Traumatic brain injury (TBI) is a major cause of mortality in Canada, contributing to approximately 23% of all injury-related deaths (1). TBI is also related to many adverse health outcomes, (2, 3) such as physical and mental health conditions, (4–6) cognitive impairment, (7) suicidality, (8) substance use, (9) victimization, (10) increased mortality, (11) increased health care utilization, (10) and incarceration (12). In addition to its impact on patients' health, TBI is also immensely burdensome on the health care system.
Individuals who sustain TBI have a higher burden of comorbid illness which results in increased health care utilization.(10, 13) Mental health disorders are one of the common sequelae of TBI; in the first-year post-injury, up to 77% receive a psychiatric diagnosis: commonly, anxiety, mood and substance-use disorders which often present co-morbidly (14, 15). The high rates of mental health disorders are associated with worse health outcomes, increased mental health care utilization, and poorer quality of life. Studies using Anderson’s Behavioral Model (16) have found that a variety of predisposing characteristics, enabling characteristics, and need factors are associated with mental health care utilization. Among predisposing characteristics, military service and older age are significantly associated with mental health service utilization. (17) Notably, veterans are much more likely to utilize services, which may be due to comorbidity of combat-related mental health problems and TBI (18). As adults with TBI become older, they are less likely to obtain mental health services which may be due to physical or cognitive problems affecting service access. (17, 19) Among enabling resources, TBI patients without health insurance are less likely to utilize services, indicating that lack of insurance continues to be a reason for not seeking mental health treatment in the United States (20). Not surprisingly, homelessness and unemployment are associated with greater mental health care utilization and need for care (19). Finally, need factors including worse self-reported health, medical comorbidities and psychiatric comorbidities are consistently reported as the strongest predictors of mental health care utilization. TBI is associated with higher rates of psychiatric comorbidities including posttraumatic stress disorder (PTSD), anxiety, mood disorders, schizophrenia, and substance use disorders; many of these conditions are significantly associated with increased mental health care utilization (17, 21).
Several studies have reported an association of mental health care utilization with TBI (17–19, 22). Miles et al. (2017) examined associations between TBI history and mental health service use in a subsample of returning veterans who were newly diagnosed with PTSD, depression, and/or anxiety. PTSD and TBI history, but not depression or anxiety, were associated with a greater number of psychotherapy visits when controlling for demographic and clinical variables. TBI history was related to greater mental health care utilization independent of mental health diagnoses and demographic characteristics (18). Fasoli et al. performed a prospective observational study of veterans receiving inpatient or outpatient mental health care (19). They found that the number of comorbidities, past mental health service use, and psychiatric diagnosis categories were the strongest predictors of mental health care utilization. The relationship between predisposing characteristics and enabling resources differed by type of utilization (inpatient, residential, or outpatient). Older age, lack of social support, and fewer problems with access predicted greater outpatient mental health care utilization. Drag et al. examined predictors of outpatient and inpatient health care utilization in veterans with a history of TBI (22). Mental health disorders such as mood disorders, substance use disorders, PTSD, and schizophrenia were associated with inpatient and outpatient mental health care utilization in veterans with TBI. Although psychiatric disorders were more prevalent in the TBI group and associated with increased medical and mental health care utilization within the TBI group, they did not account fully for the significant group differences (22).
TBI may not directly result in mental health care utilization; rather it may be that TBI patients who utilize mental health services may differ across known factors to influence mental health service use including age, sex, unemployment, medical comorbidities, and psychiatric comorbidities. Traditionally, regression analyses have been used to control for confounding associations between TBI and mental health care utilization. However, bias may inevitably persist. For example, the effect of TBI on mental health care utilization may be biased because the distribution of factors influencing mental health care services among TBI patients who use these services compared to uninjured patients are different.
Enrollment of patients into randomized controlled trials with TBI as a “treatment” is unethical; a popular method to resolve the challenges posed by experimental studies is to utilize propensity score matching which allows researchers to make more accurate inferences by balancing non-equivalent groups that may result from using a non-randomized design (23). When properly implemented, propensity score matching aims to mimic experimental designs by removing systematic differences in observed characteristics between those with and without the exposure. (24) This is achieved by first calculating the propensity of experiencing the exposure (i.e., TBI) based on specified confounding factors (24). By ruling out the influence of covariates, the propensity score matching approach allows the closest possible approximation to a randomized experiment (25).
The methodological novelty of this study is the use of propensity score matching to reduce selection bias, thereby producing a less biased estimate of the effect of TBI on mental health care utilization. The estimated effect, which was defined as the difference in outcome (in this case, mental health care utilization) between the scenario in which an individual receives a treatment (in this case, TBI) and the counterfactual scenario in which a similar individual does not receive the treatment, was determined (26).