Objectives: Currently used metrics for antibiotic utilization are not linked to a diagnosis and may be difficult to interpret. We aimed to describe patterns and variation in antibiotic management of three common infectious disease diagnoses in Veterans Affairs Medical Centers (VAMCs).
Methods: This descriptive retrospective cohort study included all veterans discharged between 2007 and 2009 after an acute care admission to a VAMC and a principal ICD-9-CM code for pneumonia (PNEU), skin and soft-tissue infections (SSTI), urinary tract infections (UTI) or non-infectious disease-related diagnoses (NON_INF). Systemic antibiotic use was assessed based on barcode medication administration data. A treatment period was defined as a time of uninterrupted inpatient antibiotic therapy with no gaps of >1 calendar day.
Results: Over the study period there were 1.44 million discharges with an acute care admission to 128 VAMCs included in the study. Of the discharges, 58 118 had a primary ICD-9-CM diagnosis classifiable as PNEU, 36 797 as SSTI, 30 223 as UTI and 1 243 098 as NON_INF. Empirical antibiotic use during the first 2 hospital days was frequent for all infectious disease diagnoses. Type of empirical coverage varied among facilities for all conditions, as did treatment durations. Roughly half of all empirical courses of therapy where the patient was still hospitalized on day 4 did not exhibit any change to the antibiotic regimen.
Conclusions: There is substantial variation in the management of common infectious diseases with regard to choice of empirical agents and duration of therapy. Diagnosis-specific antibiotic use metrics may prove useful for antibiotic stewardship programmes.
Keywords: antibiotic use; infectious disease; practice variation.