Abstract
IMPORTANCE: Antibiotic overuse is harmful to patients and the health care system. Antibiotic prescribing report feedback has been described in outpatient and long-term care settings but is relatively untested in inpatient settings. OBJECTIVE: To assess the association between peer comparative inpatient antibiotic prescribing feedback and changes to antibiotic prescribing rates among hospitalists in a large health care network. DESIGN, SETTING, AND PARTICIPANTS: This quality improvement study linked data on billing and prescribing of broad-spectrum antibiotics for hospital-onset infections (BS-HO antibiotics) from hospitalists at 5 diverse acute care facilities encompassing both academic and primarily community-based hospitals within the same large health care network. Data were pooled into 12 two-month periods from January 1, 2023, through December 31, 2024, in a quality improvement intervention with a stepped-wedge cluster design. Hospitalists who contributed billed patient days for at least one 2-month period during the study period received the intervention. INTERVENTIONS: Observed-to-expected ratios (OERs) for days of therapy (DOT) of prescribed antibiotics were calculated for each hospitalist per period and reported back to hospitalists in a peer comparative report disseminated via email every 2 months. MAIN OUTCOMES AND MEASURES: Hospitalists' prescribing rates for BS-HO antibiotics with activity against Pseudomonas aeruginosa were assessed, accounting for time, clustering of hospitalists, and patient characteristics including comorbidities and clinical syndrome. RESULTS: The study included 169 hospitalists (median per hospital, 30 [range, 24-50]) for a total of 1687 bimonthly observation periods. Per 2-month period, hospitalists had a mean (SD) of 126 (48) patient encounters. Among hospitalists at facilities receiving the intervention, the prescribing rate ratio (RR) for DOT was higher among clinicians caring for higher proportions of patients with sepsis (RR, 1.04; 95% CI, 1.00-1.08) and end-stage kidney disease (RR, 1.09; 95% CI, 1.05-1.14) and was lower for each sequential reporting period (RR, 0.99; 95% CI, 0.98-1.00). In multivariable models accounting for these variables and the trend over sequential periods, the intervention was not significantly associated with lower prescribing rates (RR, 0.97; 95% CI, 0.91-1.04). CONCLUSIONS AND RELEVANCE: Peer comparative inpatient prescribing reports for hospitalists were not associated with a change in hospitalists' prescribing rates of BS-HO antibiotics. The findings suggest additional efforts to augment the utility of the reports are justified.