Antimicrobial resistance monitoring in commensal and clinical Escherichia coli from broiler chickens: Differences and similarities

Abstract

Background: In the Netherlands, antimicrobial resistance (AMR) is monitored in commensal indicator Escherichia coli from healthy broilers at slaughter as part of a European monitoring programme. In a separate programme for poultry health, AMR is monitored in veterinary pathogens from diseased broilers. So far, it is unknown how the outcomes of these two AMR monitoring approaches in the same animal population are associated. Aims: This study aims to investigate the association between the outcomes of monitoring non-wildtype susceptibility (using epidemiological cut-off values, ECOFF, as prescribed by EU legislation) in commensal E. coli isolated from healthy broilers (i.e. active surveillance) with the outcomes of monitoring clinical resistance (using clinical breakpoints, to determine susceptibility for antibiotic treatment in veterinary practice) in E. coli isolated from diseased broilers (i.e. passive surveillance). Methods: Data acquired by broth microdilution was analysed for commensal indicator E. coli and clinical E. coli from the Netherlands, 2014–2019. A generalized linear multivariable model (Poisson regression) was used to determine time trends and identify differences in mean resistant proportions. Results: Observed resistant proportions of the monitored commensal E. coli and clinical E. coli were similar with overlapping confidence intervals for most time points for ampicillin, gentamicin, cefotaxime, tetracycline, colistin and trimethoprim/sulfonamide. The statistical analysis showed that only for cefotaxime and tetracycline, mean resistant proportions were different. In commensal E. coli, a decrease of resistant proportions over time was observed, except for gentamicin. In clinical E. coli, no time trend was detected in resistant proportions, except for cefotaxime and colistin. Conclusions: Generally, the resistant proportions monitored in commensal and clinical E. coli were similar. However, some relevant differences were found, which can be explained by the type of monitoring approach, i.e. active or passive surveillance. The random sample of commensal E. coli isolated from healthy animals (active surveillance), was more suitable to monitor AMR time trends. The sample of clinical isolates from diseased animals (passive surveillance), resulted in a higher chance to detect low-prevalent resistance: i.e. cefotaxime and colistin. The clinical E. coli data showed more fluctuation over time, and data from a longer period of time would be needed to determine the association. This study shows the value of both an active and a passive surveillance component for AMR monitoring.