To examine private insurance coverage and its impact on use of Veterans Health Administration (VA) care among VA enrollees without Medicare coverage.
The 1999 National Health Survey of Veteran Enrollees merged with VA administrative data, with other information drawn from American Hospital Association data and the Area Resource File.
We modeled VA enrollees' decision of having private insurance coverage and its impact on use of VA care controlling for sociodemographic information, patients' health status, VA priority status and access to VA and non‐VA alternatives. We estimated the true impact of insurance on the use of VA care by teasing out potential selection bias. Bias came from two sources: a security selection effect (sicker enrollees purchase private insurance for extra security and use more VA and non‐VA care) and a preference selection effect (VA enrollees who prefer non‐VA care may purchase private insurance and use less VA care).
VA enrollees with private insurance coverage were less likely to use VA care. Security selection dominated preference selection and naïve models that did not control for selection effects consistently underestimated the insurance effect.
Our results indicate that prior research, which has not controlled for insurance selection effects, may have underestimated the potential impact of any private insurance policy change, which may in turn affect VA enrollees' private insurance coverage and consequently their use of VA care. From the decline in private insurance coverage from 1999 to 2002, we projected an increase of 29,400 patients and 158 million dollars for VA health care services.