Secondary Logo

US Population Estimates of Health and Social Outcomes 5 Years After Rehabilitation for Traumatic Brain Injury

Corrigan, John D. PhD; Cuthbert, Jeffrey P. PhD, MPH, MS; Harrison-Felix, Cynthia PhD; Whiteneck, Gale G. PhD; Bell, Jeneita M. MD, MPH; Miller, A. Cate PhD; Coronado, Victor G. MD, MPH; Pretz, Christopher R. PhD

Section Editor(s): Dijkers, Marcel P. PhD; Gordon, Wayne A. PhD

Journal of Head Trauma Rehabilitation: November/December 2014 - Volume 29 - Issue 6 - p E1–E9
doi: 10.1097/HTR.0000000000000020
Original Articles
Free

Objective: To estimate the number of adults in the United States from 2006 to 2012 who manifest selected health and social outcomes 5 years following a traumatic brain injury (TBI) that required acute inpatient rehabilitation.

Design: Secondary data analysis.

Setting: Acute inpatient rehabilitation facilities.

Participants: Patients 16 years and older receiving acute inpatient rehabilitation for a primary diagnosis of TBI.

Main Outcome Measures: Mortality, functional independence, societal participation, subjective well-being, and global outcome.

Results: Annually from 2001 to 2007, an average of 13 700 patients aged 16 years or older received acute inpatient rehabilitation in the United States with a primary diagnosis of TBI. Approximately 1 in 5 patients had died by the 5-year postinjury assessment. Among survivors, 12% were institutionalized and 50% had been rehospitalized at least once. Approximately one-third of patients were not independent in everyday activities. Twenty-nine percent were dissatisfied with life, with 8% reporting markedly depressed mood. Fifty-seven percent were moderately or severely disabled overall, with 39% having deteriorated from a global outcome attained 1 or 2 years postinjury. Of those employed preinjury, 55% were unemployed. Poorer medical, functional, and participation outcomes were associated with, but not limited to, older age. Younger age groups had poorer mental and emotional outcomes. Deterioration in global outcome was common and not age-related.

Conclusions: Significant mortality and morbidity were evident at 5 years postinjury. The deterioration in global outcomes observed regardless of age suggests that multiple influences contribute to poorer outcomes. Public health interventions intended to reduce post–acute inpatient rehabilitation mortality and morbidity rates will need to be multifaceted and age-specific.

Department of Physical Medicine and Rehabilitation, Ohio State University, Columbus Ohio (Dr Corrigan); Research Department, Craig Hospital, Englewood, Colorado (Drs Cuthbert, Harrison-Felix, Whiteneck, and Pretz); National Center for Injury Prevention and Control, Centers for Disease Control and Prevention, Atlanta, Georgia (Drs Bell and Coronado); and Department of Education, National Institute on Disability and Rehabilitation Research, Washington, District of Columbia (Dr Miller).

Corresponding Author: John D. Corrigan, PhD, Department of Physical Medicine and Rehabilitation, Ohio State University, 480 Medical Center Dr, Columbus, OH 43210 (corrigan.1@osu.edu).

This research was supported by an interagency agreement between the US Department of Health and Human Services (HHS), Centers for Disease Control and Prevention (CDC), and the US Department of Education, Office of Special Education and Rehabilitative Services, National Institute on Disability and Rehabilitation Research (NIDRR), with supplemental funding to the NIDRR-funded Traumatic Brain Injury (TBI) Model Systems National Data and Statistical Center (grant no. H133A110006). The study was also supported by a TBI Model Systems Centers grant from the NIDRR to Ohio State University (grant no. H133A120086). This article does not reflect the official policy or opinions of the CDC or HHS and does not constitute an endorsement of the individuals or their programs—by the CDC, HHS, or other components of the federal government—and none should be inferred. No commercial party having a direct financial interest in the results of the research supporting this article has or will confer a benefit upon the authors or upon any organization with which the authors are associated. The TBI Model Systems National Database is supported by the NIDRR and created and maintained by the TBI Model Systems Centers program. This article is intended to promote the exchange of ideas among researchers and policy makers. The views expressed in it are part of ongoing research and analysis and do not necessarily reflect the position of the US Department of Education.

The authors declare no conflicts of interest.

THE CENTERS FOR DISEASE CONTROL and Prevention estimates that persons 16 years and older account for 90% of annual hospital admissions for traumatic brain injury (TBI).1 Adolescents and adults with injuries of sufficient severity that require rehabilitation have an elevated risk for premature mortality, persistent cognitive impairment, and reduced participation in society.2–4 While longitudinal studies have provided evidence of these long-term consequences from moderate and severe TBIs, the samples studied did not allow estimates of the prevalence of mortality and morbidity in the US population.

Recent publications by Corrigan and colleagues5 and Cuthbert and colleagues6 have provided a basis for making population estimates of persons 16 years and older in the United States who receive inpatient rehabilitation for a primary diagnosis of TBI. The Uniform Data System for Medical Rehabilitation7 and the American Medical Rehabilitation Providers Association's eRehabData8 serve as intermediaries between inpatient rehabilitation facilities and the Centers for Medicare & Medicaid Services. These organizations receive information on all patients treated in a rehabilitation facility (whether or not the patient has Medicare funding) and forward these data to the Centers for Medicare & Medicaid Services. Data from as much as 92% of all inpatient rehabilitation facilities, and a still higher percentage of all patients, are collected and reported by these 2 programs.5 Corrigan and colleagues5 (for the years 2001-2007) and Cuthbert and colleagues6 (for the years 2001-2010) combined core data sets from the 2 intermediaries and compared the resulting US population characteristics to the cohort enrolled in the TBI Model Systems National Database (TBIMS-NDB) for the same time periods. Both sets of analyses found overwhelming similarity between the US population and the TBIMS cohorts after accounting for differences in the proportion of older adults. Among multiple individual differences and injury characteristics compared, important discrepancies were noted only for 1- to 9-day rehabilitation lengths of stay for both younger and older patients, and primary insurance and the proportion in the oldest age categories for the 65 years and older group. The minimal number of differences led to a common conclusion in both studies: with weighting, the TBIMS-NDB can be used to make population estimates for persons older than 15 years who received acute inpatient rehabilitation for a primary diagnosis of TBI in the United States.5,6

While population characteristics can be directly extracted from the combination of the Uniform Data System for Medical Rehabilitation and eRehabData data sets, only a limited number of variables are available, and no information is collected beyond rehabilitation discharge. The ability to extrapolate population estimates from the TBIMS-NDB allows a richer array of characteristics to be estimated, including outcomes postinjury at 1, 2, 5, and every 5 years thereafter. The TBIMS-NDB includes medical record–verified cases of moderate and severe TBIs admitted for inpatient rehabilitation at a TBIMS center. Rigorous data collection procedures have resulted in minimal missing data and high frequency of successful follow-up (≥79%) up to 20 years postinjury.9

This study used the Corrigan and colleagues 2001-2007 compilation of the US population receiving rehabilitation for TBI to weight TBIMS-NDB cases between October 1, 2001, and December 31, 2007, who were followed up at 5 years postinjury. Selected health and social outcomes representing multiple domains of the International Classification of Functioning, Disability and Health10 were estimated for the prevalence among those 16 years and older who received acute inpatient rehabilitation for a primary diagnosis of TBI in the United States between October 1, 2001, and December 31, 2007.

Back to Top | Article Outline

METHODS

The US Department of Education, National Institute on Disability and Rehabilitation Research has funded the TBIMS program since 1987. Central to this program is a standardized, longitudinal database for studies of TBI treatment and outcomes.11 The TBIMS-NDB contains cases treated by 20 TBIMS centers funded between 1988 and 2012.11 In the TBIMS-NDB, TBI is defined as damage to brain tissue caused by an external mechanical force as evidenced either by medically documented loss of consciousness or posttraumatic amnesia due to brain trauma or by objective neurological findings that can be reasonably attributed to TBI on physical examination or mental status examination. In addition to having incurred a TBI meeting the aforementioned definition, additional inclusion criteria are as follows: (1) meet at least 1 of the following criteria for TBI severity indictors: posttraumatic amnesia greater than 24 hours, trauma-related intracranial neuroimaging abnormalities, loss of consciousness exceeding 30 minutes, a Glasgow Coma Scale score in the emergency department of less than 13 (unless due to intubation, sedation, or intoxication); (2) be at least 16 years of age at the time of injury; (3) present to an acute care hospital within 72 hours of injury; (4) receive both acute hospital care and comprehensive rehabilitation in a designated brain injury inpatient unit within the TBIMS; and (5) provide informed consent to participate or have a proxy provide consent.9 Data are recorded during inpatient rehabilitation and subsequently during follow-up interviews at 1, 2, and 5 years postinjury and every 5 years thereafter. The TBIMS data collection protocol uses a “best source” policy for interviewing a proxy when the individual is not able to provide valid information.9 Measures that require reporting of a subjective state (eg, life satisfaction, emotional state) are collected only from the individual with a TBI. As of June 30, 2012, the TBIMS-NDB contained 11 247 cases and had a follow-up rate of 79% for follow-up interviews conducted up to 20 years postinjury; 80% for just those at 5 years.

For the purpose of this study, the TBIMS-NDB was limited to patients admitted and discharged between October 1, 2001, and December 31, 2007, resulting in 4838 cases; however, at the time of analysis, not all 5-year follow-up interview windows for these cases had closed (5-year interviews are required to be completed between 6 months before and 6 months after TBI anniversary). Cases that remained open as of April 1, 2012, regardless of completion status, were excluded from the analyses, resulting in a final total sample of 4064 cases. These cases were then weighted using raking, an iterative proportional weighting procedure that repeats until a sample distribution closely matches that of a larger population across numerous population characteristics simultaneously.12 In this case, sample distributions were the characteristics of the TBIMS-NDB, and the population values were the categorical characteristics of the US TBI Rehabilitation population included in the article by Corrigan and colleagues.5 Variables included as part of this weighting procedure, including the categorical distributions used, are shown in Table 1. One requirement of raking is that population proportions must be 5% or more; thus, categorizations of the US TBI rehabilitation population characteristics found to be below this threshold were combined with similar categories. Within the TBIMS-NDB, missing values for variables were rare (<10%); however, those individuals with missing data across any of these variables had the missing value imputed using expectation maximization13 prior to the implementation of the raking procedure. The final US population resulting from the inclusion criteria and weighting processes are shown in Figure 1.

Figure 1

Figure 1

TABLE 1

TABLE 1

At the completion of raking, all case weights were evaluated using weight trimming, a process in which weights of extreme values are truncated to less extreme values (eg, 1%-5%), with the intent of reducing the mean squared error for outcomes of interest. For these analyses, weight trimming was not found to reduce the mean squared error for any of the health and social outcomes; thus, the weights were applied in their original configuration. All raking and weight trimming analyses were completed with SAS (version 9.3), using the IHB raking macro.14 Descriptive statistics were computed for the weighted data using SPSS (version 19). Means, medians, standard deviations, and range were used for continuous variables, and proportions were used for categorical variables.

Initially, all variables available from the TBIMS-NDB interview conducted at the 5-year follow-up were reviewed for their ability to capture health and social functions with potentially important public health implications. Fifteen variables for which objective cutoffs could be defined for desirable versus undesirable outcomes were retained (see Table 2 for variables and definitions of undesirable levels). Some available outcomes were eliminated because of low incidence (eg, incarceration, subsequent TBI). While it would have been preferable to include indices of societal participation based on PART subscale scores (“productivity,” “social relationships,” and “out and about”),15 normative data are not yet developed enough to permit confident definition of cutoff points. Instead, indicators not inherently pathognomonic but that may reflect greater challenges for successful societal participation were included. There are other ways to be productive without being competitively employed; therefore, unemployment was examined only among those employed preinjury. Being unmarried does not equate with inadequate social relationships; however, it does indicate that greater effort may be needed to maintain social networks. Not being able to drive oneself may not mean one does not get out and about in the community; however, it does imply greater dependence on others (including public transportation) to do so.

TABLE 2

TABLE 2

The weighted data were initially reviewed for integrity by examining the 5-year interview follow-up composition. Once weighted, the TBIMS-NDB was found to be 83.6% complete, with a known outcome, 10.7% lost to follow-up, and 5.7% withdrew, refused, or unknown. Of the cases with a known outcome, approximately one-fourth (24.7%) were found to be deceased at 5 years. Because a large portion of the data for most other outcomes would be missing at 5 years (37.1%, primarily due to 20.6% expired), the data were reweighted using only those cases alive at 5 years with known outcomes (unweighted n = 2556). While this reweighting may have introduced nonresponder bias by excluding cases that were lost or refused, it was deemed essential to ensure that the preponderance of estimates produced were not missing, particularly “missing due to death.” Thus, estimates (excluding those expired) are for patients with TBI who are known to have survived 5 years postinjury.

Back to Top | Article Outline

RESULTS

Table 2 shows the total population estimate and breakdown by age at injury for each of the 15 health and social outcomes. Definitions of the cutoff points used for each dichotomous comparison are also shown in Table 2. Confidence intervals for proportions are not presented, as the confidence intervals obtained from the large sample size are relatively small: in the range of ±0.43% for proportions around 1%, ±1.16% for proportions around 10%, and ±1.89% for proportions around 50%. Note that all references to age are for the patient's age at injury, not the age at follow-up.

The 5-year estimate of those who died was 20.6%, with those youngest at injury (16-19 years) the least likely to be deceased 5 years postinjury (2%), increasing to 56% for those 80 years and older at injury (see Figure 2). For those who were alive at 5 years postinjury, the estimates for institutionalization (12.2%) followed a similarly marked pattern of increasing frequency with increasing age. Rehospitalizations within 5 years postinjury had an overall estimate of 49.7% and showed an increase with age; however, even for those in the youngest age group, a substantial percentage had been rehospitalized (38.5%).

Figure 2

Figure 2

Activity limitations as measured by the Functional Independence Measure (FIM) Motor and Cognitive subscales7 and the Supervision Rating Scale16 showed that approximately one-third of all persons required some form of assistance or supervision from another person (30.1% and 36.1% for FIM Motor and Cognitive, respectively) and 33.4% received supervision overnight and part of their waking hours. Of these variables, the need for supervision and FIM Cognitive subscale demonstrated the clearest trends for incremental increases in the proportion needing assistance or supervision with advancing age at injury. The FIM Motor subscale also revealed a general pattern of increasing proportion needing assistance or supervision as age increased for those 50 years and older at injury (see Figure 3).

Figure 3

Figure 3

Two of the indices reflecting societal participation—inability to drive oneself and unemployment (if employed preinjury)—demonstrated an increasing proportion with increasing age (see driving in Figure 4). The proportion being unmarried (see Figure 5) declined from early adulthood to age 60 to 69 years at injury, increasing markedly for individuals who were older at injury. Overall, almost two-thirds of the population was unmarried.

Figure 4

Figure 4

Figure 5

Figure 5

Prevalence trends for indices of subjective well-being showed varying patterns of age and outcomes. Figure 6 shows the trend of dissatisfaction with life (as indicated by a score <21 on the Satisfaction With Life Scale17) 5 years postinjury, in which those aged 40 to 49 years were most likely to be dissatisfied (48.6%) with rates falling as age either increased (8.1% for those 80 years and older) or decreased (28.4% for those aged 16-19 years). Moderate or severe depressed mood on the Patient Health Questionnaire18 at 5 years showed a similar pattern, with those aged 40 to 49 years having the highest prevalence of depressed mood (15.4%). Rates for depressed mood were lower as age increased and decreased from 40-49 years. For alcohol misuse, a markedly different relation with age was observed, with patients aged 16 to 19 years and 20 to 29 years having the highest prevalence rates (36.8% and 37.7%, respectively) and decreasing rates as age increased. Illicit drug use 5 years postinjury showed a similar pattern, although a lower rate. Those aged 16 to 19 years at injury were most likely to use illicit drugs (31.3%), with prevalence decreasing as age increased.

Figure 6

Figure 6

Five years postinjury, 57% of the population was classified as moderately disabled or worse on the Glasgow Outcome Scale–Extended.19 Those aged 40 to 49 years at injury had the highest rate (67.1%), with rates declining as age at injury decreased or increased until 80 years and older when poor global outcome increased. Deterioration was operationalized as anyone whose 5-year Glasgow Outcome Scale–Extended score was at least 1 category lower than their status on the same measure 1 or 2 years postinjury. Overall, 39.1% declined from an earlier attained level, with a relatively consistent rate across all age groups (see Figure 7).

Figure 7

Figure 7

Back to Top | Article Outline

DISCUSSION

The purpose of this study was to estimate selected health and social outcomes for adults in the United States who are 5 years post-TBI requiring rehabilitation. These estimates have not been possible previously because population data, themselves only available since 2001, have been limited to a relatively small set of descriptors collected at the time of rehabilitation. The ability to weight the TBIMS-NDB so it reflects the US population is a new development that provides unique insight into the public health burden of these injuries.

The findings reported here raise concerns about both mortality and morbidity 5 years post-TBI. Approximately 1 in 5 of those who received rehabilitation (1 in 4 of the known outcomes) had expired by 5 years postinjury. There was a marked age effect for death, as might be expected, and while we did not compare the mortality rate to age-matched controls, other studies of deaths among the TBIMS-NDB cohort have found that all age groups except the very oldest—those 80 years and older—had standardized mortality rates exceeding those expected in the general population.20,21 Identifying factors beyond age that predispose these patients to premature mortality remains a pressing public health concern.

For those alive 5 years postinjury, the current findings indicate that there is a marked interaction between age at injury and prevalence of undesirable health outcomes, particularly institutionalization and, to a lesser degree, rehospitalization. Measures of functional independence (eg, the FIM Motor, FIM Cognitive, and need for supervision) were associated with increasing age, as were the ability to drive and return to work. Even though older persons experienced the worst outcomes, for some variables, the rates occurring for younger age groups were still relatively high (eg, rehospitalization, FIM Cognitive, driving, and return to work). This high burden of injury translates into significant societal costs due to the length of time for which lost earnings and lost opportunity costs accrue.

The relatively higher rates of diminished health and functional dependence among older age groups are also observed in the general population; the extent to which prevalence among those who received TBI rehabilitation exceeded normal aging was beyond the scope of this study. At a minimum, for those age-related outcomes that showed high rates among younger age groups, there is a substantial likelihood that negative consequences exceeded general population rates. Whether or not unfavorable outcomes exceed population rates, high prevalence still suggests a public health burden.

Not all negative consequences reflected an interaction between older age and poorer outcome. Poorer subjective states, specifically life dissatisfaction and depression, were most common among those in the middle age groups (30-59 years), followed by those younger at injury. Older age groups had higher rates of positive subjective states as age at injury increased. Alcohol misuse and illicit drug use were most common in the youngest age at injury groups. The observation that younger people were more likely to misuse substances is consistent with prevalence in the general population, although the rate is lower than that for same-aged peers.22

It is interesting that those aged 30 to 59 years at injury experience the greatest life dissatisfaction and depression when persons older at injury have the worst outcomes for health status and functional independence. The middle age groups also showed the highest prevalence of poor global outcome, matched only by those 80 years and older at injury. This concurrence may not only explain the greater dissatisfaction but also raises the question why worse global outcome did not mimic the linear relation between increasing age and prevalence of poorer outcomes. This discrepancy between global outcome and both health status and functional dependence may suggest that there are multiple influences predisposing these individuals to negative consequences. Perhaps, cognitive and social consequences are more limiting for those aged 30 to 59 years at injury than for the younger age groups. In older age groups, the social support provided by marriage, adult children and long-established social networks may serve as a protective factor from negative psychosocial outcomes.

Also inconsistent with the relations observed between age at injury and prevalence of negative consequences was the lack of relation between age and the occurrence of deterioration from a previously attained global outcome. An earlier report of 10-year outcomes for the TBIMS cohort noted that older age at injury was associated with greater likelihood of deterioration.23 In the current study, deterioration was for the most part equally distributed across age at injury groups. Determining whether causes of deterioration are also equally distributed, or whether different age groups deteriorate for different reasons, was beyond the scope of the current study and will await further investigation. However, there are multiple potential sources of deterioration that could contribute to poorer outcomes, including:

  • TBI triggers a progressive, degenerative process (ie, Parkinson disease, Alzheimer disease, chronic traumatic encephalopathy);
  • frontal lobe damage endemic to TBI causes changes in self-regulation, which lead to death and disability from risky behaviors;
  • TBI causes loss of functional independence, which interacts with aging, either normal aging or other chronic medical conditions, to increase poor health; and/or
  • TBI exacerbates social factors (eg, lack of social support or environmental stressors), including health disparities (eg, lack of resources for additional rehabilitation or poor access to follow-up healthcare) to result in poorer health.

The 2012 Galveston Brain Injury Conference focused on long-term consequences and arrived at the following consensus statement:

Injury to the brain can evolve into a lifelong health condition termed chronic brain injury (CBI). CBI impairs the brain and other organ systems and may persist or progress over an individual's life span. CBI must be identified and proactively managed as a lifelong condition to improve health, independent function and participation in society.24(p6)

The current study underscores the concerns expressed in this statement by demonstrating that at least among those requiring rehabilitation for TBI, both premature death and deterioration may occur as early as 5 years postinjury. These findings point to the need for further research to identify causes for deterioration so that preventive strategies can be proposed and tested.

There are several potential limitations to this study, although foremost is the caution necessary when population estimates are derived from weighting cohort data. For instance, while the TBIMS-NDB data were weighted to reflect demographic characteristics and indices of functional severity (ie, admission FIM Motor and FIM Cognitive, rehabilitation length of stay), injury severity characteristics were not available for weighting. It is possible that older adults enter rehabilitation for TBI due to consequences that are an interactive function of both the brain injury severity and other comorbid conditions. The broader US population of older adults in rehabilitation for TBI may have a different “mix” of brain injury versus other etiologies contributing to their need for rehabilitation than those older adults included in the TBIMS NDB, which could result in different estimates of 5-year outcomes. In addition, the rates reported were among those living; thus, the estimates of the prevalence of negative consequences are attenuated because of “survivor bias.” Given that a large proportion of those who incurred a TBI expired by the 5-year follow-up interview and that weighting was recalculated to exclude known cases of expiration, the negative effects of time post-TBI may well be underestimated. The cohort used to calculate the outcome analyses could have some inherent characteristic that enhanced their chance of survival, and as such the estimates presented here would be inflated by this form of selection bias.25 As such, these estimates should be applied cautiously and within the appropriate context.

Back to Top | Article Outline

CONCLUSIONS

The current findings suggest that significant mortality and morbidity occur by 5 years post-TBI in patients who have received rehabilitation. Poorer medical, functional, and participation outcomes were associated with, but not limited to, older age. Other age groups had poorer mental and emotional outcomes, including life dissatisfaction. Both the high rate of premature mortality and the deterioration observed for global outcome may suggest that there are multiple influences affecting poorer health and function, underscoring the need for proactive interventions to be available after the acute phase of recovery.

Back to Top | Article Outline

REFERENCES

1. National Center for Injury Prevention and Control, Centers for Disease Control and Prevention. National Hospital Discharge Survey. Atlanta, GA: Centers for Disease Control and Prevention; 2010.
2. Dikmen SS, Corrigan JD, Levin HS, Machamer J, Stiers W, Weisskopf MG. Cognitive outcome following traumatic brain injury. J Head Trauma Rehabil. 2009;24(6):430–438.
3. Rutherford GW, Wlodarczyk RC. Distant sequelae of traumatic brain injury: premature mortality and intracranial neoplasms. J Head Trauma Rehabil. 2009;24(6):468–474.
4. Temkin NR, Corrigan JD, Dikmen SS, Machamer J. Social functioning after traumatic brain injury. J Head Trauma Rehabil. 2009;24(6):460–467.
5. Corrigan JD, Cuthbert JP, Whiteneck GG, et al. Representativeness of the traumatic brain injury model systems national database. J Head Trauma Rehabil. 2012;27(6):391–403.
6. Cuthbert J, Corrigan JD, Whiteneck GG, et al. Extension of the representativeness of the traumatic brain injury model systems national database: 2001 to 2010. J Head Trauma Rehabil. 2012;27(6):E15–E27.
7. UDS. Uniform Data System for Medical Rehabilitation. http://www.udsmr.org. Updated 2013. Accessed May 22, 2013.
8. American Medical Rehabilitation Providers Association. eRehabData.com. https://wwwv.erehabdata.com. Updated 2013. Accessed May 22, 2013.
9. TBI Model Systems National Data and Statistical Center. Traumatic brain injury model systems national database. http://www.tbindsc.org. Updated 2013. Accessed February 9, 2013.
10. World Health Organization. International Classification of Functioning, Disability, and Health. Geneva, Switzerland: World Health Organization; 2001.
11. Dijkers MP, Harrison-Felix C, Marwitz JH. The traumatic brain injury model systems: history and contributions to clinical service and research. J Head Trauma Rehabil. 2010;25(2):81–91.
12. Kalton G, Florres-Cervantes I. Weighting methods. J Off Stat. 2003;19(2):81–97.
13. Dempster AP, Lair NM, Rubin DB. Maximum likelihood from incomplete data via the EM algorithm. J R Stat Soc. 1977;39(1):1–38.
14. Izrael D, Hoaglin DC, Battaglia MP. A SAS macro for balancing a weighted sample. Paper presented at: Proceedings of the Twenty-fifth Annual SAS Users Group International Conference; 2000; Indianapolis, IN, Paper 258.
15. Bogner JA, Whiteneck GG, Corrigan JD, Lai JS, Dijkers MP, Heinemann AW. Comparison of scoring methods for the participation assessment with recombined tools-objective. Arch Phys Med Rehabil. 2011;92(4):552–563.
16. Boake C. Supervision Rating Scale: a measure of functional outcome from brain injury. Arch Phys Med Rehabil. 1996;77(8):765–772.
17. Pavot W, Diener E. Review of the satisfaction with life scale. Psychol Assess. 1993;5:164–172.
18. Kroenke K, Spitzer RL, Williams JB. The PHQ-9: validity of a brief depression severity measure. J Gen Intern Med. 2001;16(9):606–613.
19. Wilson JT, Pettigrew LE, Teasdale GM. Structured interviews for the Glasgow Outcome Scale and the extended Glasgow Outcome Scale: guidelines for their use. J Neurotrauma. 1998;15(8):573–585.
20. Harrison-Felix C, Kreider SE, Arango-Lasprilla JC, et al. Life expectancy following rehabilitation: a NIDRR traumatic brain injury model systems study. J Head Trauma Rehabil. 2012;27(6):E69–E80.
21. Harrison-Felix C, Kolakowsky-Hayner SA, Hammond FM, et al. Mortality after surviving traumatic brain injury: risks based on age groups. J Head Trauma Rehabil. 2012;27(6):E45–E56.
22. Substance Abuse and Mental Health Services Administration. Results From the 2008 National Survey on Drug Use and Health: National Findings. Rockville, MD: Office of Applied Studies; 2009. NSDUH Series H-36, HHS Publication No. SMA 09-4434.
23. Kolakowsky-Hayner SA, Hammond FM, Wright J, et al. Ageing and traumatic brain injury: age, decline in function and level of assistance over the first 10 years post-injury. Brain Inj. 2012;26(11):1328–1337.
24. Hammond FM, Malec JF. Rethinking brain injury. Brain Inj Prof. 2013;10(1):6–10.
25. Szklo M, Nieto F. Epidemiology: Beyond the Basics. 2nd ed. Sudbury, MA: Jones & Bartlett; 2007.
Keywords:

craniocerebral trauma; epidemiology; outcomes; prevalence; rehabilitation; traumatic brain injury

© 2014 Wolters Kluwer Health | Lippincott Williams & Wilkins