Do More Hospital Beds in an Area Induce Excess Demand Final rept. (July 3, 2003, July 2, 2005)

摘要

The purpose of the report is to examine the extent to which hospitals with more 'available' beds induce use of those beds. Studies of geographical variations in hospitalization rates suggest that the availability of hospital beds in an area is an important factor determining hospitalization rates. However, there are methodological problems with the traditional way in which area bed supply is determined, the effect of which is to induce a correlation between bed availability and utilization even if there is no supply-induced demand. We use a Bayesian hierarchical model to examine the relationship between bed availability and hospital induced demand in a way that avoids biases associated with traditional analyses. We consider hospital admissions and outpatient treatment in Massachusetts in 2000 of Medicare patients with one of 15 medical conditions. Using Bayesian hierarchical models, we estimate 3 effects that underlie observed utilization patterns: (1) the amount of disease in an area; (2) 'true' area-specific hospital utilization (i.e., if there were no supply-induced demand); and (3) a hospital induced utilization factor for each hospital. We then examine the relationship between beds per expected person (or day) based on true utilization and the hospital induced utilization factor. The Bayesian models result in estimates of the correlations that were similar to those from the traditional approach, in the .50 to .60 range. Biases associated with traditional approaches offset each other. As a result, the correlation resulting from traditional approaches is a reasonable estimate of the strength of underlying relationships.