155. Oleson C, Burns A, Ditunno J, Geisler F, Coleman W. Prognostic Value of Pinprick
Preservation in Motor Complete, Sensory Incomplete Spinal Cord Injury. Arch Phys Med
Rehab. 2005;86:988-992.
155
156. Van Middendorp J, Hosman A, Donders A, et al. A clinical prediction rule for
ambulation outcomes after traumatic spinal cord injury: a longitudinal cohort study.
Lancet. 2011;377:1004-1010.
157. Weinstein D, Ko H, Graziani V, Dittuno J. Prognostic Significance of the Delayed
Plantar Reflex Following Spinal Cord Injury. J Spinal Cord Medicine. 1997;20:207-211.
158. Kay E, Deutsch A, Wuermser L. Predicting Walking at Discharge From Inpatient
Rehabilitation After a Traumatic Spinal Cord Injury. Arch Phys Med Rehab.
2007;88:745-750.
159. Marino R, Graves D. Metric Properties of the Asia Motor Score: Subscales Improve
Correlation with Functional Activities. Arch Phys Med Rehab. 2004;85:1804-1810.
160. Ota T, Akaboshi K, Nagata M, et al. Functional assessment of patients with spinal cord
injury:measured by the motor score and the Functional Indepedence Measure. Spinal
Cord. 1996;34:531-535.
161. McKinley WO, Johns JS, Musgrove JJ. Clinical presentations, medical complications,
and functional outcomes of individuals with gunshot wound-induced spinal cord injury.
Am J Phys Med Rehabil. Mar-Apr 1999;78(2):102-107.
162. Zorner B, Blanckenhom W, Dietz V, Group E-SS. Clinical Algorithm for Improved
Prediction of Ambulation and Patient Stratification after Incomplete Spinal Cord Injury. J
Neurotrauma. 2010;27:241-252.
163. Lazar R, Yarkony G, Ortolano D, et al. Prediction of Functional Outcome by Motor
Capacity after Spinal Cord Injury. Arch Phys Med Rehab. 1989;70:819-822.
164. Crozier K, Cheng L, Graziani V, Zorn G, Herbison G, Ditunno J. Spinal Cord Injury:
prognosis for ambulation based on quadriceps recovery. Paraplegia. 1992;30:762-767.
165. Vazquez X, Rodriguez M, Penaranda J, Concheiro L, Barus J. Determining prognosis
after spinal cord injury. Forensic and Legal Medicine. 2008;15:20-23.
166. Haisma JA, van der Woude LH, Stam HJ, et al. Complications following spinal cord
injury: occurrence and risk factors in a longitudinal study during and after inpatient
rehabilitation. J Rehabil Med. May 2007;39(5):393-398.
167. McKinley WO, Jackson AB, Cardenas DD, DeVivo MJ. Long-term medical
complications after traumatic spinal cord injury: a regional model systems analysis. Arch
Phys Med Rehabil. Nov 1999;80(11):1402-1410.
168. Aito S. Complications during the acute phase of traumatic spinal cord lesions. Spinal
Cord. Nov 2003;41(11):629-635.
169. Chen D, Apple DF, Jr., Hudson LM, Bode R. Medical complications during acute
rehabilitation following spinal cord injury--current experience of the Model Systems.
Arch Phys Med Rehabil. Nov 1999;80(11):1397-1401.
156
170. Levi R, Hultling C, Seiger A. The Stockholm Spinal Cord Injury Study: 2. Associations
between clinical patient characteristics and post-acute medical problems. Paraplegia. Oct
1995;33(10):585-594.
171. Hitzig SL, Tonack M, Campbell KA, et al. Secondary health complications in an aging
Canadian spinal cord injury sample. Am J Phys Med Rehabil. Jul 2008;87(7):545-555.
172. Vidal J, Sarrias M. An analysis of the diverse factors concerned with the development of
pressure sores in spinal cord injured patients. Paraplegia. May 1991;29(4):261-267.
173. Groah SL, Weitzenkamp D, Sett P, Soni B, Savic G. The relationship between
neurological level of injury and symptomatic cardiovascular disease risk in the aging
spinal injured. Spinal Cord. Jun 2001;39(6):310-317.
174. Werhagen L, Budh CN, Hultling C, Molander C. Neuropathic pain after traumatic spinal
cord injury--relations to gender, spinal level, completeness, and age at the time of injury.
Spinal Cord. Dec 2004;42(12):665-673.
175. Krassioukov AV, Furlan JC, Fehlings MG. Medical co-morbidities, secondary
complications, and mortality in elderly with acute spinal cord injury. J Neurotrauma. Apr
2003;20(4):391-399.
176. Furlan JC, Krassioukov AV, Fehlings MG. The effects of gender on clinical and
neurological outcomes after acute cervical spinal cord injury. J Neurotrauma. Mar
2005;22(3):368-381.
177. Meade MA, Cifu DX, Seel RT, McKinley WO, Kreutzer JS. Medical procedures,
complications, and outcomes for patients with spinal cord injury: A multicenter
investigation comparing African Americans and whites. Arch Phys Med Rehabil. Mar
2004;85(3):368-375.
178. Fawcett J, Curt A, Steeves J, et al. Guidelines for the conduct of clinical trials for spinal
cord injury as developed by the ICCP panel: spontaneous recovery after spinal cord
injury and statistical power needed for therapeutic trials. Spinal Cord. 2007;45:190-205.
179. Bondurant FJ, Cotler HB, Kulkarni MV, McArdle CB, Harris JH, Jr. Acute spinal cord
injury. A study using physical examination and magnetic resonance imaging. Spine
(Phila Pa 1976). Mar 1990;15(3):161-168.
180. Hayashi K, Yone K, Ito H, Yanase M, Sakou T. MRI findings in patients with a cervical
spinal cord injury who do not show radiographic evidence of a fracture or dislocation.
Paraplegia. Apr 1995;33(4):212-215.
181. Mascalchi M, Dal Pozzo G, Dini C, et al. Acute spinal trauma: prognostic value of MRI
appearances at 0.5 T. Clin Radiol. Aug 1993;48(2):100-108.
182. O'Beirne J, Cassidy N, Raza K, Walsh M, Stack J, Murray P. Role of magnetic resonance
imaging in the assessment of spinal injuries. Injury. Mar 1993;24(3):149-154.
157
183. Sato T, Kokubun S, Rijal KP, et al. Prognosis of cervical spinal cord injury in correlation
with magnetic resonance imaging. Paraplegia. Feb 1994;32(2):81-85.
184. Selden NR, Quint DJ, Patel N, d'Arcy HS, Papadopoulos SM. Emergency magnetic
resonance imaging of cervical spinal cord injuries: clinical correlation and prognosis.
Neurosurgery. Apr 1999;44(4):785-792; discussion 792-783.
185. Tewari M, Gifti D, Singh P. Diagnosis and prognostication of adult spinal cord injury
without radiographic abnormality using magnetic resonance imaging: analysis of 40
patients. Surg Neurol. 2005;63:204-209.
186. Ramon S, Dominguez R, Ramirez L, et al. Clinical and magnetic resonance imaging
correlation in acute spinal cord injury. Spinal Cord. Oct 1997;35(10):664-673.
187. Boldin C, Raith J, Fankhauser F, Haunschmid C, Schwantzer G, Schweighofer F.
Predicting neurologic recovery in cervical spinal cord injury with postoperative MR
imaging. Spine (Phila Pa 1976). Mar 1 2006;31(5):554-559.
188. Miyanji F, Furlan J, Aarabi B, Arnold P, Fehlings M. Acute Cervical Traumatic Spinal
Cord Injury: MR Imaging Findings Correlated with Neurologic Outcome--Prospective
Study with 100 Consecutive Patients. Radiology. 2007;243:820-827.
189. Shin JC, Kim DY, Park CI, Kim YW, Ohn SH. Neurologic recovery according to early
magnetic resonance imaging findings in traumatic cervical spinal cord injuries. Yonsei
Med J. Jun 30 2005;46(3):379-387.
190. Shimada K, Tokioka T. Sequential MR studies of cervical cord injury: correlation with
neurological damage and clinical outcome. Spinal Cord. 1999;37:410-415.
191. Takahashi M, Harada Y, Inoue H, Shimada K. Traumatic cervical cord injury at C3-4
without radiographic abnormalities: correlation of magnetic resonance findings with
clinical features and outcome. J Orthop Surg (Hong Kong). Dec 2002;10(2):129-135.
192. Shepard MJ, Bracken MB. Magnetic resonance imaging and neurological recovery in
acute spinal cord injury: observations from the National Acute Spinal Cord Injury Study
3. Spinal Cord. Dec 1999;37(12):833-837.
193. Schaefer DM, Flanders AE, Osterholm JL, Northrup BE. Prognostic significance of
magnetic resonance imaging in the acute phase of cervical spine injury. J Neurosurg. Feb
1992;76(2):218-223.
194. Flanders AE, Spettell CM, Friedman DP, Marino RJ, Herbison GJ. The relationship
between the functional abilities of patients with cervical spinal cord injury and the
severity of damage revealed by MR imaging. AJNR Am J Neuroradiol. May
1999;20(5):926-934.
195. Yamazaki T, Yanaka K, Fujita K, Kamezaki T, Uemura K, Nose T. Traumatic central
cord syndrome: analysis of factors affecting the outcome. Surg Neurol. Feb
2005;63(2):95-99; discussion 99-100.
158
196. Aarabi B, Harrop JS, Tator CH, et al. Predictors of pulmonary complications in blunt
traumatic spinal cord injury. J Neurosurg Spine. Sep 2012;17(1):38-45.
197. Ohshio I, Hatayama A, Kaneda K, Takahara M, Nagashima K. Correlation between
histopathologic features and magnetic resonance images of spinal cord lesions. Spine
(Phila Pa 1976). Jul 1993;18(9):1140-1149.
198. Berens SA, Colvin DC, Yu CG, Yezierski RP, Mareci TH. Evaluation of the pathologic
characteristics of excitotoxic spinal cord injury with MR imaging. AJNR Am J
Neuroradiol. Aug 2005;26(7):1612-1622.
199. Fukuoka M, Matsui N, Otsuka T, Murakami M, Seo Y. Magnetic resonance imaging of
experimental subacute spinal cord compression. Spine (Phila Pa 1976). Jul 15
1998;23(14):1540-1549.
200. Hackney DB, Finkelstein SD, Hand CM, Markowitz RS, Black P. Postmortem magnetic
resonance imaging of experimental spinal cord injury: magnetic resonance findings
versus in vivo functional deficit. Neurosurgery. Dec 1994;35(6):1104-1111.
201. Ohta K, Fujimura Y, Nakamura M, Watanabe M, Yato Y. Experimental study on MRI
evaluation of the course of cervical spinal cord injury. Spinal Cord. Aug 1999;37(8):580-
584.
202. Takahashi T, Suto Y, Kato S, Ohama E. Experimental acute dorsal compression of cat
spinal cord: correlation of magnetic resonance signal intensity with spinal cord evoked
potentials and morphology. Spine (Phila Pa 1976). Jan 15 1996;21(2):166-173.
203. Becerra JL, Puckett WR, Hiester ED, et al. MR-pathologic comparisons of wallerian
degeneration in spinal cord injury. AJNR Am J Neuroradiol. Jan 1995;16(1):125-133.
204. Quencer RM, Bunge RP, Egnor M, et al. Acute traumatic central cord syndrome: MRI-
pathological correlations. Neuroradiology. 1992;34(2):85-94.
205. Quencer RM, Bunge RP. The injured spinal cord: imaging, histopathologic clinical
correlates, and basic science approaches to enhancing neural function after spinal cord
injury. Spine (Phila Pa 1976). Sep 15 1996;21(18):2064-2066.
206. Greenberg SM, O'Donnell HC, Schaefer PW, Kraft E. MRI detection of new
hemorrhages: potential marker of progression in cerebral amyloid angiopathy. Neurology.
Sep 22 1999;53(5):1135-1138.
207. Furlan J, Fehlings M, Massicotte E. A quantitative and reproducible method to assess
cord compression and canal stenosis after cervical spine trauma: a study of interrater and
intrarater reliability. Spine. 2007;32:2083-2091.
208. Fehlings M, Rao S, Tator C, et al. The Optimal Radiologic Method for Assessing Spinal
Canal Compromise and Cord Compression in Patients with Cervical Spinal Cord Injury
Part 2: Results of a multicenter study. Spine. 1999;24:605-613.
159
209. Cadotte D, Wilson J, Mikulis D, Stroman P, Brady S, Fehlings M. Conventional MRI as
a diagnostic and prognostic tool in spinal cord injury. Expert Opin Med Diang. 2011;5:1-
13.
210. Grossman RG, Toups EG, Frankowski RF, Burau KD, Howley S. North American
Clinical Trials Network for the Treatment of Spinal Cord Injury: goals and progress. J
Neurosurg Spine. Sep 2012;17(1):6-10.
211. Harrell F, ed Regression Modeling Strategies with Applications to Linear Models,
Logistic Regression and Survival Analysis. New York: Springer-Verlag; 2001.
212. Steyerberg E, ed Clinical Prediction Models: A Practical Approach to Development,
Validation, and Updating. New York: Springer; 2009.
213. Greenland S, Finkle WD. A critical look at methods for handling missing covariates in
epidemiologic regression analyses. Am J Epidemiol. Dec 15 1995;142(12):1255-1264.
214. Little R, ed Statistical analysis with missing data. 2nd ed. Hoboken, NJ: Wiley; 2002.
215. Vach W, Blettner M, eds. Missing data in epidemiologic studies. New York, NY: Wiley;
1998. Encyclopedia of Biostatistics.
216. Rubin DB. Inference and Missing Data. Biometrika. 1976;63:581-592.
217. Little R. Regression with missing X's: a review. J Am Stat Assoc. 1992;88:125-134.
218. van Buuren S, Boshuizen HC, Knook DL. Multiple imputation of missing blood pressure
covariates in survival analysis. Stat Med. Mar 30 1999;18(6):681-694.
219. Schaefer J, ed Analysis of incomplete multivariate data. London: Chapman & Hall; 1997.
220. Wilson JR, Grossman RG, Frankowski RF, et al. A clinical prediction model for long-
term functional outcome after traumatic spinal cord injury based on acute clinical and
imaging factors. J Neurotrauma. Sep 2012;29(13):2263-2271.
221. Wyndaele M, Wyndaele J. Incidence, prevalence, and epidemiology of spinal cord injury:
what learns a worldwide literature survey? Spinal Cord. 2006;44:523-529.
222. Burke D, Linden R, Zhang Y, Maiste A, Shields C. Incidence rates and populations at
risk for spinal cord injury: A regional study. Spinal Cord. 2001;39:274-278.
223. Hawryluk G, Roland J, Kwon B, Fehlings M. Protection and repair of the injured spinal
cord: a review of completed, ongoing, and planned clinical trials for acute spinal cord
injury. Neurosurgical Focus. 2008;25:E14.
224. AmericanSpinalInjuryAssociation, ed International Standards for Neurological
Classification of Spinal Cord Injury. Chicago IL2000.
160
225. Liao C, Lui T, Chen L, Chuang C, Huang Y. Spinal cord injury without radiological
abnormality in preschool-aged children: correlation of magnetic resonance imaging
findings with neurological outcomes. J Neurosurgery. 2005;103:S17-23.
226. Dodds T, Martin D, Stolov W, Devo R. A validation of the functional independence
measurement and its performance among rehabilitation inpatients. Arch Phys Med Rehab.
1993;74:531-536.
227. van Middendorp J, Hosman A, Pouw M, Van de Meent H. Is determination between
complete and incomplete traumatic spinal cord injury clinically relevant? Validation of
the ASIA sacral sparing criteria in a prospective cohort of 432 patients. Spinal Cord.
2009;47:809-816.
228. Stineman M, Maislin G, Fiedler R, Granger C. A Prediction Model for Functional
Recovery in Stroke. Stroke. 1997;28:550-556.
229. Rubin DB. Expert Report on Multiple Imputation. 2006;
http://www.fda.gov/ohrms/dockets/ac/06/briefing/2006-4269b1_05_sponsor.pdf.
Accessed December 2012.
230. Efron B, Tibshirani R. Bootstrap Methods for Standard Errors, Confidence Intervals, and
Other Measures of Statistical Accuracy. Statistical Science. 1986;1:54-75.
231. Heinemann A, Kirk P, Hastie B, et al. Relationships between disability measures and
nursing effort during medical rehabilitation for patients with traumatic brain and spinal
cord injury. Arch Phys Med Rehab. 1997;78:143-149.
232. Wilson JR, Vaccaro A, Harrop JS, et al. The Impact of Facet Dislocation on Clinical
Outcomes after Cervical Spinal Cord Injury: Results of a Multicenter North American
Prospective Cohort Study. Spine (Phila Pa 1976). Aug 14 2013;2:97-103.
233. Dvorak M, Fisher C, Aarabi B, et al. Clinical Outcomes of 90 Isolated Unilateral Facet
Fractures, Subluxations, and Dislocations Treated Surgically and Nonoperatively. Spine.
2007;32:3007-3013.
234. Scher A. Unilateral locked facet in cervical spine injuries. AJR. 1977;129:45-68.
235. Hadley M, Walters B, Grabb P, et al. Treatment of Subaxial Cervical Spinal Injuries.
Neurosurgery. 2002;50:S156-165.
236. Rogers W. Fractures and dislocations of the cervical spine: An end-result study. J Bone
Joint Surg 1957;39:341-376.
237. Durbin F. Fracture-dislocations of the cervical spine. J Bone Joint Surg Br. 1957;39:23-
38.
238. Sonntag V. Management of Bilateral Locked Facets of the Cervical Spine. Neurosurgery.
1981;8:150-152.
161
239. Maiman D, Barolat G, Larson S. Management of Bilateral Locked Facets of the Cervical
Spine. Neurosurgery. 1986;18:542-547.
240. Payer M, Schmidt M. Management of traumatic bilateral locked facets of the subaxial
cervical spine. Contemporary Neurosurgery. 2005;27:1-4.
241. Andreshak J, Dekutoski M. Management of unilateral facet dislocations: A review of the
literature. Orthopedics 1997;20:917-926.
242. Allen B, Ferguson R, Lehmann T. A mechanistic classification of closed, indirect
fractures and dislocations of the lower cervical spine. Spine. 1982;7:1-27.
243. Olerud C, Jonsson H. Compression of the cervical spinal cord after reduction of fracture
dislocations: Report of 2 cases. Acta Orthop Scand. 1991;62:599-601.
244. Lee A, MacLean C, Newton D. Rapid Traction for Reduction of Cervcial Spine
Dislocations. J Bone Joint Surg Br. 1994;76:352-356.
245. Evans D. Reduction of cervical dislocations. J Bone Joint Surg Br. 1961;43:552-555.
246. Brunette D, Rockswold G. Neurologic Recovery following Rapid Spinal Realignment for
Complete Cervical Spinal Cord Injury. Journal of Trauma. 1987;27:445-447.
247. Hadley M, Walters B, Grabb P, et al. Initial Closed Reduction of Cervical Spine Fracture-
Dislocation Injuries. Neurosurgery. 2002;50:S44-50.
248. Ivanic P, Pearson A, Tominaga Y, Simpson A, Yue J, Panjabi M. Mechanism of cervical
spinal cord injury during bilateral facet dislocation. Spine. 2007;15:2467-2473.
249. Wilson JR, Arnold PM, Singh A, Kalsi-Ryan S, Fehlings MG. Clinical prediction model
for acute inpatient complications after traumatic cervical spinal cord injury: a subanalysis
from the Surgical Timing in Acute Spinal Cord Injury Study. J Neurosurg Spine. Sep
2012;17(1 Suppl):46-51.
250. McDonald J, Sadowsky C. Spinal-cord injury. Lancet. 2002;359:417-425.
251. Wilson J, Fehlings M. Adrenal Insufficiency as a Cause of Refractory Hypotension in
Acure Period after Spinal Cord Injury: A Perspective Statement. World Neurosurgery.
2011.
252. Da Paz A, Beraldo P, Almeida M, Neves E, Alves C, Khan P. Traumatic injury to the
spinal cord. Prevalence in Brazilian Hospitals. Paraplegia. 1993;30:636-640.
253. Janneke H, van der Woude L, Stam H. Complications following spinal cord injury:
occurence and risk factors in a lognitudinal study during and after inpatient rehabilitation.
Journal of Rehabilitaion Medicine. 2007;39:393-398.
254. Noreau L, Proulx P, Gagnon L, Drolet M. Secondary impairments after spinal cord
injury: a population-based study. Am J Phys Med Rehab. 2000;79:526-535.
162
255. Valtonen N, Karlson A, Alaranta H. Work participation among persons with traumatic
spinal cord injury and meningomyelocele. J Rehabil Med. 1998;38:192-200.
256. Fehlings M, Wilson J. Timing of surgical intervention in spinal trauma: what does the
evidence indicate? Spine. 2010;35:S159-S160.
257. Furlan J, Kailaya-Vasan A, Aarabi B, Fehlings M. A Novel Approach to Quantitatively
Assess Posttraumatic Cervical Spinal Canal Compromise and Spinal Cord Compression:
A Multicenter Responsiveness Study. Spine. 2011;784-793.
258. Reines H, Harris R. Pulmonary complications of acute spinal cord injuries.
Neurosurgery. 1987;21:193-196.
259. Westgren N, Levi R. Quality of life and traumatic spinal cord injury. Arch Phys Med
Rehab. 1998;79:1433-1439.
260. Meyers A, Andersen E, Hagglund K. A model of outcomes research: spinal cord injury.
Arch Phys Med Rehab. 2000;81:S81-S90.
261. Winslow C, Rozovsky J. Effect of spinal cord injury on the respiratory system. Arch Phys
Med Rehab. 2003;82(803-814).
262. Grossman RG, Frankowski RF, Burau KD, et al. Incidence and severity of acute
complications after spinal cord injury. J Neurosurg Spine. Sep 2012;17(1):119-128.
263. Meduri GU, Headley AS, Golden E, et al. Effect of prolonged methylprednisolone
therapy in unresolving acute respiratory distress syndrome: a randomized controlled trial.
JAMA. Jul 8 1998;280(2):159-165.
264. Annane D, Bellissant E, Bollaert PE, Briegel J, Keh D, Kupfer Y. Corticosteroids for
treating severe sepsis and septic shock. Cochrane Database Syst Rev.
2004(1):CD002243.
265. Giugliano GR, Giugliano RP, Gibson CM, Kuntz RE. Meta-analysis of corticosteroid
treatment in acute myocardial infarction. Am J Cardiol. May 1 2003;91(9):1055-1059.
266. AdministrationOnAge. Profile of Older Americas. 2012;
http://www.aoa.gov/aoaroot/aging_statistics/Profile/index.aspx. Accessed April 20, 2012.
267. Cifu D, Seel R, Kreutzer J, McKinley W. Age, outcome, and rehabilitation costs after
paraplegia cause by traumatic injury of the thoracic spinal cord, conus medularis, and
cauda equina. J Neurotrauma. 1999;16:805-815.
268. Cifu DX, Seel RT, Kreutzer JS, McKinley WO. A multicenter investigation of age-
related differences in lengths of stay, hospitalization charges, and outcomes for a matched
tetraplegia sample. Arch Phys Med Rehabil. Jul 1999;80(7):733-740.
163
269. Seel RT, Huang ME, Cifu DX, Kolakowsky-Hayner SA, McKinley WO. Age-related
differences in length of stays, hospitalization costs, and outcomes for an injury-matched
sample of adults with paraplegia. J Spinal Cord Med. Winter 2001;24(4):241-250.
270. James L, Brett J. Mediators, moderators, and tests for mediation. Journal of Applied
Psychology. 1984;69:307-321.
271. U.S.-Census-Bureau. Sixty-Five Plus in the United States. Statistical Brief 2011;
http://www.census.gov/population/socdemo/statbriefs/agebrief.html. Accessed March 12,
2013.
272. Baron R, Kenny D. The Moderator-Mediator Variable Distinction in Social
Psychological Research: Conceptual, Strategic and Statistical Considerations. Journal of
Personality and Social Psychology. 1986;51:1173-1182.
273. Cifu D, Seel R, Kreutzer J, Martwitz J, McKinley W, Wisory D. Age, outcome and
rehabilitation costs after tetraplegic spinal cord injury. NeuroRehabilitation.
1999;12:177-185.
274. Herbison GJ, Zerby SA, Cohen ME, Marino RJ, Ditunno JF, Jr. Motor power differences
within the first two weeks post-SCI in cervical spinal cord-injured quadriplegic subjects.
J Neurotrauma. Winter 1992;9(4):373-380.
275. McKinley W, Cifu D, Seel R, et al. Age-related outcomes in persons with spinal cord
injury: a summary paper. NeuroRehabilitation. 2003;18(1):83-90.
276. Fitch MT, Silver J. CNS injury, glial scars, and inflammation: Inhibitory extracellular
matrices and regeneration failure. Exp Neurol. Feb 2008;209(2):294-301.
277. Fehlings MG, Theodore N, Harrop J, et al. A phase I/IIa clinical trial of a recombinant
Rho protein antagonist in acute spinal cord injury. J Neurotrauma. May 2011;28(5):787-
796.
278. Bedbrook G, Sakae T. The managemet of spinal injuries-past and present. Int Rehabil
Med. 1982;2:45-61.
279. Lo TP, Jr., Cho KS, Garg MS, et al. Systemic hypothermia improves histological and
functional outcome after cervical spinal cord contusion in rats. J Comp Neurol. Jun 10
2009;514(5):433-448.
280. Braughler J, Hall E. Effects of multi-dose methylprednisolone sodium succinate
administration on injured cat spinal cord neurofilament degradation and energy
metabolism. J Neurosurg. 1984;61:290-295.
281. Baskin DS, Simpson RK, Jr., Browning JL, Dudley AW, Rothenberg F, Bogue L. The
effect of long-term high-dose naloxone infusion in experimental blunt spinal cord injury.
J Spinal Disord. Feb 1993;6(1):38-43.
164
282. Fehlings MG, Tator CH, Linden RD. The effect of nimodipine and dextran on axonal
function and blood flow following experimental spinal cord injury. J Neurosurg. Sep
1989;71(3):403-416.
283. Bose B, Osterholm J, Kalia M. Ganglioside-induced regeneration and reestablishment of
axonal continuity in spinal cord-transected rats. Neuroscience Letters. 1986;63:165-169.
284. Riluzole for amyotrophic lateral sclerosis (ALS)/motor neuron disease (MND). 2009.
285. Grossman RG, Fehlings MG, Frankowski RF, et al. A phase I multicenter trial to
investigate the safety and pharmacokinetics profile of riluzole in the treatment of
traumatic spinal cord injury. Session Number 123. Paper presented at: Society for
Neuroscience 2012; New Orleans, LA, USA.
286. Festoff BW, Ameenuddin S, Arnold PM, Wong A, Santacruz KS, Citron BA.
Minocycline neuroprotects, reduces microgliosis, and inhibits caspase protease
expression early after spinal cord injury. J Neurochem. Jun 2006;97(5):1314-1326.
287. Teng YD, Mocchetti I, Taveira-DaSilva AM, Gillis RA, Wrathall JR. Basic fibroblast
growth factor increases long-term survival of spinal motor neurons and improves
respiratory function after experimental spinal cord injury. J Neurosci. Aug 15
1999;19(16):7037-7047.
288. Spillmann AA, Bandtlow CE, Lottspeich F, Keller F, Schwab ME. Identification and
characterization of a bovine neurite growth inhibitor (bNI-220). J Biol Chem. Jul 24
1998;273(30):19283-19293.
289. Freund P, Wannier T, Schmidlin E, et al. Anti-Nogo-A antibody treatment enhances
sprouting of corticospinal axons rostral to a unilateral cervical spinal cord lesion in adult
macaque monkey. J Comp Neurol. Jun 1 2007;502(4):644-659.
290. Sahni V, Kessler JA. Stem cell therapies for spinal cord injury. Nat Rev Neurol. Jul
2010;6(7):363-372.
291. Bradbury EJ, Carter LM. Manipulating the glial scar: chondroitinase ABC as a therapy
for spinal cord injury. Brain Res Bull. Mar 10 2011;84(4-5):306-316.
292. Luo J, Borgens R, Shi R. Polyethylene glycol immediately repairs neuronal membranes
and inhibits free radical production after acute spinal cord injury. J Neurochem. Oct
2002;83(2):471-480.
293. Das M, Patil S, Bhargava N, et al. Auto-catalytic ceria nanoparticles offer
neuroprotection to adult rat spinal cord neurons. Biomaterials. Apr 2007;28(10):1918-
1925.
294. Harkema S, Gerasimenko Y, Hodes J, et al. Effect of epidural stimulation of the
lumbosacral spinal cord on voluntary movement, standing, and assisted stepping after
motor complete paraplegia: a case study. Lancet. Jun 4 2011;377(9781):1938-1947.
165
295. Otani K, Abe H, Kadoya S. Beneficial effects of methylprednisolone sodium succinate in
the treatment of acute spinal cord injury. Sekitsy Sekizui 1994;7:633-647.
296. Geffner LF, Santacruz P, Izurieta M, et al. Administration of autologous bone marrow
stem cells into spinal cord injury patients via multiple routes is safe and improves their
quality of life: comprehensive case studies. Cell Transplant. 2008;17(12):1277-1293.
297. Yoon SH, Shim YS, Park YH, et al. Complete spinal cord injury treatment using
autologous bone marrow cell transplantation and bone marrow stimulation with
granulocyte macrophage-colony stimulating factor: Phase I/II clinical trial. Stem Cells.
Aug 2007;25(8):2066-2073.
298. Sykova E, Homola A, Mazanec R, et al. Autologous bone marrow transplantation in
patients with subacute and chronic spinal cord injury. Cell Transplant. 2006;15(8-9):675-
687.
299. Deda H, Inci MC, Kurekci AE, et al. Treatment of chronic spinal cord injured patients
with autologous bone marrow-derived hematopoietic stem cell transplantation: 1-year
follow-up. Cytotherapy. 2008;10(6):565-574.
300. Mackay-Sim A, Feron F, Cochrane J, et al. Autologous olfactory ensheathing cell
transplantation in human paraplegia: a 3-year clinical trial. Brain. Sep 2008;131(Pt
9):2376-2386.
301. Lima C, Escada P, Pratas-Vital J, et al. Olfactory mucosal autografts and rehabilitation
for chronic traumatic spinal cord injury. Neurorehabil Neural Repair. Jan 2010;24(1):10-
22.
302. Saberi H, Moshayedi P, Aghayan HR, et al. Treatment of chronic thoracic spinal cord
injury patients with autologous Schwann cell transplantation: an interim report on safety
considerations and possible outcomes. Neurosci Lett. Sep 26 2008;443(1):46-50.
303. Knoller N, Auerbach G, Fulga V, et al. Clinical experience using incubated autologous
macrophages as a treatment for complete spinal cord injury: phase I study results. J
Neurosurg Spine. Sep 2005;3(3):173-181.
304. Lammertse DP, Jones LA, Charlifue SB, et al. Autologous incubated macrophage therapy
in acute, complete spinal cord injury: results of the phase 2 randomized controlled
multicenter trial. Spinal Cord. Sep 2012;50(9):661-671.
166
Appendices
Appendix 1: International Standards for the Neurological Classification of Spinal Cord
Injury (American Spinal Injury Association: International Standards for Neurological
Classification of Spinal Cord Injury, revised 2011; Atlanta, GA. Reprinted 2011)
167
168
Appendix 2: Emerging Therapies for Acute Traumatic Spinal Cord Injury
Modified from:
Wilson JR, Forgione N, Fehlings MG. Emerging Therapies for Acute Traumatic Spinal Cord
Injury. CMAJ, Epub Dec 10, 2012.
Abstract
At the patient level, traumatic spinal cord injury remains one of the most devastating conditions
encountered in clinical medicine. When considering losses in productivity and the costs of life
long care for affected individuals, the socioeconomic footprint of spinal cord injury is also
significant. Although a therapeutic breakthrough has yet to emerge, a recent surge in preclinical
spinal cord injury related research has led to the identification of several promising
neuroprotective and neuroregenerative therapies, which are currently in various stages of clinical
development. The current review provides general, as well as specialty physicians, with an
update on the current treatment evidence relating to the medical, surgical and cellular based
treatment of acute traumatic spinal cord injury.
Key Points:
1) Hemodynamic support (mean arterial pressure 85-90mmHg) and monitoring in an
intensive care unit setting for the first week after injury are recommended.
2) Decompressive surgery performed before 24 hours after injury has been shown safe,
feasible and, in prospective non-randomized trials, to be associated with improved rates of
neurological recovery.
3) Acute administration of intravenous methylprednisolone is not a standard of care but
rather a neuroprotective option that may be associated with an increased risk of
complications.
4) The safety of cellular transplantation in human spinal cord injury is under study in early
phase clinical trials and, at present, is purely an investigational therapy.
5) An improved understanding of pathophysiology has facilitated early phase clinical trials
with potential regenerative approaches involving Rho and Nogo inhibitors and
investigational neuroprotective therapies including riluzole, minocycline and hypothermia.
169
Keywords: Spinal Cord Injury; Neurotrauma; Drug Therapy; Spine Surgery; Cellular
Transplantation;
Introduction
At present, approximately 85,000 Canadians live with a diagnosis of spinal cord injury
and of these, more than half are secondary to trauma5. It is anticipated that as the population
continues to age, the annual incidence, and overall prevalence, of traumatic spinal cord injury
will continue to rise primarily as a result of increased fall related injuries amongst the elderly6.
Given this information, it is clear that the treatment of these injuries is relevant not only to spine
surgeons and physiatrists, but also to primary care physicians who may treat and counsel spinal
cord injury patients in the emergency department or family practice setting on a day to day basis.
In the current review, we present relevant pathophysiology and recent evidence pertaining to the
medical, surgical and cellular-based treatment of acute traumatic spinal cord injury. Most of the
identified pharmacologic studies were randomized trials or early phase non-randomized
prospective studies. Studies relating to the remaining topics were predominately observational in
design(Box1).
What are the key mechanisms that underlie neural injury and neural repair after spinal
cord injury?
The initial spinal cord trauma, or primary injury, initiates a sequence of pathological
events, collectively referred to as secondary injury. These secondary mechanisms begin within
seconds of the primary injury and continue for several weeks thereafter, leading to an expanded
region of tissue destruction. Initial disruption of the spinal cord vasculature leads to the
development of grey and white matter micro-hemorrhages, interstitial edema and release of
coagulation factors as well as vaso-active amines37
. These events promote thrombosis and
vasospasm of the cord microvasculature causing tissue hypoxia and impaired neuronal
homeostasis. At the cellular level, impairments include neuronal ionic imbalance, membrane
lipid peroxidation, free radical formation and release of toxic levels of the excitatory
neurotransmitter glutamate44
. Neuroprotective agents discussed below act to mitigate secondary
injury elements to reduce the extent of neural injury.
170
The regenerative capacity of Central Nervous System (CNS) neurons is severely limited
compared to neurons in the Peripheral Nervous System due largely to the production of
inhibitory molecules that thwart axonal growth, preventing regeneration of injured nerve tracts.
Nogo is a family of inhibitory proteins that bind to the Nogo receptor found on regenerating
axons276
. This binding leads to the activation of the Rho pathway, causing inhibition of axonal
growth and neuronal cytoskeletal development277
. As opposed to neuroprotective therapies that
limit the extent of acute neural injury, neuroregenerative therapies facilitate neuronal regrowth
by several mechanisms, including the blockade of these inhibitory pathways.
What supportive and surgical management is effective in spinal cord injury?
Historically, it was commonplace for injured patients to be placed in unmonitored
hospital ward beds for prolonged periods while elements of the bony injury healed. However this
approach has been supplanted by aggressive medical and surgical methods focused on
maintaining cord perfusion, avoiding complications, decompressing the spinal cord and
restoring spinal stability115
.
Medical Support
The negative consequences of hypotension on the injured CNS are well established93
.
There is consistent evidence that avoiding hypotension and maintaining aggressive blood
pressure targets during the acute phases post injury improves neurological recovery and reduces
mortality100
. Based on the existing, largely retrospective data, the American Association of
Neurological Surgeons recommends that patients’ mean arterial pressure be maintained at 85-
90mmHg for the first 7 days after injury95
. When volume replacement is inadequate to achieve
this goal, intravenous vasopressor medications may be introduced. In addition to blood pressure
maintenance, it is also recommend that patients, particularly those with severe cervical injuries,
be treated in an intensive care unit(ICU) setting with continuous cardiac, hemodynamic and
respiratory monitoring for the first 7-14 days post injury. The standardized admission of injured
patients to an ICU , has been associated in observational studies with reduced mortality and
morbidity, as well as improved neurological recovery96
.
Surgical Decompression
171
The preclinical literature provides a strong biological imperative to decompress the spinal
cord early after injury118
. In spite of compelling laboratory findings, for many years, surgeons
were reluctant to operate acutely, due to concerns that perioperative hemodynamic changes
would compromise cord perfusion 278
. Also, until the proliferation of instrumented spinal fixation
techniques in the 1990’s, decompression meant further destabilizing an already unstable spine.
In 2002, the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS), was
initiated to investigate the efficacy and safety of performing acute decompressive surgery 119
. In
this prospective, multi-center non-randomized cohort study, 313 patients were enrolled and
underwent either early (<24 hours after injury) or late (≥24 hours after injury) surgical
decompression. Early surgery was found to be safe with a trend towards reduced incidence of in-
hospital complications in the early (24.2%) versus the late (30.5%) surgical group (p=0.21). At
6-months follow-up, early surgery was associated with higher odds of neurological recovery as
defined by a 2 American Spinal Injury Association Impairment Scale grade improvement
(OR=2.57,95%CI:1.11,5.97). Although previous retrospective studies provide conflicting results
surrounding the impact of early surgery on neurological recovery, all studies on the topic have
shown early decompression to be safe121
. Moreover, these results validate the findings of recent
consensus surveys indicating surgeon preference towards early decompression122
.
In the specific case of cervical spine dislocation associated with spinal cord injury, rapid
closed reduction of the spine, using skeletal traction, remains a valid treatment option to achieve
spinal cord decompression. In this case, surgery is performed after closed reduction to re-
establish spinal stability.
Therapeutic Hypothermia
The rationale for therapeutic hypothermia after spinal cord injury emanates from
preclinical studies suggesting that cooling mitigates secondary injury mechanisms279
. Initial
clinical studies involving direct cooling of the cord during surgery failed to demonstratebenefit.
However, in a recent phase I trial, researchers at the University of Miami investigated the acute
use of modest (33ºC) systemic intravascular hypothermia 104
. In this study, there was no
difference in the complication profile observed in the cooled 14 patients and the 14 matched
control patients. As regards neurological recovery, 6 out 14 patients (42.9%) converted from
complete (no motor or sensory function below the level of the injury) to incomplete (motor or
172
sensory function present below the level of injury) status at 1-year follow-up which compares
favorably with the historical literature reported rate of 20%. While at present there is insufficient
evidence to support the use of systemic hypothermia, a multicenter efficacy trial exploring this
therapy is currently being planned
What Pharmacological Therapies have been evaluated for the treatment of spinal cord
injury?
To date, a total of five drug therapies have undergone evaluation in the context of phase
III trials (Table1). While controversy continues to loom over the use of methylprednisolone,
none of these agents have become standard of care.
Neuroprotective Agents
Methylprednisolone Sodium Succinate
Methylprednisolone, which attenuates membrane lipid peroxidation and post-traumatic
inflammation, has consistently been associated with improved neurobehavioral outcomes in
preclinical studies280
. However, the use of methylprednisolone in the clinical setting remains
controversial. In the most recent Cochrane review, the results of 3 studies were combined in
meta-analysis format109
. In the overall analysis, no treatment effect was observed for
methylprednisolone (24-hour regimen) with respect to motor recovery at 6-months follow-up
(Weighted Mean Difference=0.85, 95%CI:-1.79,3.49). However, when commenced within 8-
hours of injury, methylprednisolone administration was associated with an additional 4 points in
motor recovery (Weighted Mean Difference=4.06,95%CI:0.58,7.55). Trends were found
towards increased rates of GI hemorrhage (RR=2.18,95%CI:0.80,5.93) and wound infections
(RR=2.11,95%CI: 0.81,5.49) in methylprednisolone treated patients. Critics of
methylprednisolone cite these complication trends, as well as the use of subgroup analyses to
prove effect, as arguments against its use110
. Balancing the available evidence, consensus
guidelines recommend that 24-hour infusion of methylprednisolone, started within 8-hours of
injury, is a treatment option, that should only be undertaken with knowledge of potential for
treatment related complications107
.
173
Naloxone
The opioid antagonist naloxone demonstrated efficacy as a neuroprotectant in animal
studies by blocking the neurotoxic effects of the endogenous opioid Dynorphin A281
. In the
second National Acute Spinal Cord Injury Study, although overall group values were not
quantified, no difference in motor recovery was observed between the naloxone andplacebo
treated groups at 6-months follow-up105
.
Tirilazad-Mesylate
In the third National Acute Spinal Cord Injury Study, the efficacy of another putative
neuroprotective agent Tirilazad, a non-glucocorticoid 21-aminosteroid, developed to inhibit
neuronal membrane peroxidation, was evaluated in comparison to 24-hour methylprednisolone
infusion106
. At 6-months follow-up, there was no difference between groups with respect to
motor recovery (overall value for Tirilazad group not reported). There has been no large
systematic comparison of Tirilazad to placebo performed to date.
Nimodipine
Nimodipine, a calcium channel blocker, demonstrated efficacy in animal studies by
blocking the calcium-dependent activation of destructive cellular enzymes, and by blocking pre-
synaptic glutamate release282
. However, in a phase III trial evaluating this drug, there was no
difference in motor recovery between patients receiving placebo and those receiving Nimodipine
at 1-year follow-up.
Neuroregenerative Agents
GM-1(Sygen)
Gangliosides are complex glycolipid molecules that comprise an important structural
component of neuronal membranes. Laboratory studies have demonstrated that the
administration of gangliosides can enhance axonal regeneration after injury283
. In addition, a
variety of neuroprotective effects have been attributed to these compounds. In a randomized trial
involving 760 patients, ganglioside compound GM-1(Sygen) was compared to placebo 24
. At 6-
174
months follow-up, there was no difference between these groups in the proportion of patients
achieving marked neurologic recovery (overall quantitative group comparison not reported).
What Pharmacologic Therapies are in development for the treatment of spinal cord
injury?
Several neuroprotective and neuroregenerative agents, targeting specific pathological
mechanisms are now in the midst of clinical translation. While promising, these agents have yet
to demonstrate efficacy in phase III trials.
Neuroprotective Agents
Riluzole
Riluzole is a sodium channel blocker approved by the FDA and Health Canada for the
treatment of Amyotrophic Lateral Sclerosis (ALS). When started at diagnosis, riluzole reduces
motor neuron degeneration and prolongs survival for ALS patients 284
. In preclinical spinal cord
injury models, riluzole has shown to mitigate secondary injury by blocking pathological sodium
channel activation and reducing neuronal glutamate release44
. A phase I/II trial evaluating the
safety and pharmacokinetics of riluzole in human injury was initiated in 2010 and completed in
January 2012, with a preliminary analysis of results recently presented 113,285
.
Minocycline
Minocycline, a chemically modified form of tetracycline, also holds promise as a
neuroprotectant, although the exact mechanisms underlying its efficacy in animal injury models
remain incompletely understood 286
. In light of its preclinical efficacy, and favorable human
safety profile gleaned from its use in other clinical conditions such as acne, minocycline was
brought to trial by investigators at the University of Calgary. In this randomized placebo-
controlled phase II trial, minocycline administration was associated with a trend towards
improved motor recovery, with treated patients experiencing 6 additional points of American
Spinal Injury Association motor score recovery at 1-year follow-up(p=0.20 95% CI:-3,14).
Minocycline administration was also found to be safe, with a single case of transiently elevated
serum transaminases as the only drug related complication.
175
Basic Fibroblast Growth Factor
Injection of this factor has been shown to improve functional and respiratory parameters
in animal injury models, through a presumed mechanism of reducing glutamate mediated
excitotoxicity287
. A recombinant version of this molecule is the subject of a phase I/II trial
currently recruiting patients.
Neuroregenerative Agents
Cethrin
BA-210 is a bacterial derived toxin that inhibits the Rho pathway of inhibitory proteins
and has shown to promote axonal growth in-vitro. When combined with a biohemostatic
adhesive, BA-210 forms a permeable paste called Cethrin that is applied to the spinal cord dura
post injury. Based on documentation of preclinical efficacy, a phase I/IIa trial was undertaken
during which one of two Cethrin dosages was applied to the dura at surgery in 48 patients with
complete injuries114
. At 1-year no serious complications were attributed to Cethrin. Further, in
the 1mg and 3mg dosage groups, cervical injured patients experienced mean American Spinal
Injury Association motor score improvements of 27 and 21 points respectively, which compares
favorably to 10 points of motor recovery observed for similar patients in historical series.
Anti-Nogo
Nogo-A is a protein that has been shown to block axonal growth in the human CNS288
.
Anti-Nogo is a monoclonal antibody engineered to target Nogo-A and promote neural
regeneration 289
. This drug is currently in the midst of early stage clinical investigation.
What is the current status of Cellular Transplantation as a treatment for spinal cord
injury?
The transplantation of stem cells and autologous non-stem cells has been intensively
studied in preclinical injury models. A variety of cellular subtypes have been utilized for this
purpose with each theorized to act through one, or several, of three key mechanisms: 1) release
of growth promoting trophic factors, 2) environmental modification (i.e. reduction of scar or
inflammation) and 3) cellular replacement. The exploration of numerous cellular subtypes is a
176
reflection of the desire to implant the cell with the optimal balance of these functions. In
preclinical studies, cellular transplantation, either alone or in combination with other therapies,
has been associated with enhanced neurobehavioural recovery, with no one cellular subtype
demonstrating superiority290
. In the clinical realm, although prematurely used to treat injured
patients at institutions in several countries throughout the world, no study has established
efficacy for the transplantation of any cellular line. However, in the existing early phase trials,
major adverse events related to transplantation have been rare. Table 2 describes previous and
existing clinical studies involving transplantation of the relevant cellular subtypes. When
interpreting the results of these studies it is important to consider that independent of treatment,
most patients will experience some natural neurologic recovery which plateaus at 4-6 month
after injury. This is relevant since non-controlled studies in which patients are transplanted
before this plateau may report improvement in neurological function after treatment. In this
setting it is impossible to discern whether improvements are related to treatment or simply to
individuals’ natural recovery potential and hence results must be interpreted with caution.
Overall, cellular transplantation is purely an investigation therapy, which at present, should only
be undertaken in the context of clinical trials.
Future Directions
A substantial volume of recent laboratory work has identified a variety of promising
therapies, yet to appear on the clinical landscape. Chondroitinase ABC is a bacterial derived
enzyme which has shown beneficial effects in rodent injury models by degrading elements of the
glial scar that prevents post-traumatic axonal growth291
. Another compound, Polyethylene
glycol, a hydrophilic polymer, when combined with magnesium, has demonstrated
neuroprotective properties in animal models, by preserving neuronal membrane integrity292
. Both
of these treatments appear poised for eventual translation to the clinic. More recently, several
groups have begun to investigate the role of nanomedicine in promoting neuroprotection and
neuroregenerative after injury293
. Cerum-oxide and gold nano-particles have demonstrated
positive results in both in-vitro and in-vivo laboratory studies. Lastly, apart from pharmacologic
therapies, researchers are in the early phases of investigating neuromodulatory approaches, such
as epidural spinal cord stimulation, to aid rehabilitation efforts during the chronic phases after
injury294
. Although only described in case report form, such approaches appear promising, and in
177
the future may augment benefits contributed by acute therapeutics to maximize patients’ long-
term potential for recovery.
Box 1. Evidence used in this review:
We performed a comprehensive computerized literature search using MEDLINE with the key
word “spinal cord injury” and the subheading “treatment”. The literature search was limited to
clinical articles in English language journals, published between 1980 and 2012. Animal studies,
review articles and case reports were excluded. The search strategy was supplemented by
searching the Cochrane Database of Systematic Reviews with the search term “spinal cord
injury”. Through this process a total of 401 abstract were obtained and reviewed by the first
author for relevance to this topic. Of these abstracts, 45 full text articles were obtained and used
as the basis for this review. In addition, we also reviewed the 2002 American Association of
Neurological Surgeons/Congress of Neurological Surgeons cervical spinal cord injury consensus
guidelines as well as the Consortium for Spinal Cord Medicine 2008 spinal cord injury early
acute management clinical practice guidelines.
Box 2. How to use the results of this review in practice:
A 25 year old unrestrained male driver presents to the emergency department 30 minutes after a
high speed motor vehicle accident. His mean arterial pressure is 65mmHg. Neurological
examination reveals lack of motor or sensory function below the 5th
cervical neurologic level.
Subsequent CT scan and MRI reveal a fracture/dislocation of the cervical spine causing
compression of the spinal cord. No other major injuries are identified.
Immediate attention should be placed on optimizing oxygenation and hemodynamic status as
well as on ensuring immobilization of the craniospinal axis. A mean arterial pressure of 85-90
mmHg should be achieved and maintained for the first week following injury using volume
resuscitation augmented with vasopressors as needed. Consultation with the spine surgery
service should be obtained as early as possible to evaluate for suitably of immediate
decompressive/stabilization surgery. Although the patient presents within 8 hours of injury,
administration of methylprednisolone is not standard therapy, but rather is a treatment option that
may lead to increased complications. This patient should be admitted to an intensive care unit
with continuous cardiac, hemodynamic and respiratory monitoring for the first 1-2 weeks after
injury. Although a variety of additional treatments including pharmacologic agents, systemic
hypothermia and cellular transplantation appear promising in early phase trials, none are
recommended for routine clinical use at present.
178
Table 1: Acute pharmacologic agents previously studied in phase III efficacy trials for the
treatment of traumatic spinal cord injury Drug Purported Drug Mechanism Number of RCTs
evaluating drug
Evidence for use
Methylpredni-
solone Sodium
Succinate
(MPSS)
Neuroprotective Agent:
-Attenuates neuronal membrane
peroxidation
-reduced TNF-a release
-improved spinal cord perfusion
-reduced neuronal calcium
influx
-3 (evaluating high
dose 24hour MPSS
vs. placebo)105,111,295
-1 (evaluating high
dose 48hour MPSS
vs high dose 24hour
MPSS)106
- High dose 24hour MPSS infusion:*
- No difference in NASCIS motor
score recovery in overall analysis as
compared to placebo (WMD =0.85[-
1.79,3.49])
- Improved NASCIS motor score
recovery in subgroup receiving
treatment within 8 hours of injury as
compared to placebo (WMD
=4.06[0.58,7.55])
-Trend towards increased wound
infection rates (RR:2.11[0.81,5.49])
and GI bleeding (RR:2.18[0.80,5.93])
in steroid group
- High dose 48hour MPSS infusion:
-No difference in NASCIS motor
score recovery in overall analysis as
compared to 24 hour MPSS
(MD=3.37[-0.54,7.28])
- Trends towards increased rates of
severe sepsis (RR:4.0[0.45,35.38]) and
pneumonia (RR:2.25[0.71,7.15]) as
compared to 24hour MPSS
Naloxone Neuroprotective Agent:
-Blocks the neurotoxic effects
of the endogenous opioid
Dynorphin A
-1 (Naloxone vs.
placebo)105
- No difference in NASCIS motor
score recovery between treatment and
placebo group
Nimodipine Neuroprotective Agent:
- L-type calcium channel
blocker
-Prevents activation of calcium
dependent apoptotic enzymes
and blocks presynaptic
glutamate release
-1 (Nimodipine vs.
placebo)111
- No difference in motor neurologic
status between treatment and placebo
groups with ASIA Motor Scores of
67[50-95] and 72[50-94] at 1-year
respectively.
Tirilazad
Mesylate (TM)
Neuroprotective Agent:
-Attenuates neuronal membrane
peroxidation
-1 (TM vs. 24hour
MPSS)106
- No difference in NASCIS motor
score recovery between TM and
24hour MPSS group
-no placebo controlled evaluation
available
GM-1
Ganglioside
(Sygen)
Neuroregenerative Agent:
- Important component of CNS
neuronal membranes
-Facilitates axonal regrowth and
regeneration
-Several neuroprotective
properties also attributed to this
compound
-1 (GM-1 vs.
placebo)24
- No difference in marked neurological
recovery between treatment and
placebo groups as defined by at least a
2 grade improvement in modified
Benzel scale grade
WMD: Weighted Mean Difference MD: Mean Difference
NASCIS: National Acute Spinal Cord Injury Study
ASIA: American Spinal Injury Association
Modified Benzel Sale: Ordinal Scale between 1 and 7 with higher number reflecting superior neurologic status
*Reflects the results of meta-analysis combining results of three studies
179
Table 2: Summary of Cellular Sub-Types and Published or Ongoing Clinical Research Studies
Cell Type Clinical studies published or
underway
Patient Population and
time of Cell
Administration
Study Findings
Bone marrow
derived stem
cells (BMC)
1)Geffner 2008296
(Autologous BMC given via
intra-SC injection, IV or
topically)
2)Yoon 2007297
(Autologous BMC with
GM-CSF injected in SC at
lesion site)
3)Sykova 2006298
(Autologous BMCs given IA
or IV)
4) Deda 2008299
(Autologous BMCs intra-
SC)
1) 8 SCI patients: 4 acute
(5days-6months and 4
chronic (5-21yrs).
2) 35 SCI patients: 17
acute(<2weeks), 6
subacute (2-8weeks) and
12 chronic(>8weeks).
3) 20 SCI patients: 7
acute(10-30days) and 13
chronic(2-17mos)
4) 9 complete SCI
patients with chronic SCI
1) No major adverse events.
Improvements in QoL and
bladder function reported as
measured by Barthel Scale and
study specific bladder function
index respectively.
2) No major adverse events in
patients implanted. At least 1 AIS
grade improvement in 30.4% of
patients as compared to 0% in
historical cohort.
3)No major adverse events. 5/7
acute patients and 1/13 chronic
patients experienced neurological
improvement
4) All patients improved at least 1
AIS grade. No complications
reported.
Olfactory
Ensheathing
cells (OEC)
1)Mackay-Sim 2008300
(Autologous OEC injected
intra-SC)
2) Lima 2010301
(Olfactory mucosa
containing OECs injected
intra-SC)
1)6 complete thoracic
chronic (>6mos) SCI
patients
2) 20 complete chronic
(18-89mos) SCI patients
1) No adverse events. No
functional improvement. Sensory
improvement in 1 patient.
2) One case of aseptic meningitis.
At least 1 AIS grade improvement
in 55% of patients.
Schwann Cells
(ShC)
1)Saberi 2008302
(Autologous ShCs obtained
from sural nerve injected
intra-SC)
1)Interim safety report of
4/33 with chronic (28-
80mos) thoracic SCI
1) No major adverse events.
Transient paresthesia and muscle
spasms noted in all 4 patients.
Activated
Autologous
Macrophages
(AAM)
1)Knoller 2005303
(AAM injected intra-SC
immediately caudal to
lesion)
2)Lammertse 2012304
(AAM injected intra-SC
immediately caudal to
lesion)
1)8 complete acute
(<14days) SCI patients
2)43 patients (26
treatment, 17 control)
complete acute
(<14days) SCI patients
1) No major adverse events
related to implantation. At least 2
AIS grade improvement in 3
patients (38%)
2) At least 1 AIS grade
improvement in 7 active treatment
patients (27%) and 10 control
patients (59%). One case of post-
op spinal instability and one case
of post-op atelectasis attributed to
treatment.
Human
Embryonic
stem-cells
(hESC)
1)Geron Corp, Menlo Park,
CA
(hESC OPCs injected intra-
SC)
1)Target enrollment of 8
complete acute (7-
14days) thoracic SCI
patients
1)Trial halted prior to completion
after implantation of 4 patients.
Tissue derived
adult neural
stem cells
1)Curt (ongoing)
(allogeneic HuCNS-SC ®
injected intra-SC)
1)Target enrollment of
12 thoracic chronic
(>6weeks) SCI patients
1) Currently enrolling, no data
reported.
Intra-SC: Intra-Spinal Cord
GM-CSF: Granulocyte Macrophage Colony Stimulating Factor
IA: Intra-arterial IV: Intravenous
OPC: Oligodendroglial Precursor Cell
180
HuCNS-SC: Human CNS Stem Cells (Type of Adult neural stem cell registered to Stem Cell Inc.)
QoL: Quality of Life
AIS: ASIA (American Spinal Injury Association) Injury Scale: An ordinal scale with 5 levels ranging between grade
‘A’ (the most severe) and ‘E’ (perfect neurological status).
181
Appendix 3: Determination of Level of Evidence for papers included in Systematic Review
Author
year
Study Design Patients at similar
point in course of
disease or treatment
Patients followed
long enough for
outcomes to occur
Complete
follow-up
of > 85%
Controlling for
prognostic
factors
Evidence
class
Prospective
cohort
Retrospective
cohort
Case-
control
Case-
series
Aarabi 2012 + + + + + I
Aito 2003 + + + + - III
Boldin 2006 + _ + + - II
Bondurant 1990 + _ + + _ II
Bravo 1996 + + - + + III
Brown 1991 + + - III
Browne 1993 + + + + - II
Burns 1997 + + - - - III
Calancie 2004 + + + - + II
Chen 1999 + + + + - III
Cifu 1999 + + - - + III
182
Author
year
Study Design Patients at similar
point in course of
disease or treatment
Patients followed
long enough for
outcomes to occur
Complete
follow-up
of > 85%
Controlling for
prognostic
factors
Evidence
class
Prospective
cohort
Retrospective
cohort
Case-
control
Case-
series
Coleman 2004 + + - - III
Crozier 1992 + + + - III
DeVivo 1990 + + + + + II
Ditunno 1987 + + - - III
Ditunno 1992 + + + - III
Ditunno 2000 + + + - II
Dryden 2004 + - + + - III
Fasset 2007 + + + - III
Falnders 1999 + + + + _ III
Furlan 2005 + + + + + II
Furlan 2009 + + + - II
Groah 2001 + + + + + II
Haisma 2007 + + + - + II
183
Author
year
Study Design Patients at similar
point in course of
disease or treatment
Patients followed
long enough for
outcomes to occur
Complete
follow-up
of > 85%
Controlling for
prognostic
factors
Evidence
class
Prospective
cohort
Retrospective
cohort
Case-
control
Case-
series
Hayashi 1995 + - + + - III
Hitzig 2008 + + + + + II
Katoh 1995 + + + + III
Kay 2007 + + - - III
Ko 1999 + + - + III
Krassioukov 2003 + + + + - III
Lazar 1989 + + - - III
Levi 1995 + _ + + + III
Mange 1992 + + - - III
Marino 1999 + + + - III
Marino 2004 + + - - III
Mascalchi 1993 + + + + - II
Maynard 1979 + + + + III
184
Author
year
Study Design Patients at similar
point in course of
disease or treatment
Patients followed
long enough for
outcomes to occur
Complete
follow-up
of > 85%
Controlling for
prognostic
factors
Evidence
class
Prospective
cohort
Retrospective
cohort
Case-
control
Case-
series
McKinley 1999 + + + + III
Meade 2004 + + + + + II
Miyanji 2007 + - _ + + II
O’Beirne 1993 + _ + + _ III
Oleson 2005 + + + - III
Ota 1996 + + - - III
Pollard 2003 + + + - III
Ramon 1997 + - - + - III
Sato 1994 + + + + - III
Seldin 1999 + + + + + II
Shaefer 1992 + + + + - III
Shepard 1999 + + _ + + II
Shimada 1999 + + + + - II
185
Author
year
Study Design Patients at similar
point in course of
disease or treatment
Patients followed
long enough for
outcomes to occur
Complete
follow-up
of > 85%
Controlling for
prognostic
factors
Evidence
class
Prospective
cohort
Retrospective
cohort
Case-
control
Case-
series
Shin 2005 + + + + - II
Sipski 2004 + + + - III
Steeves 2011 + + + - III
Takahashi 2002 + + + + + II
Tewari 2005 + + + + _ III
van
Middendorp
2009 + + + -
III
van
Middendorp
2011 + + + -
III
Vazquez 2008 + + + + III
Vidal 1995 + - + + + III
Waters 1992 + + + + II
Waters 1993 + + + + II
Waters 1994a + + + + II
186
Author
year
Study Design Patients at similar
point in course of
disease or treatment
Patients followed
long enough for
outcomes to occur
Complete
follow-up
of > 85%
Controlling for
prognostic
factors
Evidence
class
Prospective
cohort
Retrospective
cohort
Case-
control
Case-
series
Waters 1994b + + + + II
Waters 1995 + + + + II
Waters 1995b + + + + II
Weinstein 1997 + + - + III
Werhagen + - + + + III
Wu 1992 + + + - III
Yamazaki 2005 + + + + - III
Zariffa 2011 + + - - III
Zorner 2010 + + - + III
Level 1: Prospective Cohort Study; Level 2: Retrospective study; Level 3: Case Control; Level 4: Case Series; Level 5: Expert
Opinion; (Evidence maybe downgraded depending on patient follow-up, differences in disease time-course and adjustment for
confounders).
187
Appendix 4: Summary of Data in Combined Dataset
Dataset Gender Age Etiology Energy Admis. ASIA exam
Admis GCS
MRI IM signal charac.
F/U ASIA (6or12 mos)
F/u FIM (6or12 mos)
Ster. Adm.
Surg. Time
Surg. Details
STASCIS X X X X X X X X X X X X Data present /426
426 418 423 423 426 415 58 303 238 423 426 426
NACTN X X X X X X X X X X X X Data present /303
303 303 291 291 303 296 254 155 158 285 248 248
Data present In C/H Dataset /729
729 100%
721 99%
714 98%
714 98%
729 100%
711 98%
312 43%
458 63%
394 54%
708 97%
674 92%
674 92%
Admis: Admission; ASIA: American Spinal Injury Association; IM: Intramedullary; C/H: Combined Harmonized; FIM: Function Independence Measure;
Ster.:Steroids;Surg.:Surgery;