Session number: 1027
Abstract title: VA-Medicare Enrollees’ Reliance on VA Care: Controlling for endogeneity of comprehensive Medicare
Author(s):
FH Wang - Center for Health Quality, Outcomes and Economic Research
Y Shen - Center for Health Quality, Outcomes and Economic Research
J Gardner - Center for Health Quality, Outcomes and Economic Research
A Jayadev - Center for Health Quality, Outcomes and Economic Research
Objectives: Health insurance affects people’s demand for health care and is itself determined by their perceived need for care (endogeneity). We examined factors influencing VA-Medicare dual enrollees’ decision to purchase Medigap/HMO coverage, and how coverage and other factors affect VA/non-VA utilization. We accounted for the endogeneity of insurance to eliminate estimation bias.
Methods: ‘1999 enrollee survey data’ (insurance coverage, VA/non-VA use, personal characteristics) was merged with AHA data and Area Resource File (to measure access to VA/non-VA care), N=66,274. Enrollees have “comprehensive Medicare” if they reported Medigap/HMO coverage. We measured reliance on VA care by respondents’ self-reports of using, in the preceding 12 months, VA alone, both systems, or non-VA alone. We employed probit model to examine how characteristics (VA priority or health status) affect the ikelihood of having comprehensive coverage. Ordered probit model tested the impact of Medicare coverage and access to VA/non-VA care systems on the choice to use VA, controlling for other characteristics. Maximum-likelihood was applied for unbiased estimation, controlling for the endogeneity of coverage.
Results: Overall, 58% of respondents reported having comprehensive Medicare coverage. 38% reported using VA exclusively, 48% used both VA and non-VA and 14% used non-VA exclusively. Enrollees were more likely to have comprehensive coverage if they were female, white, employed/retired, married, older, having chronic diseases, and VA priority 6/7. For example, priority 7 (high income) enrollees are 21% more likely to have insurance. The endogeneity of insurance was significant. Enrollees without comprehensive coverage were more likely to use VA as well as those who lived close to VA hospitals. Enrollees were more likely to use VA if they were male, nonwhite, unemployed, unmarried, younger age, catastrophically disabled (priority 4), low income (priority 5) and 50%+ service connected-disabled (priority 1).
Conclusions: Many factors affect VA-Medicare dual enrollees’ purchase of insurance and their reliance on VA care. Proper econometric methods should be used to control for the endogeneity of insurance to eliminate estimation bias.
Impact statement: Identifying groups of enrollees relying on VA heavily is an important step for VA to deliver specific needs for target groups.