Abstract
INTRODUCTION: Urinary tract infections (UTI) are common community-based bacterial infections. Empiric antibiotic recommendations are guided by local resistance rates. Previous research suggests that cultures are overused for uncomplicated cystitis, but practice patterns have not been described in detail. Variations in culturing have implications for the interpretation of antibiotic resistance rates. METHODS: We used a retrospective cohort study to analyze variations in urine culturing among physicians, controlling for patient and physician characteristics. We identified all outpatient physician visits among adults and children for cystitis in British Columbia between 2005 and 2011 using administrative data and linked these to laboratory data on urine cultures. Using hierarchical generalized linear mixed models we explored variations in urine culture submissions for cystitis (ICD code 595) and the associations with patient and physician characteristics, stratified by patient sex. RESULTS: Urine cultures were associated with 16% of visits for cystitis among females and 9% among males, and 59% of visits overall were associated with antibiotic treatment. Older patients, patients with a recent antibiotic prescription, and long term care residents were significantly less likely to have a culture associated with a cystitis visit, whether male or female. Female physicians and physicians with 16-35 years' experience were more likely to culture, while international medical graduates were less likely-particularly for female visits. Notably, there was substantial unexplained variation among physicians after controlling for physician characteristics: we found a 24-fold variation in the odds of culturing a female UTI between physicians who were otherwise similar. CONCLUSION: Individual physicians show substantial variation in their propensity to submit cultures for cystitis visits. Reducing such variation and encouraging appropriate levels of culturing would support effective antibiotic use.