A multilevel analysis of trimethoprim and ciprofloxacin prescribing and resistance of uropathogenic Escherichia coli in general practice

Objectives: A retrospective analysis of databases was performed to describe trimethoprim and ciprofloxacin prescribing and resistance in Escherichia coli within general practices in the West of Ireland from 2004 to 2008. Methods: Antimicrobial susceptibility testing was performed by disc diffusion methods according to the CLSI methods and criteria on significant E. coli isolates (colony count >10(5) cfu/mL) from urine samples submitted from general practice. Data were collected over a 4.5 year period and aggregated at practice level. Data on antimicrobial prescribing of practices were obtained from the national Irish prescribing database, which accounts for similar to 70% of all medicines prescribed in primary care. A multilevel model (MLwiN) was fitted with trimethoprim/ciprofloxacin resistance rates as outcome and practice prescribing as predictor. Practice and individual routinely collected variables were controlled for in the model. Results: Seventy-two general practices sent between 13 and 720 (median 155) samples that turned out to be E. coli positive. Prescribing at practice level was significantly correlated with the probability of antimicrobial-resistant E. coli with an odds ratio of 1.02 [95% confidence interval (CI) 1.01-1.04] for every additional prescription of trimethoprim per 1000 patients per month in the practice and 1.08 (1.04-1.11) for ciprofloxacin. Age was a significant risk factor in both models. Higher variation between practices was found for ciprofloxacin as well as a yearly increase in resistance. Comparing a 'mean' practice with 1 prescription per month with one with 10 prescriptions per month showed an increase in predicted probability of a resistant E. coli for the 'mean' patient from 23.9% to 27.5% for trimethoprim and from 3.0% to 5.5% for ciprofloxacin. Conclusions: A higher level of antimicrobial prescribing in a practice is associated with a higher probability of a resistant E. coli for the patient. The variation in antimicrobial resistance levels between practices was relatively higher for ciprofloxacin than for trimethoprim.