Comparative analysis of clinical breakpoints (CB), normalised resistance interpretation (NRI), and epidemiological cut-offs (ECOFFS), in interpreting antimicrobial resistance of E. coli isolates originating from poultry in different farm types in Tanzania.

Existing breakpoint guidelines are not optimal for interpreting antimicrobial resistance data from animal studies and low-income countries, and therefore their utility for analysing such data is limited. There is need to integrate diverse data sets, such as those from low-income populations and animals, to improve data interpretation. There is very limited research on the relative merits of clinical breakpoints, epidemiological cut-offs, and normalised resistance interpretation breakpoints in interpreting microbiological data, particularly in animal studies and studies from low-income countries. The aim of this study was to compare antimicrobial resistance in E. coli isolates using ECOFFS, CLSI, and NRI breakpoints. A total of 69 non-repetitive poultry isolates were selected for investigation based on lactose fermentation on MacConkey agar and subsequent identification and confirmation as E. coli using chromogenic agar and uidA Polymerase Chain Reaction. Kirby Bauer disc diffusion technique was used for susceptibility testing. For each antimicrobial agent, inhibition zone diameters were measured, and ECOFFs, CLSI, and NRI bespoke breakpoints used for resistance interpretation. According to the interpretation of all breakpoints except ECOFFs, tetracycline resistance was significantly higher (TET) (67.8% – 69.5%), than ciprofloxacin (CIPRO) (18.6%– 32.2%), imipenem (IMI) (3.4% – 35%), and ceftazidime (CEF) (1.7% – 45.8%). Prevalence estimates of antimicrobial resistance (AMR) using CLSI and NRI bespoke breakpoints did not differ for CEF (1.7% CB and 1.7% COWT), IMI (3.4% CB and 4.0% COWT) and TET (67.8% CB and 69.5% COWT). However, with ECOFFs, antimicrobial resistance estimates for CEF, IMI, and CIP were significantly higher (45.8%, 35.6%, and 64.4%, respectively; P < 0.01). Across all the three breakpoints, resistance to ciprofloxacin varied significantly (32.2% CB, 64.4% ECOFFs, and 18.6% COWT). Antimicrobial resistance interpretation is influenced by the breakpoint used, necesscitating further standardisation, especially for microbiological breakpoints, in order to harmonise outputs. The AMR ECOFFs estimates in the present study were significantly higher compared to CLSI and NRI.