OBJECTIVES: Improving and promoting rational drug use is a great interest. We aimed to assess the effectiveness and cost-effectiveness of prescribing audit and feedback intervention in improving physician prescribing. METHODS: A four-arm randomized trial with economic evaluation conducted in Tehran. Three interventions (routine feedback, revised feedback, and printed educational material) and a no intervention control arm compared. Physicians working in outpatient practices were randomly allocated to one of the four arms using stratified randomized sampling. The interventions were developed based on a review of literature, physician interviews, current experiences in Iran and with theoretical insights from the Theory of Planned Behavior. Effects of the interventions on improving antibiotics and corticosteroids prescribing assessed in regression analyses. Cost data assessed from a health care provider's perspective and incremental cost-effectiveness ratios calculated. RESULTS: Comparing the new-design feedback arm and the no intervention arm, we observed significant reductions in the proportion of prescriptions including Dexamethasone injectable (1.64 difference in percentage change; p value: 0.006) and Cefixime (0.99 difference in percentage change; p value: 0.01). We also observed significant reductions in the printed educational material arm's proportion of prescriptions including Cefixime (0.93 difference in percentage change; p value: 0.04) as compared with the no intervention arm. ICER values corresponding to Dexamethasone injectable and Cefixime were 0.41 and 1.03 US$ per unit reduction in the number of prescriptions respectively. CONCLUSIONS:

According to the results of this study, we recommend that a careful attention to the format, design and the way the messages are conveyed in feedback forms is an important indicator of audit and feedback's potential success in improving prescribing behavior. Considering the incremental cost-effectiveness ratio, the cost-effectiveness of new-design feedback intervention arm has been proved.