2008 HSR&D National Meeting Abstract
1075 — Does Fragmented Financing Undermine Coordination of Care?
Pizer SD (Health Care Financing & Economics (HCFE) and BU), Gardner JA
Veterans who come to VA facilities for care may also receive care in community settings financed by Medicare, Medicaid, commercial insurance, or others. This fragmentation of financing reflects a division of responsibility among health care organizations that makes care coordination more difficult and could lead to unfavorable health outcomes. On the other hand, fragmented financing could improve outcomes by covering a broader array of services. To assess the relative importance of these two possible effects, this study examines the relationship between fragmented financing and risk-adjusted outcomes, controlling for differences in utilization.
We analyze administrative data from VA, Medicare, and Medicaid for a large sample (n=794,235 in 1999) of veterans with any history of use of VA services from FY1998 to FY2001. We estimate probit regressions predicting the probability of preventable hospitalization and mortality, adjusting for co-morbidities and demographics. We measure the fragmentation of financing by calculating the proportions of care delivered by VA and non-VA sources and taking the maximum. We use a system of three equations to model the simultaneous determination of outcomes, health services utilization, and fragmentation of financing.
We find that fragmented financing is strongly associated with higher probability of risk-adjusted preventable hospitalization, holding utilization constant (P < 0.0001). However, we did not find statistically significant effects on the probability of risk-adjusted mortality in a similar model.
The estimated effect of fragmented financing on preventable hospitalization is consistent with coordination failures between VA and non-VA providers. The size of the estimate indicates that fragmented financing may have large deleterious effects on patients who rely on both systems. However, the lack of a significant effect on mortality raises the possibility that the observed hospitalization effect could be due to insufficiently controlled variation in coverage and utilization.
The VA continues to invest in equipment and processes that are designed to provide complete, up-to-date information to VA clinicians. The effectiveness of these investments is limited by the frequency of non-VA care received by VA patients and the lack of communication between a patient’s VA and non-VA providers.