[Systematic bias in measuring of the utilization of ambulatory medical services]

The results of utilization studies on ambulatory medical care are usually disappointing. Since these studies have used the regression linear model with cross-sectional data, they have rarely considered the utilization of medical services as a dynamic process unfolding over a certain period of time. The utilization of medical services was observed during three consecutive two-week periods with the help of a sample, drawn from the files of the RAMQ, of 41,811 individuals living in Laval and in the Metropolitan region of Quebec. Transition rates towards user or non-user status between each period are predicted, rather than volumes of utilization. Results show that the variables which predict the volume of utilization are not those which predict the transition towards non-user status, though volume of utilization arises from the cumulation over time periods of the using status. Differences appear between the utilization processes of persons aged 65 and over and other adults. Consequently, studies on the utilization of medical services should no longer use multiple linear regression. Thus, the conclusions which these studies draw about the impact of policies of accessibility to health care are also biased.