NUMEROUS STUDIES HAVE addressed the incidence of

Transcription

1 332 ORIGINAL ARTICLE Trends in New Injuries, Prevalent Cases, and Aging With Spinal Cord Injury Michael J. DeVivo, DrPH, Yuying Chen, MD, PhD ABSTRACT. DeVivo MJ, Chen Y. Trends in new injuries, prevalent cases, and aging with spinal cord injury. Arch Phys Med Rehabil 2011;92: Objective: To determine the characteristics of the newly injured and prevalent population with spinal cord injury (SCI) and assess trends over time. Design: Prospective cohort study. Setting: SCI Model Systems and Shriners Hospital SCI units. Participants: The study population included people whose injuries occurred from 1935 to 2008 (N 45,442). The prevalent population was estimated based on those who were still alive in Losses to follow-up (approximately 10%) were excluded from the prevalent population. Interventions: Not applicable. Main Outcome Measures: Demographic and injury characteristics, mortality, self-reported health, rehospitalization, FIM, Craig Handicap Assessment and Reporting Technique, and the Diener Satisfaction with Life Scale. Results: Mean age at injury increased 9 years since the 1970s. Injuries caused by falls and injuries resulting in high-level tetraplegia and ventilator dependency are increasing, while neurologically complete injuries are decreasing. Discharge to a nursing home is increasing. The mean age of the prevalent population is slightly higher than that of newly injured individuals, and the percentage of incident and prevalent cases older than 60 years is the same (13%). Prevalent cases tend to be less severely injured than incident cases, and less than 5% of prevalent cases reside in nursing homes. Within the prevalent population, life satisfaction and community participation are greater among persons who are at least 30 years postinjury. These findings are a result of very high mortality rates observed after 60 years of age. Conclusions: Within the prevalent population, the percentage of elderly persons will not increase meaningfully. Those who reach older ages will typically have incomplete and/or lower-level injuries and will have relatively high degrees of independence and overall good health. Key Words: Epidemiology; Mortality; Rehabilitation; Spinal cord injuries by the American Congress of Rehabilitation Medicine NUMEROUS STUDIES HAVE addressed the incidence of SCI throughout the United States and the rest of the world The most common causes of SCI as well as the demographic characteristics and nature of injury of persons who get injured are well established In the United States, motor vehicle crashes are the leading cause of SCI, followed by falls, acts of violence (typically gunshot wounds), and sports. Acts of violence and sports decline with advancing age, while falls increase with advancing age and become the leading cause of SCI among persons who are at least 50 years of age In other countries, motor vehicle collisions are also typically the leading cause of SCI, but acts of violence are much less common. In the United States, the average age at time of injury has been increasing in conjunction with the average age of the general population, while approximately 80% of new SCIs continue to occur among men Unfortunately, very little is known about the prevalence of SCI. Given the long life expectancies currently enjoyed by persons with SCI as well as trends toward increasing age at time of injury, there has been increased speculation and concern about the ability of our health care system to deal effectively with the potential needs of an aging SCI population. In 1980, the prevalence of SCI in the United States was estimated to be 906 per million population, or approximately 200,000 persons. 18 In 1988, a second study was undertaken with substantial funding from the Paralyzed Veterans of America and resulted in a more accurate estimate of 721 per million population, or 176,965 persons alive with SCI. 19,20 Because of increased life expectancy and, to a lesser extent, changes in incidence over time (in part because of an increased population at risk), the prevalence of SCI is increasing. Based on the 1988 prevalence figures developed by Berkowitz 19 and Harvey 20 coupled with assumptions about incidence and life expectancy, the prevalence of SCI in the United States was estimated to increase to 276,281 by Moreover, while growth in prevalence was slowing over time, the prevalence would be expected to continue to increase for the foreseeable future. The most recent attempt to characterize the prevalent SCI population in the United States occurred in The current median age of persons with SCI was 41 years, 29% were female, 89.4% were white, 51.1% were married, 26.5% were college graduates, 25.4% were veterans, and 43.7% had cervical injuries. Almost half (45.4%) of the prevalent SCI popula- From the Department of Physical Medicine and Rehabilitation, University of Alabama at Birmingham, Birmingham, AL. Presented in part to the American Spinal Injury Association, September 23 26, 2009, Dallas, TX. Supported by the National Institute on Disability and Rehabilitation Research (grant no. H133A060039). No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit on the authors or on any organization with which the authors are associated. Correspondence to Yuying Chen, MD, PhD, Spain Rehabilitation Center, Room 515, 1717 Sixth Ave S, Birmingham, AL , yychen@uab.edu. Reprints are not available from the author /11/ $36.00/0 doi: /j.apmr AIS CHART IRB NSCISC SCI SSDI List of Abbreviations American Spinal Injury Association Impairment Scale Craig Handicap Assessment and Reporting Technique institutional review board National Spinal Cord Injury Statistical Center spinal cord injury Social Security Death Index

2 SPINAL CORD INJURY EPIDEMIOLOGY, DeVivo 333 tion in the United States obtained their injury as a result of a motor vehicle crash, while 16.8% were a result of falls, 16.3% were a result of sports, 5.2% were a result of violence, and 16.3% were a result of all other causes. There have been 2 studies of SCI prevalence conducted outside the United States. In Australia, the prevalence of SCI was estimated to be 681 per million population in 1997, while in Helsinki, Finland, SCI prevalence was estimated to be only 280 per million population. 22,23 Both of these estimates are consistent with the lower SCI incidence rates in those countries compared with the United States. Characteristics of the prevalent SCI population in Australia were not reported; however, in Helsinki, 76% of persons with SCI were male, 18% were at least 60 years of age, 46% had tetraplegia, and 43% had neurologically complete injuries. The purposes of this study were to develop new estimates of the characteristics of the prevalent SCI population in the United States that would assist in evaluating the future health care needs of this aging population, and to compare the characteristics of the prevalent population with those of newly injured persons. METHODS Setting Since 1970, the United States Department of Education has used a competitive grant application process to fund hospitals throughout the United States to serve as SCI Model Systems. Competitions are held every 5 years, with some SCI Model Systems losing funding and dropping out of the program while others remain and new SCI Model Systems join the program. Overall, 26 such hospitals have been funded, with 14 funded during the most recent grant period. 24,25 SCI Model Systems have been located in Massachusetts, New York (n 3), New Jersey, Pennsylvania (n 2), Virginia (n 2), Georgia, Florida, Alabama, Ohio, Michigan (n 2), Illinois, Wisconsin, Missouri, Louisiana, Texas, Colorado, Arizona, California (n 2), Washington, and Washington, DC. In addition, there are 3 designated SCI units within the Shriners Hospital network located in Pennsylvania, Illinois, and California. Collectively, SCI Model Systems treat approximately 15% of all new SCIs that occur in the United States each year. All SCI Model Systems are required to submit data on persons with SCI to the NSCISC database, while Shriners Hospital SCI units have submitted data on children with SCI to a parallel database also housed at the NSCISC. Participants The study population included 45,442 persons whose SCI occurred as a result of a traumatic event and who were treated at either a SCI Model System or a Shriners Hospital SCI unit. Most of these persons were enrolled in the NSCISC database or the comparable National Shriners Hospital SCI database. 17,24,25 Eligibility criteria for these databases include having a traumatic SCI; being admitted to the SCI Model System within 1 year of injury; residing within the geographic catchment area of the SCI Model System (to facilitate follow-up); completing rehabilitation within the SCI Model System, recovering within 7 days without rehabilitation, or dying during the SCI Model System hospitalization; and informed consent. In each case, the treating physician confirms eligibility for enrollment in the database. However, for purposes of this study, the usual eligibility criteria for inclusion in these databases were relaxed to include a broader range of persons with SCI that would be more suitable for an epidemiologic study. For example, the usual requirements that persons be admitted within 1 year of injury and reside in the geographic catchment of the SCI Model System were waived. Therefore, although the first SCI Model Systems were funded in 1970, some persons included in this study were injured as long ago as 1935 and had very delayed admissions to the SCI Model System. Nonetheless, 99% of persons in this study were injured since Demographic and Injury Profile Demographic and injury characteristics were collected by trained personnel at each hospital using a standardized protocol during initial hospital care. 24,25 Age at injury, sex, and etiology of injury are obtained either from the hospital admission record or by personal interview. Education level is the highest level of formal education completed at the time of injury and is obtained by personal interview. Discharge disposition is obtained from the hospital discharge summary, with institutional discharge defined to include nursing homes; medi-centers; skilled nursing facilities; institutions licensed as hospitals but providing essentially long-term, custodial, chronic disease care; or assisted living units in retirement villages. Neurologic examinations were performed by physicians or specially trained nurses or physical/occupational therapists in accordance with the version of the International Standards for Neurological Classification of SCI that was in use at the time the examinations were performed. 26 Prior to August 1993, neurologic extent of injury was assessed by using the Frankel classification scale. 27 The major difference between the Frankel classification scale and the International Standards is that some injuries would be classified incomplete by the Frankel classification scale but complete by the International Standards. Ventilator dependency was defined as requiring either partial or total respiratory support (including electrophrenic pacers) on a daily basis. Ventilator dependency was measured at either discharge from the SCI Model System or first anniversary of injury, so persons who were initially ventilator-dependent but were successfully weaned were not included in this category. Additional background information related to data collection forms and instructions is located on the NSCISC web site at Survival Status Determination Deaths among persons in the study were identified either from routine follow-up conducted by data collection personnel at each hospital or periodic searches of the SSDI and other online databases such as state vital statistics and newspaper obituary files available at The most recent search of online files was conducted in March Persons not found in the SSDI or other databases and not reported as deceased by data collection personnel at each hospital were presumed alive for purposes of inclusion in the estimated SCI prevalent population. The SSDI has been established to be 92.4% sensitive and 99.5% specific in identifying survival status for persons in the NSCISC database. 28 Study participants from SCI Model Systems who lost funding and dropped out of the program as described previously whose identity is not known and who were not previously reported as deceased were considered lost to follow-up and not included in the estimate of the current prevalent SCI population, approximately 10% of total participants in the database. Physical and Psychosocial Outcome Measures Physical and psychosocial outcomes were obtained among eligible NSCISC and Shriners Hospital database participants at

3 334 SPINAL CORD INJURY EPIDEMIOLOGY, DeVivo 1, 5, and every 5 years postinjury. These data were obtained by personal interview during a clinic visit, phone interview, or mailed questionnaire. However, this information was not collected on those persons who were enrolled in this study through the expansion of eligibility criteria described previously. Participants were asked at each follow-up how many times they had been hospitalized during the past year for what reasons and for how long. Subjective health was measured by a single item from the Medical Outcomes Study Short Form-36 that reads, In general, would you say that your health is excellent (5), very good (4), good (3), fair (2) or poor (1)? 29 The motor component of the FIM was used to quantify activity limitation by assessing performance in 4 areas self-care, sphincter control, mobility, and locomotion with total scores ranging from 13 (complete dependence) to 91 (complete independence). 30 FIM data were collected only by interview and not by mailed questionnaire. Participation was measured by 1 of the 4 subscales of the CHART: physical independence, mobility, occupation, and social integration. 31 Each of the subscale scores ranges from 0 to 100, with higher scores reflecting greater participation. The Diener Satisfaction with Life Scale 32 was used to assess subjective overall life satisfaction. This scale consists of 5 statements that are each rated on a 7-point Likert-type scale with responses ranging from strongly disagree to strongly agree. Total scores range from 5 to 35, with higher scores reflecting greater life satisfaction. Statistical Analysis Estimates of the characteristics of the incident and prevalent population are expressed as means for continuous items and percentages for categoric items. The incident population included 45,442 new injuries that occurred from 1935 to The prevalent population included 24,631 persons who had a SCI from 1935 to 2008 and were still alive by December The physical and psychologic outcomes of those prevalent cases were also compared across years postinjury using data obtained from 2006 to Tests of statistical significance of trends over time were based on the Pearson chi-square test for categoric data and 1-way analysis of variance for continuous items. 33,34 Annual mortality rates were determined by creating a person-year data set in which each year of follow-up for each person was treated as a separate observation. 35 Thus, a person who was followed for 5 years and died during the fifth year would contribute 5 observations to the data set. The person in this example would be considered as alive at the end of each of the first 4 observations and dead for the fifth observation. Using this approach resulted in a person-year data set of 541,181 observations from which to calculate mortality rates. IRB Approval Each participating hospital has approval from its IRB to enroll patients in the NSCISC or Shriners Hospital databases. In addition, the NSCISC has approval from the IRB at the University of Alabama at Birmingham to maintain the NSCISC and Shriners Hospital databases as well as the expanded database of additional patients used in this study. The NSCISC also has IRB approval to conduct studies using the NSCISC database. RESULTS Trends in the characteristics of newly injured persons appear in table 1. Average age at time of injury has increased 8.8 years since 1970 (P.0001). Moreover, the percentage of persons who were older than 60 years of age at time of injury increased from 4.6% in the 1970s to 13.2% since 2005 (P.0001). This is consistent with the increase in average age of the general population of the United States. However, possible changes in underlying age-specific incidence rates also might have contributed to this shift toward older age at injury. Unfortunately, there have not been any new population-based studies of the incidence of SCI in the United States in over a decade. The percentage of females has increased only slightly over time (P.0001). Since 2005, 21.7% of new injuries occurred among females. This is consistent with the increase in average age at injury because injuries occurring at older ages are more likely to occur among females than injuries occurring at younger ages. Motor vehicle crashes continue to be the leading cause of SCI in the United States. However, injuries as a result of falls have increased and now rank second. Again, this is consistent with the increased average age at injury because falls are the leading cause of SCI among persons who are more than 45 years of age at injury. The new injuries resulting from violence peaked in the early 1990s and subsequently have decreased. Average education levels of newly injured persons also have increased along with the average age at time of injury and decline in injuries caused by acts of violence. Over time, this increased education level may lead to improved postinjury employment prospects, although this has not yet happened. 36 The increase in average age at injury also has led to an increase in the percentage of persons who are discharged to nursing homes from 5% in the 1970s to 10.8% since 2005 (P.0001). The percentage of persons with high cervical injuries has increased from 12.3% during the 1970s to 27.2% since 2005 (P.0001). Concurrently, the percentage of people whose injuries are neurologically incomplete also has increased from 43.9% to 52.7% (P.0001), and the percentage of injuries resulting in ventilator dependency has increased from 1.5% to 4.6% (P.0001). These trends are likely a result of a combination of improved acute survival among persons with higher level injuries, changes in etiology, and increased age at injury (falls among the elderly usually result in incomplete cervical injuries). Characteristics of the prevalent SCI population in the United States compared with those of new injuries occurring between 2005 and 2008 appear in table 2. As expected, the average current age of the prevalent population is higher than that of new injuries (45y vs 37.1y). However, contrary to expectations, the percentage of persons who are greater than 60 years of age was almost identical (13.7% vs 13.2%), and the percentage of females was identical (21.7%). Persons in the prevalent population had lower injury levels, less likelihood of ventilator dependency, and a reduced likelihood of residing in a nursing home compared with newly injured persons, all of which would be expected given differential long-term survival rates over time. Although neurologically complete lesions were also more common in the prevalent population (51.3% vs 47.3%) than among persons injured since 2005, they were less common compared with injuries occurring in previous periods. As seen in table 3, the mean current age of persons in the prevalent population is higher than it is for newly injured persons for all injury levels and AIS grades. Mean age increases and the age gap decreases as injuries become more incomplete. Interestingly, however, ventilator-dependent persons in the prevalent population are actually younger on average than newly injured ventilator-dependent persons. Again, this is a result of differential survival rates among older and younger ventilator-dependent persons. 37

4 SPINAL CORD INJURY EPIDEMIOLOGY, DeVivo 335 Table 1: Characteristics of New Injuries by Calendar Year Characteristic N* Age at injury Mean (y) Age 60y (%) Female (%) Etiology (%) Motor vehicle Fall Violence Sports Other Education (%) High school High school graduate High school Institutional discharge (%) Injury level (%) C C Paraplegia Normal AIS grade (%) A B C D E Ventilator use (%) NOTE. All comparisons across calendar years are statistically significant, P *Sample size varies by characteristics because of unknown and missing responses. Outcomes of persons in the prevalent SCI population are generally as good or better among 30-year survivors compared with those of persons injured more recently (table 4). The CHART physical independence and Diener life satisfaction scores consistently increased over the 30-year time span postinjury. Rehospitalization rate, self-reported health, and CHART mobility score were better at year 30 compared with earlier years. Again, this relates to differential long-term survival rates in that those with poorer short-term outcomes often do not live 20 or 30 years after injury. Table 2: Characteristics of Incident and Prevalent Populations Characteristic New Injuries Prevalent Cases (N* 3434) (N* 24,631) Age Mean (y) Age 60y (%) Female (%) Institutional residence (%) C C Paraplegia Normal AIS grade (%) A B C D E Ventilator use (%) *Sample size varies by characteristics because of unknown and missing responses. Table 3: Age Distribution by Various Characteristics of Incident and Prevalent Populations Characteristic New Injuries Prevalent Cases (N* 3434) (N* 24,631) Injury level C C Paraplegia Normal AIS grade A B C D E Ventilator use Yes No NOTE. Values are mean age (y). *Sample size varies by characteristics because of unknown and missing responses.

5 336 SPINAL CORD INJURY EPIDEMIOLOGY, DeVivo Table 4: Outcomes of Prevalent Population by Time Postinjury Postinjury Year Outcome P N* 1,591 1, Current age (y) Self-reported health FIM motor score Diener life satisfaction CHART Physical independence Mobility Social integration Rehospitalization (%) NOTE. Data are expressed as mean unless otherwise specified. *Sample size varies by outcomes because of unknown and missing responses. Annual mortality rates by current age and injury severity appear in table 5 and are offered to help explain the trends seen in tables 1 to 4. Among persons with neurologically complete (AIS A) cervical injuries, the annual percentage who die is 7.2% between ages 55 and 59 years, 8.9% between ages 60 and 64 years, 12.2% between ages 65 and 69 years, 17% between ages 70 and 74 years, and 33.3% between ages 75 and 79 years. Therefore, a 55-year-old person with complete cervical injury has only a 69% chance to reach age 60 years, a 43% chance to reach age 65 years, a 23% chance to reach age 70 years, and a 9% chance to reach age 75 years. Less severely impaired persons have somewhat lower annual mortality rates. DISCUSSION Contrary to expectations, it appears from these results that there will not be a significant increase in the near future in numbers of elderly persons with more severe SCI who are in need of substantial services such as attendant care, nursing home placement, and inpatient admissions resulting from secondary medical complications. Although health and function generally decline with advancing age, those persons who survive more than 20 years postinjury usually have good selfperceived health, are reasonably physically independent, report good satisfaction with life, and are relatively well integrated back into the community. Table 5: Percentage of Persons Who Die Each Year by Age and Injury Severity Age (y) Non-SCI 44 All SCI AIS A Cervical AIS A These findings are consistent with prior studies of long-term outcomes. For example, Whiteneck et al 38 reported that average CHART scores for each subscale are generally higher for persons with injuries of longer duration but lower for older persons, all other things equal. Similarly, Charlifue et al 39 found that perceived health scores were not meaningfully different over time postinjury, while satisfaction with life scores were higher and days rehospitalized per year were lower for persons with longer injury duration. Similarly, Hall et al 40 reported average motor FIM scores stratified by neurologic level increased over time throughout the first 5 years after injury. The characteristics of the prevalent SCI population in the United States are similar to those reported for Helsinki. 23 In the United States, 13.7% of the prevalent SCI population is at least 60 years of age compared with 18% in Helsinki, 21.7% are female compared with 24%, and 48.8% have cervical injuries compared with 46%. The characteristics of the current prevalent SCI population in the United States are also consistent with those reported in In 1988, the median age of the prevalent SCI population was 41 years, so age has increased only slightly in the past 20 years. The percentage of females was slightly higher in 1988 compared with today (29% vs 21.7%), while the percentage of cervical injuries was slightly lower in 1988 (43.7% vs 48.8%). In the United States, the average age of persons in the prevalent SCI population will continue to increase slowly but will probably plateau in the mid-50s unless there are significant declines in age-specific mortality rates for persons age 50 years or greater. Moreover, the percentage of less severely injured persons in the prevalent population will also likely increase because of their better long-term survival rates. The average age of newly injured persons will probably continue to increase with the advancing age of the general population. This will likely result in an increasing percentage of injuries as a result of falls that cause neurologically incomplete cervical injuries. Future primary prevention programs might best be targeted at falls among the elderly as well as motor vehicle crashes. Study Limitations There are several important limitations of this study. First, the study population is not population-based. Therefore, it is not possible to estimate either incidence or prevalence rate from this study. Although it is possible to estimate the proportion of persons in the prevalent population who have certain

6 SPINAL CORD INJURY EPIDEMIOLOGY, DeVivo 337 characteristics, it is not possible to estimate the actual numbers of such persons directly from this study. However, it would be possible to apply these proportional data to previous estimates of prevalence to approximate numbers of people with each characteristic. 21 Moreover, because the NSCISC database has previously identified enrollment biases, and most persons in this study were enrolled in the NSCISC database, estimates of the prevalent population will also be slightly biased. Persons with more severe injuries, nonwhites, males, and injuries caused by acts of violence are slightly overrepresented in the NSCISC database and as a result will also be overrepresented in this study. 41 This study includes only SCI resulting from trauma. The characteristics of nontraumatic SCI are different and will vary substantially based on the actual nontraumatic cause of impairment. Because SCI Model Systems began in 1970, only people admitted to a model system since 1973 were included in this study. Some persons injured before 1973 were included if they were admitted on a very delayed basis. Nonetheless, people injured before 1973 who might still be alive are probably slightly underrepresented in this study. These individuals are likely to be slightly older on average with lower injury level and more incomplete injuries than those who were included in the study. Some misclassification of survival status also likely occurred as a result of the use of the SSDI. A previous study of persons with traumatic brain injury revealed that deaths among young persons, females, nonwhites, and those who were unmarried were more likely to go unreported to the SSDI. 42 As a result, these groups may be slightly overrepresented in the current prevalent population profile. Moreover, those most likely to be deceased (more severe injuries) may also be overrepresented from misclassification of their status as still being alive in Finally, losses to follow-up may also create a small bias in the estimated prevalent population profile. Overall, 20% of persons in the NSCISC database had follow-up terminated before 2008 with a status of still being alive when follow-up ceased. However, most of these losses are from SCI Model Systems that are no longer part of the program. These losses should not significantly bias results because all such persons are uniformly lost, and there are no characteristics that will differentiate lost from not lost persons. Nonetheless, some losses do occur among currently participating SCI Model Systems. Previous investigations of losses to follow-up among persons enrolled in the NSCISC database have revealed that losses are more likely to occur among persons with AIS D injuries, those who are unmarried, and those who do not have good insurance. 43 As a result, these persons may be underrepresented in the final estimates of prevalent population characteristics. CONCLUSION The percentage of elderly persons in the prevalent population will not increase meaningfully until significant progress is made in reducing the high annual mortality rates currently observed among older persons with SCI. Those who reach older ages will typically have incomplete and/or lower level injuries and will have relatively high degrees of physical independence and overall good health and community participation. References 1. Acton PA, Farley T, Freni LW, et al. Traumatic spinal cord injury in Arkansas, 1980 to Arch Phys Med Rehabil 1993;74: Johnson RL, Gabella BA, Gerhart KA, et al. Evaluating sources of traumatic spinal cord injury surveillance data in Colorado. Am J Epidemiol 1997;146: Price C, Makintubee S, Herndon W, et al. Epidemiology of traumatic spinal cord injury and acute hospitalization and rehabilitation charges for spinal cord injuries in Oklahoma, Am J Epidemiol 1994;139: Surkin J, Gilbert BJ, Harkey HL, III, et al. Spinal cord injury in Mississippi: findings and evaluation, Spine (Phila Pa 1976 ) 2000; 25: Thurman DJ, Burnett CL, Jeppson L, et al. Surveillance of spinal cord injuries in Utah, USA. Paraplegia 1994;32: O Connor PJ. Forecasting of spinal cord injury annual case numbers in Australia. Arch Phys Med Rehabil 2005;86: Ahoniemi E, Alaranta H, Hokkinen EM, et al. Incidence of traumatic spinal cord injuries in Finland over a 30-year period. Spinal Cord 2008;46: O Connor RJ, Murray PC. Review of spinal cord injuries in Ireland. Spinal Cord 2006;44: Exner G, Meinecke FW. Trends in the treatment of patients with spinal cord lesions seen within a period of 20 years in German centers. Spinal Cord 1997;35: Maharaj JC. Epidemiology of spinal cord paralysis in Fiji: Spinal Cord 1996;34: Biering-Sorensen E, Pedersen V, Clausen S. Epidemiology of spinal cord lesions in Denmark. Paraplegia 1990;28: Divanoglou A, Levi R. Incidence of traumatic spinal cord injury in Thessaloniki, Greece and Stockholm, Sweden: a prospective population-based study. Spinal Cord 2009;47: Hart C, Williams E. Epidemiology of spinal cord injuries: a reflection of changes in South African society. Paraplegia 1994; 32: Jackson AB, Dijkers M, DeVivo MJ, et al. A demographic profile of new traumatic spinal cord injuries: change and stability over 30 years. Arch Phys Med Rehabil 2004;85: 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 1999;80: DeVivo MJ. Sir Ludwig Guttmann Lecture: trends in spinal cord injury rehabilitation outcomes from model systems in the United States: Spinal Cord 2007;45: DeVivo MJ, Vogel LC. Epidemiology of spinal cord injury in children and adolescents. J Spinal Cord Med 2004;27(suppl 1): S DeVivo MJ, Fine PR, Maetz HM, et al. Prevalence of spinal cord injury: a reestimation employing life table techniques. Arch Neurol 1980;37: Berkowitz M, Harvey C, Greene C, et al. The economic consequences of traumatic spinal cord injury. New York: Demos Publications; Harvey C, Rothschild BB, Asmann AJ, et al. New estimates of traumatic SCI prevalence: a survey-based approach. Paraplegia 1990;28: Lasfargues JE, Custis D, Morrone F, et al. A model for estimating spinal cord injury prevalence in the United States. Paraplegia 1995;33: O Connor PJ. Prevalence of spinal cord injury in Australia. Spinal Cord 2005;43: Dahlberg A, Kotila M, Leppanen P, et al. Prevalence of spinal cord injury in Helsinki. Spinal Cord 2005;43: DeVivo MJ, Go BK, Jackson AB. Overview of the national spinal cord injury statistical center database. J Spinal Cord Med 2002; 25: Stover SL, DeVivo MJ, Go BK. History, implementation, and current status of the National Spinal Cord Injury Database. Arch Phys Med Rehabil 1999;80:

7 338 SPINAL CORD INJURY EPIDEMIOLOGY, DeVivo 26. Marino RJ. International standards for neurological classification of spinal cord injury. J Spinal Cord Med 2003;26:S Frankel HL, Hancock DO, Hyslop G, et al. The value of postural reduction in the initial management of closed injuries of the spine with paraplegia and tetraplegia. I. Paraplegia 1969;7: DeVivo MJ, Stover SL, Black KJ. Prognostic factors for 12-year survival after spinal cord injury. Arch Phys Med Rehabil 1992; 73: McHorney CA, Ware JE, Jr, Raczek AE. The MOS 36-Item Short-Form Health Survey (SF-36): II. Psychometric and clinical tests of validity in measuring physical and mental health constructs. Med Care 1993;31: Ottenbacher KJ, Hsu Y, Granger CV, et al. The reliability of the functional independence measure: a quantitative review. Arch Phys Med Rehabil 1996;77: Whiteneck GG, Charlifue SW, Gerhart KA, et al. Quantifying handicap: a new measure of long-term rehabilitation outcomes [comment]. Arch Phys Med Rehabil 1992;73: Diener E, Emmons RA, Larsen RJ, et al. The satisfaction with life scale. J Pers Assess 1985;49: Welch BL. The generalisation of Student s problems when several different population variances are involved. Biometrika 1947;34: Levene H. Robust tests for equality of variance. In: Oldkin I, editor. Contributions to probability and statistics. Palo Alto: Stanford Univ Pr; p Strauss D, Shavelle R, DeVivo MJ, et al. An analytic method for longitudinal mortality studies. J Insur Med 2000;32: Pflaum C, McCollister G, Strauss DJ, et al. Worklife after traumatic spinal cord injury. J Spinal Cord Med 2006;29: Shavelle RM, DeVivo MJ, Strauss DJ, et al. Long-term survival of persons ventilator dependent after spinal cord injury. J Spinal Cord Med 2006;29: Whiteneck G, Tate D, Charlifue S. Predicting community reintegration after spinal cord injury from demographic and injury characteristics. Arch Phys Med Rehabil 1999;80: Charlifue S, Lammertse DP, Adkins RH. Aging with spinal cord injury: changes in selected health indices and life satisfaction. Arch Phys Med Rehabil 2004;85: Hall KM, Cohen ME, Wright J, et al. Characteristics of the Functional Independence Measure in traumatic spinal cord injury. Arch Phys Med Rehabil 1999;80: Go BK, DeVivo MJ, Richards JS. The epidemiology of spinal cord injury. In: Stover SL, DeLisa JA, Whiteneck GC, editors. Spinal cord injury: clinical outcomes from the model systems. Gaithersburg: Aspen; p DeVivo MJ, Underhill AT, Fine PR. Accuracy of world-wide-web death searches for persons with traumatic brain injury. Brain Inj 2004;18: Richards JS, Go BK, Rutt RD, et al. The national spinal cord injury collaborative database. In: Stover SL, DeLisa JA, Whiteneck GC, editors. Spinal cord injury: clinical outcomes from the model systems. Gaithersburg: Aspen; p Arias E. United States life tables, Vol. 56, No. 9. National vital statistics reports. Hyattsville: National Center for Health Statistics; Available at: nvsr56/nvsr56_09.pdf. Accessed December 16, 2010.