Decision analysis as a quality-assurance screening tool

The purpose of this pilot study was to examine whether the technique of decision analysis, including sensitivity analysis, could be used in a clinical quality-assurance program.

This research was performed in a family practice residency clinical practice. A computerized decision analysis model was developed for selection of initial drug therapy for hypertension. The medical records of 52 resident-managed patients with hypertension were then reviewed. The residents' drug prescribing was evaluated by faculty reviewers and also by the decision analysis model, including a sensitivity analysis.

Faculty reviewers rated the residents' drug choices as "most appropriate" or "acceptable" in 59.6% of cases. There was good agreement between faculty reviewers and the computerized decision analysis model. For example, in those cases in which the resident chose the computer model's first-choice drug, faculty deemed the management as "most appropriate" or "acceptable" 93.3% of the time. When residents selected the computer model's third or fourth choice, faculty judged the residents' therapy as "inappropriate" or "an alternate drug would have been more desirable" in 61.9% of cases.

The results suggest that computerized decision analysis techniques may be a useful adjunct to other clinical quality-assurance procedures in residency training programs.