Estimating the impact of health services in a community

The advantages of clinical and mechanical combination of observations for prediction are mutually reinforced by an application of Bayesian statistics. The technique is shown to be particularly advantageous in a situation which is characterized by a paucity of observations available to suppliment a prior expert judgment. The approach also presents a basis for evaluating relative expertise and tracing the learning experience of experts. Taking the community as the organizational level of analysis, data on the classification of aged persons in Durham County, North Carolina, and data on the changes in classification of these individuals within the population over time are gathered and studied. These data are combined with clinical judgments of changes as a demonstration of the merits of the technique developed in this paper.