Data feedback and behavioural change intervention to improve primary care prescribing safety (EFIPPS):multicentre, three arm, cluster randomised controlled trial

Objectives: High-risk prescribing in primary care is common and causes significant harm. Data feedback has been shown to be effective in many contexts, but not for prescribing safety. The aim of this study was to evaluate the effectiveness of prescribing safety feedback compared to moderately enhanced usual care.Design: Three-arm, highly pragmatic cluster-randomised trial. Practices were randomised to one of three arms by an independent statistician. Allocation was masked to the research team throughout. Analysis was intention to treat, included all eligible patients at every time point, and was adjusted for practice baseline rates for each outcome. There were no protocol changes after the trial started.Setting and participants: 262/278 (94.2%) primary care practices in three Scottish Health Boards. Interventions: Practices were randomised to: (1) ‘usual care’ consisting of e-mailed educational material with support for searching to identify patients (88 practices at baseline, 86 analysed; (2) usual care plus feedback on practice high-risk prescribing sent quarterly on five occasions (87 practices, 86 analysed), or (3) usual care plus the same feedback incorporating a behaviour change component (87 practices, 86 analysed). Main outcome measures: The primary outcome was a patient-level composite of six prescribing measures relating to high-risk use of antipsychotics, non-steroidal anti-inflammatories and antiplatelets. Secondary outcomes were the six individual measures. The primary analysis compared high-risk prescribing in the two feedback arms to usual care at 15 months. Secondary analysis examined immediate change and change in trend of high-risk prescribing associated with intervention implementation within each arm. Results: In the primary analysis, high-risk prescribing measured by the primary outcome fell from 6.0% to 5.0% in the usual care arm, compared to 6.0% to 4.5% in the feedback-only arm (OR 0.88 compared to usual care, 95%CI 0.80 to 0.96, p=0.007) and 6.2% to 4.5% in the feedback plus behaviour change component (OR 0.86, 95%CI 0.78 to 0.95, p=0.002). In the pre-specified secondary analysis of change in trend within each arm, the usual care educational intervention had no effect on the existing declining trend in high-risk prescribing. Both types of feedback were associated with significantly more rapid decline in high-risk prescribing after the intervention compared to before.Conclusions: Feedback of prescribing safety data was effective at reducing high-risk prescribing. The intervention would be feasible to implement at scale in contexts where electronic health records are in general use.