2018 Conference Presentation
Background: Since 2001, the United States has deployed more than 2.7 million troops. Upon returning home, many injured veterans face substantial barriers to work-related integration and unmet physical and mental health needs. These challenges have negative ramifications for the mental health of informal caregivers. The Department of Veterans Affairs (VA) offers a range of medical and social services to eligible veterans. Moreover, in 2010, Congress established the nationwide VA Program of Comprehensive Assistance for Family Caregivers (PCAFC) to increase caregiver knowledge of and integration into the VA system. Prior work found that enrollment in PCAFC was associated with increased use of veteran outpatient health services (Van Houtven et al, 2017). Building on this work, we propose to examine whether institutional support for family caregivers can enhance veteran use of the following VA social services: supported employment, vocational rehabilitation and education assistance.
Methods: Instrumental variable estimation addresses endogenous selection into the PCAFC program. Six month lagged medical-center approval rates for PCAFC, which vary in value and over time at 151 medical centers, are exogenous to individual patient and caregiver characteristics but strongly predict program enrollment, effectively quasi-randomizing dyads to treatment or control. Those who apply, but are not accepted into PCAFC, serve as the non-equivalent control group. The first stage probit model predicting PCAFC enrollment, in addition to the IV, also controls for baseline patient demographics (e.g. race, ethnicity, age, marital status), economic status, mental and physical health conditions from ICD9 codes, prior outpatient utilization, relationship to the caregiver (spouse, adult child, other), along with medical-center and time fixed effects. The second stage model, a Cox proportional hazard model with two-stage residual inclusion, will estimate the effect of institutional support of caregivers on time until veteran social service use, using bias correction methods suggested by Martinez-Camblor et al, 2017. The three social service use outcomes, time to use of supported employment, vocational rehabilitation, and education assistance, are modeled separately.
Results: The cohort contains 15,836 veterans whose caregivers were enrolled in and 9,106 veterans whose caregivers applied, but were not accepted into PCAFC. Enrolled veterans were a lower proportion female (7.6% versus 10.9%) and were on average 2.4 years younger than control group veterans. A higher proportion of enrolled veterans had musculoskeletal disorders (64.8% versus 58.9%) and post-traumatic stress disorder (73.7% versus 60.2%) compared to control group veterans. Preliminary first stage probit models indicate that the IV was empirically strong (F-Statistic>10) depending on the model specification. The second stage models will be estimated in April 2018, with completed results expected prior to the ILPN conference.
Conclusions: Reintegrating into work after military service injury could yield lifelong positive welfare gains for disabled adults and their families and positive system outcomes if it reduces use of disability benefits and/or reduces costly health services use. This research could have important implications for international policy efforts to train and support informal caregivers by quantifying the extent to which trained and supported caregivers can be an effective mechanism to help disabled individuals engage in social services.