Individual health insurance data of antibiotic delivery in previous months as a tool to predict bacterial resistance of urinary tract infection: A prospective cohort study

Objectives: We aimed to quantify the individual risk of antimicrobial resistance among patients with community acquired Escherichia coli urinary tract infection (UTI) according to their antibiotic exposure over the previous 18 months. Patients and methods: French patients were prospectively recruited in two centers in 2015-2017. Resistance of isolates to amoxicillin (AMX), amoxicillin-clavulanate (AMC), third-generation cephalosporins (3GC), trimethoprim-sulfamethoxazole (TMP-SMX), fluoroquinolones (FQ) and fosfomycin (FOS) was analysed according to previous intra-class and inter-class antibiotic exposure documented in health insurance files. Results: Previous antibiotic exposure was found in 588 (81.4 %) of the 722 UTI cases analysed (564 patients). Recent exposure (three months before UTI) was associated with stronger intra-class impact on E. coli resistance compared to remote exposure (18 months before UTI) for AMX, AMC, FQ and TMP-SMX, with respective adjusted odds ratios [95 % confidence interval] of 1.63 [1.20-2.21], 1.59 [1.02-2.48], 3.01 [1.90-4.77], and 2.60 [1.75-3.87]. AMX, FQ, and TMP-SMX also showed significant inter-class impact. Resistance to 3GC was not significantly associated with intraclass exposure (adjusted OR: 0.88 [0.41-1.90]). FOS resistance was remarkably low (0.4 %). Duration of the antibiotic-free period required for resistance risk to drop below 10 %, the threshold for empirical use in UTI, was modelled as < 1 month for 3GC, >18 months for AMX and TMP-SMX and uncertain for AMC (5.2 months [2.3 to > 18]) and FQ (17.4 months [7.4 to > 18]). Conclusions: Resistance of E. coli causing UTI is partially predicted by previous personal antibiotic delivery.