189 - Validating a Hospitalist-Specific Antibiotic Prescribing Metric across Four Acute Care Hospitals

Background: Peer comparison reduces unnecessary outpatient antibiotic prescribing, but no prescribing metric has been validated for inpatient comparison. We aimed to evaluate if an electronically derived antibiotic prescribing metric correlated with indicated antibiotic days in hospitalized patients.

Methods: We previously created a hospitalist-specific adjusted antibiotic use metric (observed:expected [O:E]) for National Healthcare Safety Network-defined broad-spectrum antibiotics. From May-Oct 2019 at four Emory Healthcare hospitals, we identified outlier hospitalists prescribing in the top (high O:E) and bottom (low O:E) 15^th percentile. We randomly selected 10 days of antibiotic administration from each outlier and reviewed days with > 2 days of consecutive days of antibiotics. For pneumonia, chronic obstructive pulmonary disease (COPD), or urinary tract infection (UTI) we determined if each day of antibiotics was indicated, assuming the diagnosis was accurate. We compared high vs. low O:E providers and used regression modeling to determine if the metric predicted indicated days of antibiotics.  

Results: Among 997 days, 510 (51%) were from high and 487 (49%) from low O:E providers. High O:E providers had a greater proportion of days with > 2 prior days of antibiotics (60%) compared to low O:E providers (54%, p = 0.03). In the subset of days with > 2 prior days of antibiotics (n = 569), high O:E providers had more patient-days with longer hospital stays, diabetes and Charlson comorbidity index (CCI) >3, and fewer days supervising (resident/advanced practice provider, Table 1). The primary diagnosis was pneumonia, COPD exacerbation or UTI in 260 (25%) days; 91% were indicated based on duration with no difference between high and low O:E providers (88% vs. 94%, p = 0.1). After controlling for days of hospitalization, CCI, immunocompromised status, and supervisory role, a high O:E was not associated with indicated antibiotic use (OR 0.5, 95% CI 0.2 – 1.3).Description of days with a patient on greater than two days of antibiotics, comparing high- versus low-metric providers


Conclusion: A high hospitalist antibiotic prescribing metric correlated with patients receiving > 2 consecutive days of antibiotics on any given day but did not predict unindicated antibiotic use for a subset of diagnoses. Evaluating indicated use by validating diagnoses may improve metric performance.