|Home | About | Journals | Submit | Contact Us | Français|
To identify sex-specific predictors of inpatient rehabilitation outcomes among patients with a traumatic brain injury (TBI) from a population based perspective.
Retrospective cohort study
Patients in inpatient rehabilitation for a TBI within one year of acute care discharge between 2008/09 and 2011/12 (N=1,730, 70% male, 30% female).
Inpatient rehabilitation length of stay, total Functional Independence Measure (FIM™) score, and motor and cognitive FIM™ ratings at discharge.
Sex, as a covariate in multivariable linear regression models, was not a significant predictor of rehabilitation outcomes. While many of the predictors examined were similar across males and females, sex-specific multivariable models identified some predictors of rehabilitation outcome that are specific for males and females; mechanism of injury (p<.0001) was a significant predictor of functional outcome only among females while comorbidities (p<.0001) was a significant predictor for males only.
Predictors of outcomes after inpatient rehabilitation differed by sex, providing evidence for a sex-specific approach in planning and resource allocation for inpatient rehabilitation services for patients with TBI.
Traumatic brain injury (TBI) is defined as an “insult to the brain that affects its structure or function, resulting in impairment of cognition, communication, physical function, or psychosocial behavior” . The indirect cost of TBI among Canadians aged 15 years and older is expected to increase from $7.3 billion in 2011 to $8.2 billion by 2031, far exceeding that of other common neurological conditions such as epilepsy (~$2.5 billion), multiple sclerosis (~$1.5 billion), and Alzheimer's disease and other dementias (<$1 billion) . In the United States, it was estimated that the direct and indirect medical cost of TBI in 2010 was $786.5 billion, with fatal and severe TBIs that require hospitalization accounting for 90% of the medical costs . Given the substantial cost of a TBI to the patient and the healthcare system, appropriate and adequate support is essential to improve outcome and reduce the demand on costly healthcare services.
Inpatient rehabilitation has long been identified as key in improving outcomes and prognosis after a TBI . Concurrently, rehabilitation has the potential to compensate the long-term costs associated with injury . Important to note is that findings have highlighted females as having worse outcome post-TBI [6-12] compared to their male counterparts, which may be attributable to differences in comorbidities [7, 13, 14], cause of injury [7, 15], or injury severity [7, 12, 13, 16]. Moreover, it is important to acknowledge that sex (i.e., biological construct) and gender (i.e., social constructs) are inter-related and influence health experiences . Specifically, research on sex/gender difference has resulted in evidence that “biological variances between males and females and gendered social norms result in distinct experiences of brain injury for men and women” . For example, female survivors of a TBI often report neuroendocrine dysfunction [18-20] that is known to result in hormonal changes caused by stress [18, 21, 22]. Among women survivors, fulfilling roles that fall within the women's unpaid labour such as motherhood, caregiving, and domestic roles, are often cited as sources of stress post TBI [18, 23, 24]. More important, this gendered nature of paid work has been found to “impact the likelihood of returning to work and of receiving sufficient benefits post-injury, which is dependent on lifetime cumulative earnings” . As such, a sex/gendered approach must be taken to elucidate the sex/gender specific factors affecting outcomes after TBI.
Nonetheless, a substantial gap in TBI research persists, in which sex-specific factors that predict inpatient rehabilitation outcomes have not been comprehensively studied. While this paucity of sex specific data, in particular for women with TBI, may be due to women's traditional inequality and disadvantages in access to and control of resources , it is increasingly recognized that there is a need to incorporate sex in research, analysis and interpretation of findings [26-28]. An awareness of how rehabilitation outcomes differ by sex is critical to support healthcare planning and clinical decision-making for this population. As such, the objective of this study was to identify sex-specific predictors of rehabilitation length of stay and functional outcome (total, motor, and cognition), with the goal of informing healthcare professionals and decision makers in their planning of rehabilitation services for males and females with a TBI. In particular, this information has implications for resource allocation such that it can provide information on the factors that can assist in reducing rehabilitation LOS and improving functional outcome.
Population based healthcare administrative data were used and obtained through the Ontario Cancer Data Linkage Program (‘cd-link’). The cd-link is an initiative of the Ontario Institute for Cancer Research/Cancer Care Ontario Health Services Research Program whereby risk reduced coded data from the Institute for Clinical Evaluative Sciences (ICES) Data Repository managed by ICES is provided directly to researchers with the protections of a comprehensive Data Use Agreement.
Acute care data were extracted from the Canadian Institute for Health Information Discharge Abstract Database (DAD), which contains demographic and clinical information on all acute care admissions and discharges, including transfers and deaths, in Canada . Data on inpatient rehabilitation were extracted from the National Rehabilitation Reporting System (NRS), which contains information on all clinical outcomes and rehabilitation activities from all inpatient rehabilitation beds within acute care hospitals and free-standing rehabilitation hospitals in Canada . There is mandatory reporting of acute care and inpatient rehabilitation data in Ontario, Canada, and residents of this province have universal access to healthcare. Thus, the DAD and NRS captures all patients in the province of Ontario that used acute care and/or inpatient rehabilitation services during the study period examined.
All patients admitted to inpatient rehabilitation for a TBI within one year of acute care discharge between fiscal years 2008/09 and 2011/12 in Ontario, Canada, were included in this study. Patients with TBI were identified using International Classification of Diseases Version 10 (ICD-10) codes (S02.0, S02.1, S02.3, S02.7, S02.8, S02.9, S07.1, S06, F07.2, T90.2, and T90.5). Specifically, patients with a TBI ICD-10 diagnostic code in the DAD were linked to the NRS via unique encoded identifiers to ensure that a patient was only captured once and that data on acute care and inpatient rehabilitation were matched to each patient.
The outcome variables examined were (1) inpatient rehabilitation length of stay (LOS); (2) total FIM™ score at discharge; (3) motor FIM™ rating at discharge; and (4) cognitive FIM™ rating at discharge.
Independent variables were predictors of interest that were first identified through previous research on TBI and rehabilitation [8, 31-35] and subsequently extracted from the DAD and the NRS, if available. These variables were categorized into three categories: factors that predispose (i.e., predisposing factors), create a need (i.e., need factors), and enable (i.e., enabling factors) health service use and affect outcomes, as suggested by the Andersen Behavioural Model of Health Service Utilization. This model has been widely used to assess the determinants of health service use .
Predisposing factors are those that predispose individuals to the use of health services  and included:
Need factors reflect the characteristics that create a need for the use of health services  and included:
Enabling factors are considered conditions that enable service utilization  and included:
Multivariable linear regressions were used to identify significant factors that predicted (1) inpatient rehabilitation LOS; (2) total FIM™ score at discharge; (3) motor FIM™ rating at discharge; and (4) cognitive FIM™ rating at discharge. Interaction variables for (1) sex and age and (2) sex and comorbidities were included in all four outcome models. These interaction terms were created by dummy coding the variable sex (referent category: males) and then multiplying this by the dummy coded variable age (referent category: adults <65 years of age) and by the number of ADGs, respectively. Sex-specific multivariable linear regression analyses were conducted to identify factors associated with rehabilitation outcomes for males and females separately.
To arrive at the final multivariable models presented in this paper, bivariate analyses were first conducted between each independent variable and outcome. All Andersen Behavioral variables were considered, with the exception of the AIS score because 20% of this cohort had an “unknown” AIS score (please see Table 1). As such, the AIS may not provide an accurate assessment of injury severity. Independent variables with a p-value of 0.2 or less in the bivariate analyses were included in the multivariable model building process. A p-value of <.05 was considered statistically significant and as such, all variables with p>.05 were removed, one at a time, starting with the highest p-value. Only statistically significant variables were included in the final multivariable linear regression models, with the R2 and adjusted R2 presented for each model.
Collinearity was assessed using a tolerance value of <.10, a variance inflation factor (VIF) value of >10, and condition index of >30. Heteroscedasticity was assessed using the Bruesch-Pagan Test and a visual examination of residual plots. Model specification was assessed using the Link Test. Finally, normality of residuals was assessed using the Q-Q plots of the residuals. To meet these criteria, the outcome variable rehabilitation LOS was log-transformed. Also, as the Breusch-Pagan Test suggests the presence of heteroscedasticity, heteroscedasticity consistent standard errors, 95% confidence intervals, and p-values were used. Finally, although the Link Test suggests additional independent variables are missing in our model, our efforts to include additional variables were limited by the availability of data elements in the DAD and NRS. This limitation is further elaborated in the discussion. All analyses were conducted using SAS 9.4.
This study was approved by the Toronto Rehabilitation Institute, University Health Network Research Ethics Board.
Between 2008/09 and 2011/12, there were 20,516 patients in acute care with a TBI diagnosis. Of these patients, 15.3% (n=3,137) were discharged to inpatient rehabilitation within one year of acute care discharge and among these patients, 56.7% (n=1,730) were admitted to inpatient rehabilitation for a TBI. As such, this paper comprised of a total 1,730 patients in inpatient rehabilitation for a TBI within 1-year of acute care discharge. By sex, 51.1% of female were older adults, compared to 32.2% of males (please see Table 1).
Sex, as a covariate, was not a significant predictor of rehabilitation LOS. However, sex-specific multivariable linear regressions showed that rehabilitation client grouping, admission delay, and informal support (not required vs. required and received) were predictors among males only while English language was a significant predictor among females only. Common predictors of rehabilitation LOS included older adults, acute care LOS, ALC days, FIM™ motor and cognitive rating, and injury via MVC (adjusted R2=40.5% and 33.3% for the final model for males and females, respectively). Please see Table 2.
Sex, as a covariate, was not a significant predictor of functional outcome at discharge. Increasing number of ADGs was associated with lower FIM scores at discharge among males only while injury via a MVC was associated with higher functional outcomes among females only (adjusted R2 = 56.4% to 61.1% and 51.7% to 65.1% for the final models for males and females, respectively). Please see Tables 3 to to55.
This population based study examined sex-specific predictors of inpatient rehabilitation outcomes among persons with TBI. The results provided evidence for sex differences in the demographics of this inpatient rehabilitation population. Consistent with trends seen in the United States , the current data indicates that the majority of females with TBI in inpatient rehabilitation were older adults. In Ontario, Canada, between fiscal years 2003/04 and 2005/06, 44.6% of patients with TBI discharged to inpatient rehabilitation were older adults aged 65 years and older  while the present study found that 51.1% of females were older adults. It has been projected that the number of Ontario residents aged 75 years and older will increase from 923,000 in 2013 to 2.7 million by 2041, with 18% more female than male . Concurrently, Ontario has seen a steady increase in the number and rate of TBI in the older adult population; between fiscal years 2003/04 and 2009/10, the rate of hospitalized TBI episodes of care among older adults increased by up to 63%, with the rate among patients aged 85 years and older 4.63 times the rate among patients between 65 and 74 years . Given that older adults also make significant gains in inpatient rehabilitation  and continue to recover and improve 6 months after inpatient rehabilitation , decision makers and healthcare professionals must allocate resources to and prepare for the increasing proportion of older adults, in particular older females, in the inpatient rehabilitation setting.
This study also built upon existing research by exploring sex-specific predictors of inpatient rehabilitation outcomes in the TBI population. Results showed that sex, as a covariate in the multivariable linear regression models, was not a significant predictor of rehabilitation outcomes. However, sex-specific predictors of rehabilitation outcomes were evident when data were stratified and analyzed by sex. For example, enabling factors did not significantly influence functional outcomes for males, however, the availability of supplemental health insurance, as suggested by injury in a MVC, was a significant predictor for females. Further, with the exception of higher cognitive FIM™ rating at admission, this study failed to identify factors significantly associated with higher cognitive outcome among males. However, injury in a MVC was associated with higher cognitive outcome among females, again proposing the role of supplemental health insurance in improving outcome . Despite the publicly funded nature of Canada's healthcare system, approximately 30% of healthcare spending among Canadians is covered by private insurance plans and out of pocket spending . Concurrently, research has provided evidence that having health insurance improves health and increases healthcare utilization . However, it is known that sex/gender, pay equity and income, casual and temporary work have implications on the availability of insurance and benefits and are very much inter-related; research showed that women are significantly less likely to be engaged in full-time employment [18, 48, 49] and have a significantly lower average pre-injury income . There is also evidence that the ability to mobilize resources among higher income individuals plays a crucial role in disease prevention and treatment . These factors can impact the availability of benefits post-injury. Thus, the finding that the availability of supplemental health insurance was associated with higher functional outcome among females, and not males, may reflect the gendered nature of paid work and its implications on pay inequity, employment opportunities and thus, availability of insurance and associated benefits. While this paper supports the finding that health insurance was significantly associated with higher functional outcome, the data sources for this study do not provide us with information on employment, income, and availability of benefits and insurance. Colantonio and colleagues have found that having other payer sources, including workers’ compensation or private insurance, was significantly associated with shorter wait times for inpatient rehabilitation, suggesting that the availability of other payer sources may “facilitate better case management, facilitating inpatient rehabilitation” . As such, future studies, in particular qualitative studies, that strive to elucidate the influence of insurance and support on health outcomes is important. This can provide policy-makers with evidence that can guide efforts to promote health for both men and women and understand how best to improve access to health service for this population. In particular, efforts to improve outcomes in inpatient rehabilitation for females should focus on enabling factors such as improving access to permanent work options that provide stability and health coverage.
This study also identified sex-specific enabling factors that predicted rehabilitation LOS. Of particular interest is the finding that, among females, English language was significantly associated with shorter rehabilitation LOS. Interesting, the variable ‘English language’ was not a significant predictor of functional outcome, suggesting that current rehabilitation treatments may be adequately meeting the healthcare needs of patients with TBI, regardless of language proficiency. This has been shown in research on hospital LOS and mortality, in which English proficiency did not predict in-hospital mortality while patients with limited English proficiency had longer hospital LOS . It has also been suggested that delivery of health care may be negatively altered when a patient has limited English proficiency , with a lower quality of healthcare perceived by these patients [53, 54]. The finding that English proficiency may also significantly predict LOS in the rehabilitation setting among female patients is of particular importance in the Canadian context, given the linguistic diversity of this nation. The 2011 Census of Population found that 14.2% of the Canadian population reported “speaking a language other than English or French most often at home” . Concurrently, there has been an increase in the number of immigrants to Canada, the majority of whom were females . As such, research into the reasons why English proficiency may impact rehabilitation LOS may provide an opportunity to improve healthcare for females with TBI.
Finally, data from this study indicate that there are sex-specific need factors that influence rehabilitation outcomes. In particular, comorbidities did not predict functional outcomes among females, yet, each increase in the number of ADGs was associated with lower motor and cognitive FIM™ ratings at discharge. Most data to date have indicated that comorbidities influence outcomes [7, 13, 14], however, this paper suggests this may not hold true for females. Given the higher risk for a TBI among males [42, 57, 58], current data on comorbidities and their influence on outcomes are likely driven by this male-dominant profile and thus, may not be transferable to the female population. This speaks to the significance of sex-specific research on comorbidities in the TBI population. Future research should also explore the influence of specific types of comorbidities on rehabilitation outcomes. As our TBI inpatient rehabilitation population shifts from younger males to older females, the types of comorbidities that present with a TBI in this healthcare setting may also change. A multidisciplinary approach is crucial in the treatment of a TBI, including guidelines on how best to incorporate and manage comorbidities for males and females specifically in the inpatient rehabilitation setting to maximize outcome.
Limitations of data availability in the data sources must be recognized, including the unavailability of variables of interest. This limitation is reflected in the Link Test, which suggested additional independent variables are missing in our models. For example, research has shown that marital status and race/ethnicity [8, 12, 59] are important factors to include, however, they are unavailable in the DAD and NRS. As such, this paper used the variable “English language” as a proxy for race/ethnicity to capture diversity in Canada. Further, AIS was used to assess injury severity, however, this was limited by the high proportion of “unknown” cases due to the absence of detailed ICD-10 diagnostic codes. Similarly, even though the DAD contains information on the Glasgow Coma Scale (GCS), data completion rates are low  and as such, it was not feasible to include this variable in multivariable linear regression models presented in this paper. The GCS is an important predictor to control for the severity of injury and may account for the remaining variance in these models. Given the increasing use of healthcare administrative data for population based research, education on the importance of coding accuracy and specificity is encouraged. The acute care LOS and use of intensive care units were included as additional proxies for injury severity in this paper. Finally, age groups in this paper were coded broadly as 0 – 19 years, 20 – 64 years, and 65+ years, which may differ from published literature on rehabilitation outcomes after TBI; comparisons of findings should be made with this in mind.
Nonetheless, there is mandatory reporting of acute care and inpatient rehabilitation data in Ontario, Canada. Thus, all TBI patients admitted to inpatient rehabilitation for a TBI within one year of acute care discharge between fiscal years 2008/09 and 2011/12 were included in this study. Further, the linkage of the DAD with the NRS allowed for the extraction of additional variables of interest such ALC days, use of intensive care units, and acute care LOS. This is also the first study, to the best of our knowledge, to provide sex-specific information on inpatient rehabilitation outcomes among patients with TBI.
Although sex, as a covariate in multivariable linear regression models, was not a significant predictor of rehabilitation outcomes, there are sex-specific factors that were significantly associated with rehabilitation outcomes. Comorbidities significantly predicted functional outcomes among males only while injury in a motor vehicle collision, which suggests the presence of supplemental health insurance, significantly predicted functional outcome among females only. The present data provide evidence to a support sex-specific approach in research and healthcare planning to improve rehabilitation outcomes in the TBI population. In particular, efforts to improve inpatient rehabilitation outcomes must work to understand and identify comorbidities that are present among males that influence rehabilitation outcomes and how one can improve access to financial support and benefits for females. Health service planning and resource allocation must take into account sex differences, particularly as the TBI inpatient rehabilitation demographics shift from younger males to older females.
This study was supported through the provision of data by the Institute for Clinical Evaluative Sciences (ICES) and Cancer Care Ontario (CCO) and through funding support to ICES from an annual grant by the Ministry of Health and Long-Term Care (MOHLTC) and the Ontario Institute for Cancer Research (OICR). The opinions, results, and conclusions reported in this paper are those of the authors. No endorsement by ICES, CCO, OICR, or the Government of Ontario is intended or should be inferred. Parts of this material are based on data and information compiled and provided by CIHI. However, the analyses, conclusions, opinions, and statements expressed herein are those of the author, and not necessarily those of CIHI. Research reported in this publication was supported by the Eunice Kennedy Shriver National Institute of Child Health & Human Development of the National Institutes of Health under Award Number R24HD065702 (PI: Dr. Kenneth Ottenbacher). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Dr. Angela Colantonio received support through the Canadian Institutes of Health Research (CIHR) Chair in Gender, Work and Health (#CGW-126580). Vincy Chan received support from the CIHR and the Pediatric Oncology Group of Ontario for a Doctoral Research Award, Brain Canada and CIBC for a Brain Cancer Training Award, and the Jane Gillett Pediatric ABI Studentship from the Ontario Neurotrauma Foundation. Dr. Tatyana Mollayeva received support through the Frederick Banting and Charles Best Canada Gradate Scholarships Doctoral Award from the CIHR.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.