Secondary Logo

Journal Logo

CME Article 2001 Series Number

Nontraumatic vs. Traumatic Spinal Cord Injury: A Rehabilitation Outcome Comparison

McKinley, William O. MD; Seel, Ronald T. PhD; Gadi, Ramakrishna K. MD; Tewksbury, Michael A. MBA

Author Information
American Journal of Physical Medicine and Rehabilitation: September 2001 - Volume 80 - Issue 9 - p 693-699


The pathogenesis of nontraumatic spinal cord injury (NT/SCI) includes such etiologies as spinal stenosis, tumorous compression, vascular ischemia, infection, and congenital disease. 1–4 Its annual incidence may be as high as 8 per 100,000. 5 Nontraumatic SCI has been shown to represent a significant percentage of individuals with SCI admitted to acute rehabilitation units, e.g., as much as 39% in previous studies at this institution. 6–10 Studies 6–10 have revealed that spinal stenosis accounted for 16–21% and neoplastic SCI for as many as 10–14% of SCI admissions. Impairments to movement, sensation, bladder, bowel, and sexuality often necessitate inpatient rehabilitation to enhance community reintegration and maximize medical and functional outcomes and patient/family education.

Previous studies 6–9 comparing NT/SCI with traumatic SCI (T/SCI) have revealed NT/SCI patients to be older, more often married, female, retired, paraplegic, and with incomplete injuries. These demographic and clinical presentation differences lend importance to further studies of NT/SCI and to comparisons with T/SCI (e.g., motor vehicle accident, acts of violence, falls). Previous study comparisons 8 between these two groups have shown similar FIMTM11 efficiencies and discharge-to-home rates; however, several differences were noted among clinical subgroups. Additional studies 9,12 have compared neoplastic SCI with T/SCI and found that although individuals with T/SCI achieved greater overall functional improvement, patients with neoplastic spinal cord compression had shorter rehabilitation length of stay (LOS) and achieved similar FIM efficiency and discharge-to-community rates. Rehabilitation outcomes have also been studied in patients with spinal stenosis, 6 revealing significant functional gains after inpatient rehabilitation similar to those of T/SCI individuals.

The objective of this study was to study the functional outcome of inpatient rehabilitation in individuals with NT/SCI. A comparison between NT/SCI and T/SCI controlling for age, level of injury, and American Spinal Injury Association (ASIA) classification 13 was used to assist in more clearly contrasting outcomes between these groups. It is our hope that the information obtained in this study will assist professionals in planning rehabilitation for individuals with NT/SCI.



Patients were selected from rehabilitation inpatients with SCI admitted to level I trauma centers between 1992 and 1999. All trauma centers were participants in the National Spinal Cord Injury Model Systems program. Each model systems center includes emergency medical services, intensive and acute medical care, inpatient rehabilitation, and a spectrum of community services. Data for T/SCI patients were compiled on the NIDRR SCI Database. Data for a comparison group of patients with NT/SCI were selected from one of the NIDRR model systems centers. The etiology of nontraumatic injuries included spinal stenosis, tumor, ischemia, transverse myelitis, and infection.

Patients were designated as eligible for inclusion in this study based on the following criteria: (1) admittance to the acute care hospital within 60 days of injury; (2) complete data available for age, injury level, completeness of injury, inpatient rehabilitation lengths of stay and charges, and FIM motor scores, and (3) having received at least 5 days of inpatient rehabilitation care.

Ninety-three percent of all patients were admitted to a model systems facility within 21 days of injury. Nearly 68% of the injuries were paraplegic. In regard to injury completeness, the majority of all injuries were incomplete. Patients with ASIA “C” and “D” ratings accounted for 36% and 41% of the sample, respectively. With regard to age category, nearly two-thirds of the matched patients were aged 30–59 yr.


To obtain a comprehensive understanding of treatment and recovery, the following information was collected for analysis.

Length of Stay.

The duration of stay in acute medical care and inpatient rehabilitation was calculated separately.


This was actual hospital charges for each patient.

FIM Motor Score.

Per standard protocols, 14,15 FIM motor scores ranging from 13 to 91 were assigned with higher scores denoting greater levels of independence. Scores were derived at the time of rehabilitation admission and discharge.

FIM Motor Change Scores.

These scores were calculated based on the difference between rehabilitation discharge and rehabilitation admission scores.

FIM Motor Efficiency Scores.

These scores were calculated based on dividing patients’ change scores by corresponding rehabilitation lengths of stay.


A comprehensive program of inpatient rehabilitation was provided to patients and tailored to meet their needs and abilities. Within each center, the following services were provided: nursing, occupational therapy, physiatry and related medical services, physical therapy, psychologic and neurologic assessment, recreational therapy, and social services. Each program’s admission and discharge standards were based on Commission on Accreditation of Rehabilitation Facilities standards. 16 In addition, admission and discharge decisions were based on perceptions of patients’ needs, with input obtained from third-party payment sources.

Patients were evaluated at admission to and discharge from inpatient rehabilitation. Appropriate members of the interdisciplinary rehabilitation team, using standard protocols, determined scores for functional outcome. Values were derived from assessments generally completed within 72 hr of admission and discharge. Experienced clinicians at each center collected data from a variety of sources including medical records, team conferences, and patient and family interviews. Demographic information, including age, gender, ethnicity, pre- and postinjury residence, and traumatic etiology, was obtained.

Matching Procedure.

A block design, matching procedure was used to control for the covariant effects of injury characteristics and age at injury on etiology effect. 17 Three matching variables were selected: (1) neurologic level of injury; (2) ASIA impairment classification; and (3) age category at time of injury. Age category at time of injury was comprised of two groups: a “younger” group, aged 30–59 yr, and an “older” group, aged 60 yr. These categories were based on research relating age and outcome after paraplegia SCI. 18 Each patient was identified by an injury type/age group classification (e.g., C5, ASIA “D,” aged 30–59 yr) and categorized based on etiology characteristic. Patients were selected from each etiology group to create matched dyads on the basis of their injury type/age group classification. When multiple traumatic and nontraumatic injured patients were identified within the same injury type/age classification, the patients were randomly matched until no more traumatic/nontraumatic dyads could be created. Patients without exact injury type/age group classification counterparts were excluded. Overall, 174 patients were selected, creating 87 dyads matched for neurologic level, completeness, and age category (n = 87 for both the nontraumatic and traumatic SCI groups).

Analysis of Demographic Factors.

To rule out the potential confounding effects of demographic differences among etiology groups, Cramer’s V 19 was used. Cramer’s V is a product moment r value corrected for sample size and number of categories. Cramer’s V can range from 0.000 to 1.000, with higher values denoting more substantial group differences. Analysis of gender revealed between-group differences (Cramer’s V = 0.320;P < 0.001). The nontraumatic injury group had a greater proportion of female patients (47%) in comparison with the traumatic injury group (17%). Analysis of ethnicity also revealed between-group differences (Cramer’s V = 0.305;P < 0.001). The nontraumatic injury group had a greater proportion of African-American patients (53%) than the traumatic injury group (23%). Because the etiology groups differed in gender and ethnic composition, a series of analysis of variances (ANOVAs) was then conducted to assess whether these variables were related to functional outcomes. Analyses revealed no main effects (P > 0.05) for either gender or ethnicity on functional outcomes. Thus, ethnicity and gender were not considered confounding variables and were not included in the blocked design.

Statistical Procedure for Evaluating the Matching Process.

As noted in the “Subjects” section, patients were matched on at least 25 neurologic levels of preservation, 4 levels of injury completeness, and 2 age categories, yielding at least 200 potential blocks (levels) for the injury type/age group classification variable. However, 200 blocks were too extensive for post hoc tests to yield meaningful results. Thus, the 200 blocks were consolidated into 12 blocks (6 injury type groups × 2 age groups) to enhance measurability. See Table 1 for summary information regarding the group composition of injury and age characteristic blocks.

Table 1
Table 1:
Injury level, American Spinal Injury Association (ASIA) classification, and age group composition of matched nontraumatic and traumatic groups

One-way ANOVAs were then conducted to identify main effects on two functional outcome variables: FIM motor score (M) at rehabilitation admission and FIM motor score at rehabilitation discharge. In the first set of univariate analyses, six injury classification blocks were examined. In the second set of univariate analyses, the two age groups were examined.

Injury Characteristics and Functional Outcomes.

Robust main effects (P < 0.001) for injury classification were identified for FIM motor scores at time of rehabilitation admission and discharge. When examining Tukey post hoc test differences among FIM motor scores at rehabilitation admission, patients with “ASIA A, B, C; C1–8” injuries (M = 19.73) showed lower scores than all other injury classification groups. Patients who sustained “ASIA D; C1–8” (M = 34.50) and “ASIA A, B, C; T1–12” (M = 34.77) injuries showed lower motor scores than patients with “ASIA D; T1–12” (M = 44.89) and all lumbar injuries (M = 49.00 and 46.00, respectively). Tukey post hoc test differences among FIM motor scores at rehabilitation discharge revealed that patients with ASIA A, B, C; C1–8 injuries (M = 43.20) continued to show lower scores than all other injury classification groups. Patients who sustained ASIA A, B, C; T1–12 (M = 57.35) injuries were rated as less functionally independent than patients with ASIA D; C1–8 (M = 69.42) and all lumbar injuries (M = 72.67 and 69.04, respectively). The magnitude of the ANOVA and post hoc test differences suggest that injury characteristics are related to outcomes.

Age at Injury and Functional Outcomes.

Main effects (P < 0.05) for age at injury were also identified for FIM motor scores at time of rehabilitation admission and discharge. Younger patients had significantly greater FIM motor scores (M = 37.58) at rehabilitation admission than their older counterparts (M = 32.77). For FIM motor scores at rehabilitation discharge, younger patients (M = 63.08) continued to show better outcomes than older patients (M = 56.40). The ANOVA and post hoc test differences indicated that age at injury was related to outcomes and warranted inclusion in the blocked design.

Statistical Analyses for Primary Outcome Measures.

Qualitative and quantitative statistical analyses were conducted to examine differences among traumatic and nontraumatic injured SCI patients on specific variables. For qualitative variables (e.g., discharge disposition), Cramer’s V 19 was computed. For acute care LOS, inpatient rehabilitation LOS and charges, FIM motor, FIM motor change, and FIM motor efficiency scores, analyses were conducted based on the block design, matching procedure.

Etiology-related differences for all quantitative variables were then analyzed using 2 × (2 × 6) mixed design, repeated measure ANOVAs. 17 Main effects for etiology, interaction of etiology × age category, and interaction of etiology × injury characteristic were calculated for each outcome variable. The main effects for injury characteristics and age category were treated as a covariant and removed from each estimate of error variance.

The number of planned comparisons and the potential for non-normal sampling distributions of means were considered when selecting an α level. Considering family-wise error rates, an α level of P < 0.05 was deemed acceptable. When a main effect was found using an ANOVA procedure, a Tukey post hoc test 20 was conducted to identify significant differences among variable levels.


Length of Stay and Rehabilitation Hospital Charges.

A series of 2 × (2 × 6) mixed design, repeated-measure ANOVAs compared traumatic and nontraumatic injured patients’ LOS with rehabilitation charges (Table 2). No etiology-related differences were revealed for acute care LOS (P = 0.152). However, a robust main effect was found for etiology on rehabilitation LOS (P < 0.001). Patients with T/SCI had significantly greater rehabilitation stays (M = 41.49 days) than patients with NT/SCI (M = 22.46 days). With regard to rehabilitation charges, a significant main effect for etiology was also found (P = 0.003). Spinal cord injury patients with traumatic etiologies had significantly higher charges (M = $66,117) than NT/SCI patients (M = $25,165).

Table 2
Table 2:
Length of stay (LOS) and rehabilitation charges: nontraumatic vs.traumatic spinal cord injury (SCI)

To control for the covariate effects of rehabilitation lengths of stay on charges, patients’ total rehabilitation charges were divided by rehabilitation LOS values to derive a daily charge value. A comparison of rehabilitation daily charges also revealed a main effect for etiology (P = 0.019). Mean rehabilitation daily charges for T/SCI patients ($1,629) were significantly greater than charges for NT/SCI patients ($1,148). Daily charges differed by almost $500/day between the two etiology groups. No significant interaction effects (etiology × injury characteristics; etiology × age category) were found for LOS or rehabilitation charges.

Functional Status on Admission to and Discharge from Rehabilitation Care.

A series of 2 × (2 × 6) mixed design, repeated measure ANOVAs compared the functional status of traumatic and nontraumatic injured patients (Table 3). Analysis of FIM motor scores at rehabilitation admission revealed no etiology group differences (P = 0.869). Mean FIM motor scores ranged from 36.65 to 37.03, indicating that most patients required moderate levels of assistance on admission to rehabilitation care. The interaction between etiology X injury characteristic and etiology X age category was examined. No significant interaction effects were found.

Table 3
Table 3:
Functional outcome comparisons: nontraumatic vs.traumatic spinal cord injury (SCI)

An analysis of FIM motor scores at rehabilitation discharge revealed a main effect for the etiology group (P = 0.001). Patients with SCI through traumatic means were discharged with higher FIM motor ratings (M = 68.01) than their nontraumatically injured counterparts (M = 55.84). Based on mean scores per FIM motor item (FIM total score divided by 13), the T/SCI patients (M = 5.23) typically required supervised support on discharge from rehabilitation. Nontraumatically injured SCI patients (M = 4.30) typically required minimal assistance support. No interaction effects were found at rehabilitation discharge for FIM motor scores.

Differences Between Rehabilitation Admission and Discharge Functioning.

Improvement between admission and rehabilitation discharge was also examined using the 2 × (2 × 6) mixed design, repeated measure ANOVA with change in FIM motor score as the dependent variable (Table 3). A main effect for etiology group was found for improvement in FIM motor scores (P < 0.001). The SCI patients who were traumatically injured showed greater improvement in FIM scores (M = 31.37) than their nontraumatically injured counterparts (M = 18.81). In this study, traumatically injured patients had almost twice the number of days in rehabilitation care compared with nontraumatically injured patients. To control for the covariate effect of rehabilitation length of stay on FIM motor improvement, FIM efficiency scores were calculated based on patients’ change scores divided by their corresponding rehabilitation lengths of stay. The 2 X (2 X 6) ANOVA results revealed no main effect for etiology on FIM motor efficiency scores (P = 0.240). Last, no significant interactions among etiology, injury characteristic, and age category were found for FIM motor change and efficiency scores.

Discharge Disposition.

The number of patients within each etiology group discharged to home and institutional settings was determined (Table 4). For each etiology group, a “likelihood of institutionalization” ratio was calculated by dividing the total number of patients by the number institutionalized. A qualitative analysis (Cramer’s V) of etiology effect on postdischarge disposition was not significant (P = 0.899). Both traumatic and NT/SCI patients were discharged to institutional settings at equivalent rates (9:1). Patients aged >60 yr were then identified, and a likelihood of institutionalization rate was calculated for older traumatically injured and nontraumatically injured patients. Whereas a trend was observed with older traumatically injured patients (9:1) less likely to be institutionalized than their nontraumatically injured counterparts (4:1), statistical analysis indicated that these institutionalization rates were equivalent (P = 0.230).

Table 4
Table 4:
Discharge disposition comparisons: traumatic vs.nontraumatic spinal cord injury (SCI)


Individuals with SCI can undergo a dramatic change in lifestyle secondary to their physical impairment and functional disabilities. Additionally, the cost of care after SCI is extraordinary for both acute hospitalization and long-term follow-up. Nontraumatic SCI has been shown to represent a significant proportion of individuals presenting for rehabilitation 6–10; however, most previous outcome studies have focused on that of T/SCI. Studies have also revealed differences in a demographic makeup between the two groups with, among other things, a significant age difference. This article seeks to further study the functional outcome of NT/SCI in comparison with that of their traumatic counterparts by matching patients with similar injury characteristics (level and completeness of injury) and by age groups. The following paragraphs discuss issues related to the functional outcome findings.

Length of Stay and Charges.

Acute care LOS was similar between NT/SCI and T/SCI groups in this study. Previous literature 9,11 reveals similar results between T/SCI patients and spinal cord tumor patients (a subset of NT/SCI). We believe that factors potentially influencing acute care LOS for T/SCI patients include issues arising from the trauma itself (e.g., spinal and hemodynamic stabilization, concomitant abdominal injuries, multiple fractures, concomitant brain injury) and other medical complications during acute care LOS (e.g., infectious venous thrombosis and pain). Factors influencing acute LOS in NT/SCI patients include issues related to the etiologic diagnostic workup (e.g., radiologic, laboratory, and electrodiagnostic testing), potential surgical interventions (e.g., tumor resection or lumbar decompression), and acute medical complications (e.g., infectious venous thrombosis and pain).

This study reveals that NT/SCI patients had a shorter rehabilitation LOS than those with T/SCI. Factors potentially influencing longer LOS in T/SCI patients include treatment issues associated with traumatic injuries (e.g., chest trauma, gunshot wound, fractures) and medical complications. McKinley et al. 9,12 reported shorter rehabilitation LOS in patients with neoplastic SCI, noting that the rehabilitation team, patient, and families may desire earlier discharge given the patient’s potentially limited life expectancy. This shorter LOS allows patients to have more time at home with family and friends, thus enhancing quality of life. Traumatic SCI patients are expected to have longer life expectancies, and inpatient rehabilitation goals include maximizing long-term functional gains. Also, T/SCI patients present with contusion, hemorrhage, or other forms of secondary SCI that may require longer LOS for neurologic and functional improvement, whereas tumor and spinal stenosis SCI patients may have some enhanced recovery of neurologic function after decompression, resection, or radiation therapy or if concomitant edema responds to treatment with steroids.

Rehabilitation charges for T/SCI patients are also greater than those for NT/SCI patients. Longer rehabilitation LOS, in this group, was likely a factor influencing this finding. Additionally, T/SCI patients may have had more diagnostic testing performed while on the rehabilitation unit than those with NT/SCI, including laboratory and imaging studies for associated medical issues. Other reasons could include differences among rehabilitation centers.

FIM Scores and Discharge Disposition.

Admission FIM scores for NT/SCI and T/SCI groups were similar in this study. This was probably influenced by the fact that these groups were matched for age, LOS, and completeness (all of which have been shown to affect functional outcome in SCI patients). Higher discharge FIM scores were noted in the T/SCI group as compared with NT/SCI and may be a result of several factors. The longer LOS for traumatic patients allowed for more rehabilitation time and the potential for additional functional gains. Neurologic recovery within and below the zone of injury has been described in acute T/SCI literature. Increased recovery of neurologic function may have led to increase functional gains during rehabilitation LOS.

The comparable FIM efficiencies (FIM change over time) between the two groups are an important finding and seem to indicate similar rates of functional improvement, despite differences in etiology of injury. McKinley et al. 8 showed similar FIM efficiencies when comparing rehabilitation outcomes between tumor SCI and T/SCI patients. Discharge rates were also comparable between T/SCI and the NT/SCI groups. Previous studies between rehabilitation of brain tumor and their nontumor counterparts, and also T/SCI and tumor SCI patients, showed comparable discharge rates as well. Both T/SCI and NT/SCI groups had close to 90% discharge-to-home rates.


Nontraumatic SCI has been shown to represent a significant proportion of SCI rehabilitation admissions. This study adds to the increasing literature examining the functional outcome of these individuals and suggests that they can achieve outcomes similar to patients with T/SCI. Individuals with NT/SCI had similar FIM efficiencies and discharge-to-home rates as those with T/SCI; however, they had shorter rehabilitation LOS, lower FIM change, and lower charges. Future studies pertaining to outcome of these individuals is suggested and should include a larger sample size, documentation of associated morbidities, and long-term follow-up outcome maintenance. It is our hope that these and other studies on individuals with NT/SCI will increase our understanding of their rehabilitation management and functional outcomes.


1. Adams RD, Salam-Adams M: Chronic nontraumatic diseases of the spinal cord. Neurol Clin 1991; 9: 605–23
2. Dawson DM, Potts F: Acute nontraumatic myelopathies. Neurol Clin 1991; 9: 585–602
3. Schmidt RD, Markovchick V: Nontraumatic spinal cord compression. J Emerg Med 1992; 10: 189–99
4. Byrne TN, Waxman SG. Spinal Cord Compression: Diagnosis and Principles of Treatment: Contemporary Neurology Series. Philadelphia, FA Davis, 1990
5. Kurtzke JF: Epidemiology of spinal cord injury. Exp Neurol 1975; 48: 163–236
6. McKinley W, Tellis A, Cifu D, et al: Rehabilitation outcome of individuals with nontraumatic myelopathy resulting from spinal stenosis. J Spinal Cord Med 1998; 21: 131–6
7. McKinley WO, Conti-Wyneken A, Vokac C, et al: Rehabilitative functional outcome of patients with neoplastic spinal cord compression. Arch Phys Med Rehabil 1996; 77: 892–5
8. McKinley WO, Seel R, Hardman J: Nontraumatic spinal cord injury: incidence, epidemiology and functional outcome. Arch Phys Med Rehabil 1998; 79: 1186–7
9. McKinley WO, Huang M, Tewksbury MA: Neoplastic vs. traumatic spinal cord injury: an inpatient rehabilitation comparison. Am J Phys Med Rehabil 2000; 79: 138–44
10. Gibson CJ:Final Report of the Rochester Regional Model Spinal Cord Injury System: 9–30–85 to 7–29–90. Rochester NY, Rochester Regional Model Spinal Cord Injury System, 1991
11. Guide for the Uniform Data Set for Medical Rehabilitation (including the FIMTMinstrument), Version 5.1. Buffalo NY, State University of New York at Buffalo, 1997
12. McKinley WO, Huang ME, Brunsvold KT: Neoplastic vs traumatic spinal cord injury: an outcome comparison after inpatient rehabilitation. Arch Phys Med Rehabil 1999; 80: 1253–7
13. International Standards for Neurological and Functional Classification of Spinal Cord Injury–Revised. Chicago, American Spinal Injury Association, 1996
14. Hall KM, Johnston MV: Outcome evaluation in traumatic brain injury rehabilitation. Part II. Measurement tools for a nationwide data system: 1994. Arch Phys Med Rehabil 1994; 75: SC10-18
15. Ditunno JF, Cohen ME, Formal C, et al: Functional outcomes, in Stover SL, DeLisa J, Whiteneck GG (eds):Spinal Cord Injury: Clinical Outcomes from the Model Systems. Gaithersburg MD, Aspen, 1995, pp 170–84
16. Standards Manual and Interpretive Guidelines for Medical Rehabilitation. Tucson AZ, Commission on Accreditation of Rehabilitation Facilities, 1997
17. Tabachnick BG, Fidell LS: Using Multivariate Statistics, ed 2. New York, Harper Collins, 1989
18. Cifu DX, Huang ME, Kolakowsky-Hayner SA, et al: Age, outcome, and rehabilitation costs after paraplegia spinal cord injury. J Neurotrauma 1999; 16: 805–15
19. Hays WL: Statistics, ed 5. New York, Harcourt Brace, 1994
20. Bruning JL, Kintz BL:Computational Handbook of Statistics. Reading MA, Addison Wesley Longman, 1997

Spinal Cord Injury; Nontraumatic Spinal Cord Injury; Rehabilitation; Functional Outcome

© 2001 Lippincott Williams & Wilkins, Inc.