Social norm feedback reduces primary care antibiotic prescribing in a regression discontinuity study

Background
Reducing antibiotic prescribing is a priority for health authorities responsible for preventing antimicrobial resistance. Northern Ireland has high rates of antimicrobial use. We implemented a social norm feedback intervention and evaluated its impact.

Objectives
To estimate the size and duration of the effect of a social norm feedback letter to GPs who worked in the 20% of practices with the highest antimicrobial prescribing.

Methods
The letter was sent in October 2017 to 221 GPs in 67 practices. To assess the effect of the intervention, we used a sharp non-parametric regression discontinuity (RD) design, with prescribing rates in the four calendar quarters following the intervention as the outcome variables.

Results
In the quarter following the intervention (October to December 2017) there was a change of −25.7 (95% CI = −42.5 to −8.8, P = 0.0028) antibiotic items per 1000 Specific Therapeutic group Age-sex Related Prescribing Units (STAR-PU). At 1 year, the coefficient was −58.7 (95% CI = −116.7 to −0.7, P = 0.047) antibiotic items per 1000 STAR-PU. The greatest change occurred soon after the intervention. Approximately 18 900 fewer antibiotic items were prescribed than if the intervention had not been made (1% of Northern Ireland’s annual primary care antibiotic prescribing).

Conclusions
A social norm feedback intervention reduced antibiotic prescribing in the intervention practices. The diminishing effect over time suggests the need for more frequent feedback. The RD method allowed measurement of the effectiveness of an intervention that was delivered as part of normal business, without a randomized trial.