Mailed feedback to primary care physicians on antibiotic prescribing for patients aged 65 years and older: pragmatic, factorial randomised controlled trial

Audit and Feedback Interventions for Antibiotic Prescribing in Primary Care: A Systematic Review and Meta-analysis

BACKGROUND

This systematic review evaluates the effect of audit and feedback (A&F) interventions targeting antibiotic prescribing in primary care and examines factors that may explain the variation in effectiveness.

METHODS

Randomized controlled trials (RCTs) involving A&F interventions targeting antibiotic prescribing in primary care were included in the systematic review. Cochrane Central Register of Controlled Trials, MEDLINE, EMBASE, CINAHL, and ClinicalTrials.gov were searched up to May 2024. Trial, participant, and intervention characteristics were extracted independently by 2 researchers. Random effects meta-analyses of trials that compared interventions with and without A&F were conducted for 4 outcomes: (1) total antibiotic prescribing volume; (2) unnecessary antibiotic initiation; (3) excessive prescription duration, and (4) broad-spectrum antibiotic selection. A stratified analysis was also performed based on study characteristics and A&F intervention design features for total antibiotic volume.

RESULTS

A total of 56 RCTs fit the eligibility criteria and were included in the meta-analysis. A&F was associated with an 11% relative reduction in antibiotic prescribing volume (N = 21 studies, rate ratio [RR] = 0.89; 95% confidence interval [CI]: .84, .95; I2 = 97); 23% relative reduction in unnecessary antibiotic initiation (N = 16 studies, RR = 0.77; 95% CI: .68, .87; I2 = 72); 13% relative reduction in prolonged duration of antibiotic course (N = 4 studies, RR = 0.87 95% CI: .81, .94; I2 = 86); and 17% relative reduction in broad-spectrum antibiotic selection (N = 17 studies, RR = 0.83 95% CI: .75, .93; I2 = 96).

CONCLUSIONS

A&F interventions reduce antibiotic prescribing in primary care. However, heterogeneity was substantial, outcome definitions were not standardized across the trials, and intervention fidelity was not consistently assessed. Clinical Trials Registration. Prospero (CRD42022298297).

Process evaluation of two large randomized controlled trials to understand factors influencing family physicians' use of antibiotic audit and feedback reports

BACKGROUND

Unnecessary antibiotic prescriptions in primary care are common and contribute to antimicrobial resistance in the population. Audit and feedback (A&F) on antibiotic prescribing to primary care can improve the appropriateness of antibiotic prescribing, but the optimal approach is uncertain. We performed two pragmatic randomized controlled trials of different approaches to audit and feedback. The trial results showed that A&F was associated with significantly reducing antibiotic prescribing. Still, the effect size was small, and the modifications to the A&F interventions tested in the trials were not associated with any change. Herein, we report a theory-informed qualitative process evaluation to explore potential mechanisms underlying the observed effects.

METHODS

Ontario family physicians in the intervention arms of both trials who were sent A&F letters were invited for one-on-one interviews. Purposive sampling was used to seek variation across interested participants in personal and practice characteristics. Qualitative analysis utilized inductive and deductive techniques informed by the Clinical Performance Feedback Intervention Theory.

RESULTS

Modifications to the intervention design tested in the trial did not alter prescribing patterns beyond the changes made in response to the A&F overall for various reasons. Change in antibiotic prescribing in response to A&F depended on whether it led to the formation of specific intentions and whether those intentions translated to particular behaviours. Those without intentions to change tended to feel that their unique clinical context was not represented in the A&F. Those with intentions but without specific actions taken tended to express a lack of self-efficacy for avoiding a prescription in contexts with time constraints and/or without an ongoing patient relationship. Many participants noted that compared to overall prescribing, A&F on antibiotic prescription duration was perceived as new information and easily actionable.

CONCLUSION

Our findings indicate that contextual factors, including the types of patients and the setting where they are seen, affect how clinicians react to audit and feedback. These results suggest a need to test tailored feedback reports that reflect the context of how, where, and why physicians prescribe antibiotics so that they might be perceived as more personal and more actionable.

TRIAL REGISTRATION

Clinical Trial registration IDs: NCT04594200, NCT05044052.