Spinal Cord (2004) 42, 621626 & 2004 International Spinal Cord Society All rights reserved 1362-4393/04

$30.00

www.nature.com/sc

Original Article Outcomes in patients admitted for rehabilitation with spinal neurological lesions following intervertebral disc herniation
J Ronen1,2, D Goldin3, M Itzkovich1,2, V Bluvshtein1, I Gelernter4, R Gepstein2,5, R David2,5, A Livshitz2,5 and A Catz*,1,2
1

Spinal Research Laboratory, Loewenstein Rehabilitation Hospital, Raanana, Israel; 2Sackler Faculty of Medicine, Tel Aviv University, Tel Aviv, Israel; 3Tel Aviv Medical Center, Tel Aviv, Israel; 4The Statistical Laboratory, School of Mathematics, Tel Aviv University, Tel Aviv, Israel; 5Sapir Medical Center, Kfar Sava, Israel Background: Little information is available about the survival, neurological recovery, and length of stay in hospital for rehabilitation (LOS) of patients with spinal neurological decit following disc herniation (DH). Study design: Retrospective cohort study. Objective: To report on outcomes and factors aecting these. Setting: The Spinal Research Laboratory, Loewenstein Rehabilitation Hospital, Israel. Subjects: A total of 158 patients with DH spinal neurological lesions (DHSNL). Method: Data were collected retrospectively. Survival was assessed using the KaplanMeier method; relative mortality risk by the Cox proportional hazard model. Neurological recovery was evaluated by calculating the change in Frankel grades, and factors that aect it were assessed by logistic regression. LOS associations were analyzed with ANOVA. Results: The median age at lesion onset was 48 years, and the median survival 29 years. Age and gender had a signicant eect on survival, but not so lesion severity, level, or decade of onset. Of the 69 patients who had Frankel grades A, B, or C on admission, 72% achieved useful recovery to grades D or E. The severity and level of the spinal neurological lesion (SNL) had a signicant eect on recovery. The mean LOS was 87 days; it was signicantly aected by lesion severity and level and by the decade of admission to rehabilitation, and decreased with time. Conclusions: Patients with DHSNL who were admitted for rehabilitation have favorable survival and recovery rates compared with previously studied patients with other types of SNL. Their LOS is probably a function of medical requirements, but is decreasing with time. Spinal Cord (2004) 42, 621626. doi:10.1038/sj.sc.3101642; Published online 3 August 2004 Keywords: spinal cord lesion; disc herniation; survival; recovery; length of stay in hospital

Introduction
Disc herniation (DH) is usually related to degenerative disc disease, which is believed to develop owing to the combined eects of several factors, including genetic factors and changes in collagen hydration. Decreased hydration of the disc can reduce its cushioning eect, thus transmitting a greater portion of the load to the annulus brosus, which may tear, which in turn could allow extrusion of the nucleus pulposus and lead to injury to the spinal cord and nerve roots.1,2 DH is most common in the lumbar region, mainly at the L4L5 and L5S1 levels. When it causes neurological damage, it is usually by injury to nerve roots, but higher
*Correspondence: A Catz, Department IV, Spinal Rehabilitation, Loewenstein Rehabilitation Hospital, 278 Ahuza St., PO Box 3, Raanana 43100, Israel

lumbar discs may also herniate and injure the cauda equina or the conus medullaries.3 Lumbar DH usually presents with symptoms of low back pain (LBP) and sharp burning leg pain, and is often associated with numbness and tingling. In more advanced cases, sensory or motor decit may occur, sometimes with bladder and bowel diculties. DH also occurs frequently in the cervical spine. Cervical DH usually presents with cervical radiculopathy; patients complain of radiating arm pain with numbness and paresthesia and/or weakness in the muscles supplied by that nerve root. If the DH is severe and compresses the spinal cord, patients may present with signs and symptoms of myelopathy: nger numbness, weakness of the upper limbs, and diculty in walking due to spasticity and impaired sense of position.

Outcomes of disc spinal neurological lesions J Ronen et al

622

In more severe cases, bowel and bladder control dysfunction may appear. DH is less common in the thoracic spine, but herniation of the thoracic disc has been diagnosed more frequently since the introduction of new imaging techniques,4 and it frequently results in myelopathy.5 Treatment can be conservative or surgical. Many studies have assessed the management of DH,615 but have included patients without neurological decit, and outcomes frequently describe reduction of pain or are concentrated on surgical results. Only scant information is available about DH outcomes that are important for patients with SNL in rehabilitation, such as survival, neurological recovery, and length of stay as in-patients in rehabilitation.5,1619 However, DH is a relatively frequent cause of nontraumatic spinal neurological lesions (SNL) among in-patients in rehabilitation, and the etiology of almost 15% of a studied group with nontraumatic SNL in Israel.20 In other series, it appears as an underlying cause of SNL, but its prevalence is not separated from that of other conditions of degenerative spine disease.16,17,21 The present study reports on the most relevant outcome measures in patients admitted to rehabilitation with spinal neurological decit following DH (DHSNL). The information obtained by this study can serve as a basis for comparisons with future outcomes to help evaluate the eect of rehabilitation and improve patient care.

tion. Recovery rate was assessed either as any recovery, namely a recovery of at least one Frankel grade from a grade of A, B, C, or D, or as useful recovery, namely a recovery from Frankel grades A, B, or C at admission to grades D or E at discharge. The assessment of useful recovery did not include patients with an initial Frankel grade of D or E. Logistic regressions were used to examine the association between various aecting factors and recovery.22 Analyses of the length of stay in hospital for rehabilitation (LOS) due to DHSNL were performed after a square root transformation to approach a normal distribution. The associations of LOS with potentially aecting factors were analyzed by ANOVA: three-way ANOVA for age, gender, and Frankel grade, and oneway ANOVA, followed by Tukeys post hoc test, for comparisons between decade of admission and SNL severity. Statistical signicance was dened by Po0.05. Data were analyzed by SPSS for Windows, version 11 (SPSS Inc., USA).

Results
Demographic and clinical data Included in the study were 104 male (66%) and 54 female (34%) patients, with a median age of 48 years and a mean age of 48. 2 years (SD 12.7 years) at lesion onset. SNL was cervical in 38 patients (24%), thoracic in 25 (16%), and lumbar in 93 (60%). The mean time from SNL onset to rehabilitation was 18.4 months (SD 29.4 months). The initial Frankel grade was A in one patient (0.6%), B in 15 patients (9.5%), C in 53 patients (33.5%), and D in 89 patients (56.3%). In all, 14 patients were admitted before 1970, 34 between 1971 and 1980, 34 between 1981 and 1990, and 76 between 1991 and 2000. Outcome: survival Mortality data were available for 156 of the 158 patients. Of them, 123 (79%) survived at the end of the follow-up period (May 2001). The longest survival up to that time was 40.5 years. The accumulated survival was 96.5% (SE 1.6%) 5 years after the SNL onset, 88.3% (SE 2.9%) 10 years after the SNL onset, 78.7% (SE 4.3%) 15 years after the SNL onset, 73% (SE 5.1%) 20 years after the SNL onset, 62.9% (SE 6.5%) 25 years after the SNL onset, and 41.8% (SE 11.8%) 30 years after the SNL onset (Figure 1). The median survival time was 29 years. Univariate analysis showed that age at SNL onset, gender, and decade of SNL onset, but not Frankel grade or lesion level, had a signicant eect on survival (Po0.05). After controlling for gender, initial Frankel grade, SNL level, and decade of SNL onset, the mortality risk was 1.1 times higher for every additional year of age at SNL onset. After controlling for age, initial Frankel grade, SNL level, and decade of SNL onset, the mortality risk was 3.4 times higher for men

Methods
The study included 158 patients with DHSNL, admitted between 1962 and 2000 to the spinal department of Loewenstein Rehabilitation Hospital, in Israel. The patients had disabilities that required hospitalization; many had myelopathy, and most of them were admitted to rehabilitation following spinal surgery. Demographic and clinical data were retrospectively collected by reviewing the hospital charts. Mortality data were collected from the Population Registry of the Israel Ministry of Internal Aairs. Survival rates were estimated using the product limit (KaplanMeier) method, and dierences between subgroups were analyzed by logrank test (univariate analysis). The Cox proportional hazard model was used to determine the probability of mortality (hazard) in the presence of specic risk factors (multivariate analysis).22 The time of SNL onset was dened as the earliest time of DHSNL symptoms and signs mentioned in the hospital records. The severity of neurological decit below the spinal level of injury was graded according to Frankel,23 as described in a previous publication on traumatic SNL recovery.24 In most cases, Frankel grades had not been assigned during hospitalization, so for purposes of the study, they were assigned retroactively on the basis of the examination protocols. The degree of neurological recovery or regression in each patient was determined by comparing the Frankel grade of neurological decit at rst admission for rehabilitation (initial Frankel grade) with the grade at discharge from the same hospitalizaSpinal Cord

Outcomes of disc spinal neurological lesions J Ronen et al

623

than for women. The eects on the risk of mortality of initial Frankel grade, SNL level, and decade of SNL onset were not statistically signicant when controlling for the other aecting factors (Table 1).

Outcome: neurological recovery Of the 69 patients who had initial Frankel grades of A, B, or C (72.5%, or 31.6% of all DHSNL patients), 50 achieved useful recovery and showed an improvement during the course of rehabilitation to grades D or E (Table 2). Of these, 53 (76.8% or 34% of all patients with DHSNL) showed any recovery (improvement of at least one Frankel grade). The frequency of any recovery in patients who had initial Frankel grades of A, B, or C was inversely related to the severity of the original decit: grade A: 0, grade B: 73.3%, and grade C: 79.2%. Of the grade B patients,

20% achieved grade C and 53.3% grade D. Of the grade C patients, 77.4% achieved grade D and 1.9% grade E. Of the grade D patients, 37.1% achieved grade E (Table 3). When controlling each initial Frankel grade, age at admission to rehabilitation, gender, SCL level, and decade of admission to rehabilitation, for all the other listed variables, only initial Frankel grade had a signicant eect on any recovery, and only SNL level had a signicant eect on useful recovery (Po0.05; Tables 4 and 5). The odds of any or useful recovery during rehabilitation were not signicantly dierent for initial Frankel grade C versus A and B, but the odds of recovery for initial grade D were 19% of those for A and B (P 0.014; Tables 4 and 5). The odds of useful recovery during rehabilitation were not signicantly dierent for cervical versus either thoracic or lumbar lesions, but the odds of recovery for thoracic lesions were 20% of those for lumbar lesions (P 0.025; Table 5).

Table 2 Neurological recovery in DHSNL Relative proportion of the grade on admission (%) 0.6 9.5 33.5 56.3 0 Relative proportion of the grade at discharge (%) 0.6 2.5 8.9 66.5 21.4 Change in the relative proportion of the grade (%) 0 7 24.6 10.2 21.4

Frankel grade A B C D E

Figure 1 Survival following SNL due to DH. Acc survival Accumulated survival rate. Time Time from lesion onset to death or end of follow-up

Changes in the severity of neurological damage by Frankel grades between admission for rehabilitation and discharge from hospitalization, in percent of patients (N 262). Total of A, B, and C grades at admission (56.5%) minus total at discharge (23.7%) 32.8%, representing 58% of 56.5%

Table 1 Hazard (relative risk of mortality) in DHSNL, considering factors with potentially concomitant eects 95% condence interval Aecting factor Age Male gender SNL severity SNL level Cervical Thoracic Decade 19611970 19711980 19811990 Hazard 1.10 3.44 0.91 1.01 1.26 8.27 4.74 7.35 P 0.001 0.033 0.826 0.910 0.981 0.668 0.247 0.07 0.148 0.07 Lower 1.06 1.11 0.38 0.38 0.44 .860 0.58 0.87 Upper 1.15 10.70 2.16 2.69 3.55 79.78 38.88 62.28

Age and decade at lesion onset. The hazard is compared with that for lumbar lesions (level) and that for 19912000 (decade)
Spinal Cord

Outcomes of disc spinal neurological lesions J Ronen et al

624
Table 3 Relationship between severity of neurological damage and recovery Frankel grade on admission (N) A (1) B (15) C (53) D (89) 100 0 0 0 Frankel grade at discharge Aecting factor A (%) B (%) C (%) D (%) E (%) 0 26.7 0 0 0 20 20.8 0 0 53.3 77.4 62.9 0 0 1.9 37.1 SCL severity Age SCL level T versus C L versus C Male gender Decade 19611970 19711980 19811990 Odds 3.45 0.96 0.10 0.48 1.02 0.97 0.82 4.19 P 0.095 0.170 0.041 0.060 0.537 0.979 0.705 0.976 0.800 0.281 Lower 0.81 0.91 0.01 0.05 0.26 0.13 0.18 0.31 Upper 14.72 1.02 1.10 4.99 3.92 7.04 3.78 56.70 Table 5 Useful recovery in DHSNL 95% condence interval

Changes in Frankel grades between admission for rehabilitation and discharge from hospitalization, in percent of initial number of patients for each grade (N)

Table 4

Any recovery in DHSNL 95% condence interval

Aecting factor

Odds

Lower 0.460 0.051 0.96 0.10 0.24 0.59 0.10 0.35 0.37

Upper 7.32 0.72 1.02 1.26 1.44 2.70 1.94 2.37 2.29

SCL severity 0.001 FC versus FA+FB 1.84 0.386 FD versus FA+FB 0.19 0.014 Age SCL level T versus C L versus C Male gender Decade 19611970 19711980 19811990 0.99 0.384 0.26 0.36 0.111 0.59 0.246 1.26 0.552 0.759 0.44 0.280 0.92 0.860 0.92 0.863

Odds of useful recovery during rehabilitation, controlling for aecting factors. SNL severity by Frankel grading. Age and decade at admission to rehabilitation. For age, the odds are for each additional year. Each decade is compared to 19912000. C Cervical SNL; T Thoracic SNL; L Lumbar SNL

Odds of any recovery during rehabilitation, controlling for aecting factors. SNL severity by Frankel grading. Age and decade at admission to rehabilitation. For age, odds are for each additional year. Each decade is compared to 19912000. FA, FB, FC, FD Frankel grades. C Cervical SNL; T Thoracic SNL; L Lumbar SNL

decades was found statistically signicant (P 0.013). In the last decade, LOS was signicantly shorter than before 1980 (Po0.05), but the dierence in LOS between the last two decades did not reach statistical signicance. The mean LOS was 76 days (SD 58 days) for cervical SNL, 116 days (SD 88 days) for thoracic SNL, and 84 days (SD 59 days) for lumbar SNL. The dierence in LOS between cervical and thoracic SNL levels was also found signicant (Po0.05), but the dierence between the lumbar and the other SNL levels was not signicant. Outcome: main SNL complications Of the 158 patients included in this study, 26 (16.4%). had urinary complications (of them, six had vesicoureteral reux), four (2.5%) had respiratory complications, and six (3.8%) had pressure sores.

Outcome: LOS The mean LOS was 87.2. days (SD 65.1 days; range 15343 days). When controlling for two age groups at rehabilitation (p50 or X51 years), gender, and initial Frankel grade (A, B, and C or D), only Frankel grade had a signicant eect on LOS (Po0.001), while the eects of age and gender were not signicant (P 0.833 and 0.569). LOS was 296 days for the single patient with Frankel grade A; mean LOS was 159 days (SD 88 days) for patients with Frankel grade B, 107 days (SD 68 days) for C, and 61 days (SD 37 days) for D. The mean LOS was 108 days (SD 79 days) before 1970, 113 days (SD 73 days) for 19711980, 82 days (SD 55 days) for 19811990, and 74 days (SD 59 days) for 19912000. The change in LOS over the
Spinal Cord

Discussion
DHSNL may cause morbidity and death and require long hospitalization, but the orthopedic and rehabilitation literature lacks attention to survival, recovery, and LOS in series of DHSNL patients. This study has been conducted to add information to what is already available about outcomes for this patient group and to contribute to the assessment of the role of rehabilitation medicine in their care. This DHSNL patient group is too small for actuarial calculation, but a rough comparison with data from the central bureau of statistics in Israel, given a median age of 48 years at lesion onset and a median survival of 29 years, reveals that the survival of these patients is close

Outcomes of disc spinal neurological lesions J Ronen et al

625

to that of the general population in Israel during the same period. Comparisons with various SNL etiologies indicate that the mortality risk in SSSNL is the lowest.20 These, and the fact that survival was not signicantly aected by the severity or level of the lesion, may indicate that the complications of spinal neurological damage are not life-threatening in this group, as they are in patients with other SNL, including traumatic spinal cord lesions (TSCL).20,24,25 The signicant eect of decade of onset on survival, demonstrated by univariate analysis, may be misleading, as this eect did not reach statistical signicance when controlling for other aecting factors. Recovery rate over the course of rehabilitation was high: 72.5% progressed from a nonfunctional to a functional condition, as compared to 2327% in patients with TSCL studied earlier.25,26 The signicance of this nding is emphasized by the independence of useful recovery of age, gender, SNL level, and decade of admission to rehabilitation. Useful recovery was also independent of the SNL severity, but this nding indicates only that recovery rate is not signicantly dierent between patients with Frankel grades of either B or C, and that grade B lesions also had a high tendency to recover. Inference about recovery rates in DHSNL with complete lesions is impossible, as the group included only one patient with Frankel grade A. The signicantly lower recovery from a grade D lesion to normal neurological status indicates that although relatively mild, these DHSNL tend to cause a persistent neurological damage, which may have clinical implications despite the excellent functional recovery. Despite the incomplete DHSNL in the vast majority of cases and the encouraging survival and recovery, DHSNL patients can develop serious medical complications. In the present study, these complications were represented in 2.516.4% of patients by urinary impairments including vesico-ureteral reux, respiratory disorders, and pressure sores. Therefore, favorable nal outcomes cannot be taken for granted and they probably depend partly on the quality of care during rehabilitation and partly on ongoing care, socioeconomic conditions, and health-related behaviors. LOS was longer for the studied DHSNL patients than for patients with TSCL in the USA or Australia,16,17,2729 but shorter than for patients with TSCL in Europe.5,22,30 The longer LOS in Europe may be related to the dierent population in the European studies (TSCL versus DHSNL). Lesions are more severe in TSCL studied than in the DHSNL in the present study, and therefore this nding is compatible with the major impact of lesion severity on LOS found in the present study. The longer LOS in patients with more severe lesions is probably related to the greater amount of functional problems, which require therapeutic attention, and implies that the patients medical condition has a signicant impact on LOS. The shorter LOS of TSCL patients in the US and Australia, and the decrease in LOS over the decades in the present series, which is approaching US levels of LOS, might have been

related to lower or decreasing lesion severity or to improvements in medical procedure, outpatient services, and rehabilitation practices. The authors believe, however, that the trends in LOS reect mainly nancial limitations and the increasing demand for rehabilitation services. The increasing demand is reected by admissions to rehabilitation of patients with DHSNL. In our population, it doubled in the 19912000 decade compared with the previous one and increased ve-fold compared with the 19611970 decade. This may be explained partly by some increase in the need for postoperative rehabilitation as a consequence of an increase in spinal surgical procedures in Israel, which is likely to follow a similar trend reported in the US.31 Shortening of LOS and the consequent increase in the availability of rehabilitation beds may also be a contributing factor. However, our impression is that the main reasons for this tendency is the increasing awareness of the importance of rehabilitation in the care of these patients and the improved accessibility of these patients to rehabilitation services. DH in the thoracic spine is rare according to previous publications, and its incidence was noted as 1% or less of all herniated discs.4 In the present series, however, 16% of patients had thoracic DH, which probably indicates that it is more frequent among patients with DHSNL than in patients with DH without neurological compromise. This implies that the chance of a signicant neurological damage may be higher in thoracic than in other herniated discs, which is consistent with the nding that thoracic DHSNL recovered less than lumbar or cervical, and required longer LOS. The more severe and less reversible nature of thoracic DHSNL may be related to dierences in the mechanisms that create the neurological damage owing to the excess mobility of the cervical and lumbar spine. The retrospective nature of the research, and the size of the patient group imposed certain limitations on this study: additional factors that might have inuenced outcome, such as the eect of myelopathy versus root lesion or surgery versus rehabilitation, could not be analyzed; only limited information could be obtained on medical SNL complications, and the paucity of complications did not allow their further breakdown by inuencing factors; the hazard analysis may not have a sucient power to allow for conclusions to be drawn about the lack of relationship between survival and other variables. However, the hazard for SNL severity and level is 0.911.26, and its P-values are 0.660.98, so relationship between survival and these variables is unlikely. In conclusion, the present study shows that patients with DHSNL admitted for rehabilitation have favorable survival and neurological recovery, better than in previously studied patients with other types of SNL. This may be related to a combination of mostly incomplete lesions, care in the rehabilitation system, and ongoing care. LOS in DHSNL patients is related to lesion severity and it is probably adapted to patient requirements, but has been decreasing over the decades,
Spinal Cord

Outcomes of disc spinal neurological lesions J Ronen et al

626

reecting mainly an increased demand for rehabilitation services under the increasing inuence of economic factors on medical decisions. Further study with larger patient groups may clarify LOS impact on rehabilitation outcomes, help dene the optimal LOS, obtain additional information on outcomes and factors that aect them, and improve the medical and economic aspects of the rehabilitation services for these patients.

Acknowledgements
This study was supported by the Unit of Medical Services, Rehabilitation Department, Israel Ministry of Defense.

References
1 Naylor A. The biochemical changes in the human intervertebral disc in degeneration and nuclear prolapse. Orthop Clin North Am 1971; 2: 343358. 2 Humphreys SC, Eck JC. Clinical evaluation and treatment options for herniated lumbar disc. Am Fam Physician 1999; 59: 575582; 587588. 3 Tokuhashi Y, Matsuzaki H, Uematsu Y, Oda H. Symptoms of thoracolumbar disc herniation. Spine 2001; 26: E512E518. 4 Acre CA, Dohrmann GJ. Herniated thoracic disks. Neurol Clin 1985; 3: 383392. 5 Biering-Sorensen F, Pedersen V, Clausen S. Epidemiology of spinal cord lesions in Denmark. Paraplegia 1990; 28: 105118. 6 Saal JA, Saal JS. Nonoperative treatment of herniated lumbar intervertebral disc with radiculopathy. An outcome study. Spine 1989; 14: 431437. 7 Ellenberg MR et al. Prospective evaluation of the course of disc herniations in patients with proven radiculopathy. Arch Phys Rehabil 1993; 74: 38. 8 Davis RA. A long-term outcome analysis of 984 surgically treated herniated lumbar discs. J Neurosurg 1994; 80: 415421. 9 Weber H. Lumbar disc herniation. A controlled, prospective study with ten years of observation. Spine 1983; 8: 131139. 10 Loupasis GA et al. Seven to 20-year outcome of lumbar discectomy. Spine 1999; 24: 23132317. 11 Ahn UM et al. Cauda equina syndrome secondary to lumbar disc herniation: a meta-analysis of surgical outcomes. Spine 2000; 25: 15151522. 12 Shapiro S. Medical realities of cauda equina syndrome secondary to lumbar disc herniation. Spine 2000; 25: 348352. 13 Saal JS, Saal JA, Yurth EF. Nonoperative management of herniated cervical intervertebral disc with radiculopathy. Spine 1996; 21: 18771883.

14 Heckmann JG et al. Herniated cervical intervertebral discs with radiculopathy: an outcome study of conservatively or surgically treated patients. J Spinal Disord 1999; 12: 396401. 15 Kadanka Z et al. Approaches to spondylotic cervical myelopathy: conservative versus surgical results in 3-year follow-up study. Spine 2002; 27: 22052210. 16 New PW, Rawicki HB, Bailey MJ. Nontraumatic spinal cord injury: demographic characteristics and complications. Arch Phys Med Rehabil 2002; 83: 9961001. 17 Mckinley WO, Seel RT, Gadi RK, Tewksbury MA. Nontraumatic vs traumatic spinal cord injury. A rehabilitation outcome comparison. Am J Phys Med Rehabil 2001; 80: 693699. 18 Deyo RA et al. Morbidity and mortality in association with operation on the lumbar spine. J Bone Joint Surg 1992; 74: 536543. 19 Celani MG et al. Spinal cord injury in Italy: a multicenter retrospective study. Arch Phys Med Rehabil 2001; 82: 589596. 20 Ronen J et al. Survival following non-traumatic spinal cord lesions in Israel. Arch Phys Med Rehabil 2004, in press. 21 Moore AP, Blumhardt LD. A prospective survey of the causes of non-traumatic spastic paraparesis and tetraparesis in 585 patients. Spinal cord 1997; 35: 361367. 22 Dawson-Saunders B, Trapp RG. Methods for analysing survival data. In: Basic and Clinical Biostatistics. Appleton & Lange: Norwalk, CT 1994. 23 Frankel HL et al. The value of postural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia. Paraplegia 1969; 24: 179192. 24 Catz A et al. Survival following spinal cord injury in Israel. Spinal Cord 2002; 40: 595598. 25 Whiteneck GG et al. Mortality, morbidity, and psychological outcomes of persons spinal cord injured more than 20 years ago. Paraplegia 1992; 30: 617630. 26 Catz A et al. Recovery of neurologic function after spinal cord injury in Israel. Spine 2002; 27: 17331735. 27 McKinley WO, Tewksbury MA, Mujteba NM. Spinal stenosis vs traumatic spinal cord injury: a rehabilitation outcome comparison. Spinal Cord Med 2002; 25: 2832. 28 Eastwood EA, Hagglund KJ, Ragnarsson KT, Gordon WA, Marino RJ. Medical rehabilitation length of stay and outcomes for persons with traumatic spinal cord injury. Arch Phys Med Rehabil 1999; 80: 14571463. 29 Berkowitz M, OLeary PK, Kruse DL, Harvey C. Spinal cord injury: an analysis of medical and social costs. Demos: New York 1998, pp 3872. 30 Bravo P, Labarta C, Alcaraz MA, Mendoza J, Verdu A. Outcome after vertebral fractures with neurological lesions treated either surgically or conservatively in Spain. Paraplegia 1993; 31: 358366. 31 Katz JN. Lumbar spinal fusion. Surgical rates, costs, and complications. Spine 1995; 20: 78S83S.

Spinal Cord