150 - Using audit-and-feedback to improve antibiotic-prescribing in Emergency Departments: a quasi-experimental study across 4 sites in the Veterans Health Administration

Background: Antibiotic-prescribing in Emergency Departments (EDs) is often inappropriate. In this study, we evaluated whether audit-and-feedback could improve antibiotic use in EDs.Figure 1. Comparison of antibiotic-prescribing between the pretest and intervention periods at 2 intervention EDs and 2 control EDs


Methods: We pilot tested an audit-and-feedback intervention using a quasi-experimental study design at 2 intervention and 2 matched-control EDs with a 12-month pretest and a 12-month intervention period. At intervention sites, 27 of 31 (87.1%) clinicians were enrolled; at baseline, they received 1) one-on-one education about antibiotic-prescribing and 2) individualized feedback with comparisons to local peers. Feedback included personalized antibiotic-prescribing data for all ED visits and specifically for viral acute respiratory infections (ARIs); feedback was updated quarterly. The primary outcome was the antibiotic-prescribing rate for ED visits not resulting in hospitalization, and it was assessed using a segmented regression analysis of monthly time series data. Manual chart reviews were performed to assess guideline-concordant management (i.e. prescribing an antibiotic when indicated and not prescribing when not indicated) for 5 ARIs plus cystitis.

Results: In the pre-test and intervention periods, intervention sites had 28,146 and 27,396 visits compared to 31,439 and 32,295 visits at control sites. After implementation started, intervention sites saw an immediate decrease in antibiotic use (-10.3%, p=0.15) compared to a 1.5% increase (p=0.88) at control sites. By the end of the intervention period, there was an 8.9% decrease in antibiotic use at intervention sites compared to a 3.4% decrease at control sites [relative risk ratio (RRR) -3.3% (95% CI, -8.4 to +1.7), Figure 1]. Guideline-concordant management improved from 52.1% to 72.2% (p< 0.01) at intervention sites compared to 51.3% to 58.2% (p=0.13) at control sites. Intervention and control sites had similar changes in 30-day outcomes, including late antibiotic prescriptions and hospitalizations.

Conclusion: After the implementation of audit-and-feedback at 2 EDs, antibiotic use did not significantly decrease compared to 2 control EDs but guideline-concordant management improved.  Future studies should include more study sites to improve statistical power and also evaluate the effectiveness of more frequent and specific feedback.