Making sense out of utilization data

Abstract

Health care utilization studies often reach conflicting conclusions about the appropriate measures of utilization and the relationships between such measures and the explanatory variables. The purpose of this paper is to demonstrate why such ambiguities exist, provide empirical tests of different models, and suggest an appropriate analytical framework. A household survey of a rural California community was used to collect both family and individual data on utilization, "need," accessibility, attitudes, and demographics. A number of possible utilization models are presented, each containing alternative sets of dependent and independent variables. Multiple regression analysis is applied to each model, providing considerable insight into the roles of specific independent variables in explaining alternative utilization measures. The omission of certain variables, such as health status (or "need"), can result in an incorrect interpretation of the results.