The relationship between primary care antibiotic prescribing and bacterial resistance in adults in the community: a controlled observational study using individual patient data

Objectives: To examine the relationship between primary care prescribed antibiotics and the development of antibiotic resistance in perineal flora contaminating unselected urinary isolates from a large sample of asymptomatic adults representative of the general community. Patients and methods: Escherichia coli isolates contaminating urine samples were obtained from asymptomatic adults aged > 16 years registered with general practices in the former Avon and Gloucestershire health authority areas. Data on antibiotic exposure during the 12 months prior to providing the urine samples were collected from the primary care electronic and paper medical records. The main outcome measure was resistance to amoxicillin or trimethoprim or both. Results: Two thousand nine hundred and forty-three adults submitted urine samples. Susceptibility among E. coli isolates and antibiotic prescribing data were available from 618 patients. We found no evidence of an association between resistance and patients' exposure to any antibiotic prescribed in primary care in the previous 12 months [adjusted odds ratio (OR) 1.12, 95% confidence interval 0.77-1.65, P = 0.52]. Secondary analyses demonstrated greater resistance in patients exposed to antibiotics within 2 months (adjusted OR 1.95, 1.08-3.49, P = 0.03), a dose-response relationship to increasing exposure to trimethoprim in the previous 12 months (adjusted OR 1.01, 1.01-1.02, P = 0.001) and that individuals who had been prescribed any beta-lactam antibiotic in the previous 12 months had amoxicillin MICs more than twice (adjusted 95% CI 1.23-3.31, P = 0.009) that of those who had not been prescribed any beta-lactams. Conclusions: Whether or not adults receive a prescription for any antibiotic during a 12 month period does not appear to influence the antimicrobial resistance of perineal flora. However, the temporal and dose-response relationships found may be suggestive of a causative association and should be the focus of further research.