Geographic differences in the use of Veterans Administration hospitals.

This study contrasts the determinants of community hospital utilization with Veterans Administration (VA) hospital utilization using traditional planning variables. The comparisons had some expected and some unanticipated findings. Regional differences in non-VA hospital admissions and bed days are fairly well explained by measures of medical need, provider supply, community alternatives, and sociodemographic characteristics (other than those used as proxies for case mix). However, regional variations in the VA are explained less well by the same classes of variables and the unexplained differences between the two systems do not correspond geographically. This suggests that the two systems have different reasons for regional variation. Further, contrary to expectation, when other predictors are held constant, excess bed capacity in the area does not correlate with lower VA utilization. The study is important as the VA comes under increasing pressure to contain costs. It may well be that the rational planning model attributed to the public sector is less likely to overcome maldistribution than the private sector 'invisible hand'. Policy analysts need to give more attention to the political, bureaucratic determinants of resource allocation before changing eligibility criteria or merging the two systems.