Influencing NSAID prescribing in primary care using different feedback strategies

The Audit Commission endorsed the role of the prescribing adviser in promoting safe, rational and cost-effective prescribing by general medical practitioners (GPs). However, whether such roles should involve practice visits, facilitation of educational meetings, production of local bulletins, or a combination of these and other approaches is unclear. Few UK studies have investigated the best methods to influence prescribing on a large scale in primary care. The present study was designed to determine the effectiveness of active compared to passive practice specific prescribing feedback. The programme focused on non-steroidal anti-inflammatory drugs (NSAIDs) since concern has been expressed about their use, which accounts for 4% of volume and 5% of the cost of UK National Health Service drugs prescribed in primary care, Sixty-six of the 91 general medical practices contracted to Gwent Health Authority agreed to participate in the study and were randomly stratified by practice size, locality, fund-holding and dispensing status into 2 groups. Group 1 received active feedback via practice visits from the pharmaceutical prescribing adviser to present prescribing analysis and cost data (PACT) concerning NSAID use. Group 2 received passive feedback, a practice specific prescribing analysis workbook that contained similar information to that given to Group 1 practices. Practices not wishing to enter the study were used as a self selected reference group (Group 3) which received no information on NSAIDs from the prescribing adviser. Practice visits and the distribution and completion of workbooks occurred between September 1993 and March 1994, PACT data for all NSAIDs was used to identify changes in prescribing before and after the programme. A combination of 27 indicators, in terms of items and cost per 1000 patients, were chosen to identify overall changes and potential switches between individual drugs, or to generic alternatives. Comparison of the practices in each of the three groups at analysis revealed similar distribution in terms of stratification criteria. Eleven (38%) Group 2 practices returned completed workbooks. Overview indicators (those not targeted) showed similar trends of either increase or decrease, in cost and volume, across all three groups, whereas targeted indicators demonstrated a more mixed picture between groups. In summary the total number of statistically significant changes for targeted indicators in Groups 1, 2, and 3 were 10, 8 and 1 (changes in items per 1000 patients), and 12, 10 and 3 (changes in cost per 1000 patients) respectively. Targeted indicators revealed more statistically significant changes in Group 1 (active feedback) than Group 2 passive feedback) which showed more changes than Group 3 (reference group). Active feedback was more effective at bringing about a required change than the use of passive feedback; both approaches had more impact than that registered by the reference group. The Group 2 analysis presented included both responders and non-responders, thus a more marked benefit of the workbook was perhaps masked. These results have implications for the future provision of prescribing advice to practices. Advisers currently use both active and passive methods to provide prescribing feedback to practices. The high resource intensiveness of practice visits leads many authorities to rely increasingly on less effective methods such as bulletins. Evaluation of the cost effectiveness of methods used to influence prescribing are required.