Local characteristics associated with higher prevalence of ESBL-producing Escherichia coli in community-acquired urinary tract infections: an observational, cross-sectional study.

Objectives: The prevalence of ESBL-producing Escherichia coli (ESBL-E. coli) in community-acquired urinary tract infections (UTI) has been increasing worldwide since 2000, but with large geographical variations. The aim of this study was to determine whether the ESBL-E. coli rate in urine samples from individuals with community-acquired UTI was associated with the local socio-economic, environmental, agricultural and healthcare characteristics.

Methods: This was a cross-sectional study in western France using data on antibiotic susceptibility of E. coli isolated from urine samples of individuals with community-acquired UTI analysed in non-hospital laboratories from 2015 to 2017. The ESBL-E. coli rate was calculated for each laboratory. Data on socio-economic characteristics, human antibiotic consumption, hospital bed density, animal farming density and percentage of agricultural land and surface water were retrieved at the municipality level and aggregated by study area. Their association with ESBL-E. coli prevalence was quantified using multivariate linear regression models with a backward selection.

Results: From 358 291 E. coli isolates from urine samples tested in 92 laboratories, the mean ESBL-E. coli prevalence for the study period was 3.30%. In an adjusted model, the ESBL-E. coli rate was significantly (P < 0.05) and positively associated with the local percentage of people >65 years old, third-generation cephalosporin use (DDD/1000 inhabitants), number of hospital beds/km2, poultry density, pig density and percentage of agricultural land. Lower deprivation was associated with a higher ESBL-E. coli rate.

Conclusions: Several anthropogenic factors (primary care, hospitals and animal farming) are associated with the local ESBL-E. coli rate in community-acquired UTI. These results could contribute to improve risk management, including identification of at-risk patient groups.