Support for informal caregivers as a mechanism to enhance use of vocational reintegration services for disabled veterans

Background: Vocational reintegration after military service yields lifelong positive welfare and health gains. However for some veterans, substantial health and mental comorbidities impede return to work. The Department of Veterans Affairs (VA) offers a range of vocational assistance services to meet veteran needs, but these services are under-utilized. One reason for this underuse is that medical and social service delivery is fragmented—services are administered by different Bureaus in VA. VA support for family caregivers can increase this service use by informing caregivers about vocational rehabilitation services and teaching them to navigate the complex VA system. In 2010, the US Congress established the VA Program of Comprehensive Assistance for Family Caregivers (PCAFC)—a national program to support family caregivers that includes a monthly stipend payment (~$600-$2300). We examine whether participating in PCAFC enhances veteran use of VA vocational assistance services: The Post-9/11 GI Bill; Vocational rehabilitation (VRE); and Supported employment (SE).

Methods: The treatment group includes veterans under 55 years old who were accepted into PCAFC between 2011-2014. The non-equivalent control group includes Veterans who applied, but were not accepted into PCAFC. We use two-stage residual inclusion instrumental variable (IV) estimation to address endogenous selection into the PCAFC program. The IV is the six-month lagged medical-center PCAFC approval rate; this rate is exogenous to individual patient and caregiver characteristics because it reflects the combined practice styles of the examiners at the site and strongly predicts program acceptance. The first stage logit model regresses PCAFC enrollment on the IV and baseline patient demographics, economic status, mental and physical health conditions (ICD9 codes), prior outpatient utilization, relationship between the veteran and caregiver, and medical-center and time fixed effects. The second stage Cox proportional hazards model estimates the effect of the PCAFC on time until veteran use of each social service. We model social service use outcomes separately: Post 9/11 GI Bill (n=9,776); VRE (n=9,390); SE (n=19,217).

Results: For all outcomes, the IV was empirically strong (F-Statistic>11) and covariate balance was greatly improved across levels of the IV. The second stage models and bootstrapped standard errors demonstrated that veterans whose caregivers were in PCAFC were 35% more likely to receive at least one SE visit (HR=1.35, 95% CI 1.06, 1.79); no statistically significant effects were observed for use of the post-9/11 GI Bill (HR=1.00, 95% CI 0.45, 2.22) or VRE (HR=0.94, 95% CI 0.55, 1.95).

Conclusions: Veterans whose caregivers participated in PCAFC were more likely to use SE; this finding suggests that support for caregivers connects veterans with SE services. This effect was only observed for SE possibly because linkages between SE and PCAFC are more direct—SE and PCAFC are housed within the same VA Bureau. To enhance caregivers’ ability to support veteran engagement in social services, VA could 1) reframe caregiver support policies to focus on employment and education as critical determinants of health for the veteran and 2) strengthen linkages with social services across VA Bureaus.