The impact of employment on quality of life for adults with brain injury

1.1 Study purpose

Employment is often considered an important rehabilitation outcome; however, more research is needed to demonstrate the extent to which employment is associated with greater QOL for people with TBI over and above other psychosocial correlates, such as social networks and support, problem-solving confidence, and sense of community. The purpose of this study was to examine the impact of current employment status (working vs. not working) on the self-reported QOL of adults with TBI, over and above the impact of psychosocial and control variables. We hypothesized that even after controlling for the effects of severity of current symptoms and age, employment status and psychosocial variables related to sense of community, social networks and support, and problem-solving confidence would be uniquely and collectively associated with subjective QOL for working-age adults with TBI.

2.1 Participants and procedures

Participants included 116 adults with self-reported TBI who were recruited through brain injury associations across the U.S. to participate in an anonymous survey study. To be eligible for the study, participants had to be: (a) between the ages of 18 and 65; (b) diagnosed with a brain injury no less than six months previously; and (c) currently residing in the community. The research team reached out to the Brain Injury Association of America (BIAA) as well as individual state brain injury associations and alliances for support in study recruitment. Depending on the state organization preference, the study opportunity was disseminated through emails to membership listservs and/or posted on brain injury organization websites. The online survey was set up through Qualtrics survey software. Institutional Review Board (IRB) approval was obtained prior to the start of the study. Of the 164 questionnaires that were returned by respondents under 65 years old, 27 cases were excluded for not reporting QOL (the dependent variable) and/or work status. Missing values analysis found that data were missing completely at random (Little’s MCAR Test: χ²(58, N = 164) = 49.96, p = 0.765). An additional 20 cases were removed for indicating “retired” or “student” to enhance the interpretability of the “work status” variable. Mean substitution was used for continuous psychosocial predictor variables for individuals missing no more than 3 predictor variables (n = 6 cases).

Participant characteristics are summarized in Table 1. Participants ranged in age from 21 to 64 years old (M = 46.5, SD = 11.0; Mdn = 46.0). More than half of the participants were female (59.5%), and the vast majority of the sample described themselves as White (87.1%). Nearly all (95.7%) of the participants had some post-secondary education, yet most (62.9%) of the participants reported being unemployed. Of the 43 participants who were employed, over half (55.8%) reported working full-time. The average household income of participants was $47,717. When asked about relationship status, just under half of the participants reported having a partner (43.1%). The majority of participants reported a TBI caused by motor vehicle accident (68.1%), followed by fall (20.7%) and assault (16.4%). Most participants self-reported their injuries as severe (45.7%) or moderate (40.5%) and reported an average time since injury of 14.2 years (Mdn = 11.2). The majority (82.8%) reported experiencing post-traumatic amnesia, with 44 (37.9%) indicating it lasted more than 7 days.

2.3 Data analysis

Hierarchical linear regression analysis is a form of multiple regression that examines the incremental amount of variance explained by predictor sets entered as separate steps or blocks. This approach is used to statistically control for certain variables to examine if adding variables significantly improves the model’s ability to predict the dependent variable. In this study, hierarchical linear regression was used to examine the incremental impact of work status on QOL after controlling for severity of symptoms, age, and psychosocial variables. The change in R² (ΔR²) was examined as a measure of each predictor set’s contribution. Three blocks were entered to address the research question. In the first block, symptom severity and age were entered as control variables. In the second block, psychosocial variables (sense of community, emotional social support, instrumental social support, informal social network, formal social network, problem solving confidence) were entered. In the final model, work status (employed = 1, not employed = 0) was entered. This order of blocks was used to facilitate an understanding of the unique contribution of work status over and above related psychosocial variables on QOL. Significance tests for the regression coefficients for each predictor variable were assessed at each step and at the final model to assess unique relationships to the dependent variable (QOL). A priori power analysis was conducted using G^*Power (Faul et al., 2009) and estimated that to achieve a power level of 0.80 with an alpha level of 0.05, and 9 predictors in multiple regression, the required sample size to detect a medium effect size of f² = 0.15 was 114 (actual sample size = 116). Prior to the analysis, zero-order correlations were examined for correlations of 0.70 or larger and Variance Inflation Factors (VIF) were examined for scores of 10 or greater (Belsey et al., 1980). VIF scores in this study did not exceed 1.8 suggesting multicollinearity was not a concern.

3.1 Descriptive and correlation analyses

Means, standard deviations and correlations of the study variables are presented in Table 2. On average, participants reported moderately positive levels on the 5-point QOL scale (M = 3.36, SD = 1.02, range 1 to 4). QOL was significantly correlated with almost all study variables, with the exception of formal support networks. Moderate to high correlations were found between QOL and symptom severity (r = –0.42, p < 0.001), problem solving confidence (r = 0.48, p < 0.001), sense of community (r = 0.43, p < 0.001), instrumental social support (r = 0.35, p < 0.001) and emotional social support (r = 0.57, p < 0.001). Work status was also significantly correlated with QOL (r = 0.36, p < 0.001), symptom severity (r = –0.23, p < 0.01), age (r =–0.34, p < 0.001), and formal support networks (r =0.29, p < 0.01).

To address the research question, a hierarchical regression analysis was used with three sets of predictors included in the regression equation: (a) age and symptom severity; (b) psychosocial variables including sense of community, instrumental social support, emotional social support, formal social networks, informal social networks, and problem-solving confidence; and (c) work status. Results of the analysis are summarized in Table 3 and include unstandardized regression coefficients (B), standard errors (SE), and standardized coefficients (β), as well as values of change in R² (ΔR²) for the predictor variables at each step and in the final model.

The first step of the regression analysis consisted of age and symptom severity, which accounted for a significant amount of variance in QOL, R² = 0.21, F (2, 113) = 14.62, p < 0.001. Specifically, symptom severity was negatively associated with QOL such that greater symptom severity was associated with poorer reported QOL (β= –0.38, t (113) = –4.38, p < 0.001), as was also the case for younger age (β= –0.17, t (113) = –1.98, p = 0.05). Psychosocial variables were entered next and accounted for significant variance in QOL as well, R² = 0.53, ΔR = 0.33, F (8, 107) = 15.30, p < 0.001. Further examination of the standardized partial regression coefficients showed that sense of community (β= 0.24, t (107) = 3.25, p = 0.002), emotional social support (β= 0.39, t (107) = 4.42 p < 0.001), informal social networks (β= 0.19, t (107) = 2.56, p = 0.012) and problem solving confidence (β=0.20, t (107) = 2.34, p = 0.021) contributed significantly to the variance in QOL, such that increased presence of these psychosocial variables were associated with greater QOL. Formal social network was inversely associated with QOL at this step (β= –0.16, t (107) = –2.11, p = 0.037). In the final step of the model, work status was entered, which accounted for a significant increase in the variance of QOL explained, ΔR² = 0.05, F (1, 106) = 13.60, p < 0.001. Work status contributed significantly to QOL such that employment was associated with greater QOL (β= 0.27, t (106) = 3.69, p < 0.001). In the final model, symptom severity (β= –0.12, t (106) = –1.43, p = 0.138) and age (β= 0.02, t (106) = 0.33, p = 0.742) were no longer significant predictors.

Overall the full model explained 59% of the variance of QOL, R² = 0.59, F (9, 106) = 16.71, p < 0.001. This is considered a large effect size within behavioral science as outlined by Cohen1992). After controlling for demographics, severity, and psychosocial variables, work status (β= 0.27, p < 0.001) was the second strongest predictor of QOL after emotional social support (β= 0.32, p < 0.001). These results indicate that participants who are employed reported greater QOL even after controlling for psychosocial correlates.

4.1 Implications and practice and research

The findings from this study indicate that employment, in addition to psychosocial variables, offer a unique contribution to overall QOL in adults with brain injury. This means that effective vocational rehabilitation services are critical to ensuring that persons with brain injury achieve higher QOL. For people with disabilities, including those with brain injury, there are several barriers to employment, such as transportation (Rosenbloom, 2007); discrimination in the workplace (Shier et al., 2009); and employer concerns such as cost, legal liability, and lack of knowledge of accommodations (Kaye et al., 2011). Strategies to overcome these barriers have been aimed at the system itself (via legislative work, such as the Americans with Disabilities Act), employers, and/or consumers. Employers may assist individuals with TBI by offering gradual return to work options or workplace modifications (Gourdeau et al., 2018) and by dismantling workplace discrimination that individuals face (Stergiou-Kita et al., 2017).

In the United States, the state-federal vocational rehabilitation system is the oldest and most successful public program for supporting the employment of individuals with disabilities, including those with brain injury (Ditchman et al., 2021). This program serves approximately one million individuals per year and plays a large role in helping people with disabilities reach their employment goals (Martin et al., 2010; U.S. Department of Education, 2018). Services are guided by consumer choice and designed to maximize employability, independence, and community (and workplace) participation of people with disabilities by helping them prepare for, secure, maintain, or regain employment. The employment rates of people with disabilities after receiving these services are found to be around 55% to 60% (Dutta et al., 2008; Rosenthal et al., 2006; U.S. Department of Education, 2016). Specific vocational rehabilitation services that have the strongest positive impact on employment outcomes for individuals with brain injury served in this program include on-the-job training, job placement, job search assistance, and on-the-job support (Ahonie et al., 2020; Catalano et al., 2006). However, disparities in outcomes have been observed, and research suggests that while these services are beneficial, individuals with brain injury who are of ethnic/racial minority status and those with work disincentives (e.g., receiving disability benefits) are less likely to obtain employment (e.g.., Cardoso et al., 2007; Catalano et al., 2006). Differences in the effectiveness of these services have also been documented across sex, revealing that women with TBI were less likely to be successfully employed through state vocational rehabilitation services than men despite similar injury and demographic characteristics (Bounds et al., 2003).

There are a number of evidence-based strategies to support employment and return to work for individuals with disabilities, including those with brain injury. For example, motivational interviewing is an empirically supported, client-centered approach that can be effective to promote client motivation, address barriers to change, and improve the therapeutic relationship (Chan et al., 2016; Lundahl et al., 2010). It can be used for addressing motivational issues related to employment (Wagner & McMahon, 2004), and may be a useful strategy for facilitating acceptance of goals, building the working alliance, and promoting engagement in employment services for individuals with brain injury (Medley & Powell, 2010). Supported employment, or ongoing support services, such as job coaching, also have a strong research base for promoting employment outcomes of people with disabilities (Campbell et al., 2011), and supported employment can be a cost-effective and beneficial approach for adults with brain injury (Caplan et al., 2015; Targett & Wehman, 2010; Wehman et al., 2003). For more information on these practices, a training toolkit for vocational rehabilitation service providers on evidence-based practices has been developed by the Rehabilitation Research and Training Center on Evidence-Based Practices in Vocational Rehabilitation (RRTC-EBP-VR, 2016) and is available online. Although there are a number of models for brain injury vocational rehabilitation, there is a paucity of high-quality research to evaluate effectiveness (Tyerman, 2012). Recognizing that service needs likely change over time, more research is needed to understand which vocational services and interventions are most effective, for whom and at what stages post-injury.

Finally, it is important to recognize that problem-solving confidence, social support, sense of community, and network ties may be enhanced by work. In the workplace, problems frequently arise; engaging in work may increase an individual’s problem-solving confidence as she or he is required to address workplace concerns. Emotional support may increase in the workplace as co-workers, supervisors, and family members help an individual transition back to work. A sense of community may be restored as individuals become re-integrated into the workplace environment. Informal network ties may also develop with regained employment, as individuals may have financial resources and increased motivation to see friends and family outside the workplace.

4.2 Limitations and future research

There are several limitations to consider when reviewing the findings of this study that suggest areas for future research. First, the sample size was relatively small considering the prevalence of TBI. Despite the study having adequate power to complete the hierarchical regression analysis, an increased sample size could impact the incremental changes between steps. The sample was also predominantly White, female, and with postsecondary education, which does not fully represent the diverse population of those with TBI. Furthermore, individuals who are connected to brain injury networks may have unique experiences of community that may not exemplify the experience of the larger brain injury population. Future research examining these variables should employ larger, more diverse samples of participants to increase the generalizability of findings. It is also important to note that the majority of the participants described their injury as “moderate” or “severe”; however, this categorization was based on self-report rather than a specific medical definition and poses a limitation in characterizing the sample. Second, the self-report nature of this study may also limit the participant sample, in that those with more severe functional difficulties may have challenges completing the survey or submitting it. At the same time, self-report surveys provide a direct perspective from individuals with brain injury rather than proxy measures. For individuals with brain injury, proxy measures may be incongruent with individual ratings for subjective measures, such as satisfaction with social participation (Dawson et al.; Hart et al., 2010). Measuring subjective QOL most accurately captures the individual’s perspective, rather than a caregiver or spouse’s point of view. A third limitation is that the dependent variable, QOL, was measured using a single item rather than a scale. A single item measure for subjective QOL may be the best way to capture QOL, as it is a rating of the perceived QOL of an individual (Gill & Feinstein, 1994). In fact, using lengthy scales and assessments may lack the face validity that a single, global rating may provide (De Boer, et al., 2004). However, there are a number of well-validated measures of QOL available. Many of these can be accessed through the TBI-QOL measurement system that provides access to an item bank and freely available scales (Tulsky & Kisala, 2019; Tulsky et al., 2016). Finally, an important limitation of this study is that data were captured at a single time point. Given the cross-sectional design of this study, hierarchical regression analysis was the strongest statistical approach to determine predictors of QOL. However, the findings are ultimately correlational and causality cannot be assumed. Using a longitudinal approach would provide additional information about the causality of outcomes.