Exploratory analysis of risk factors associated with antimicrobial resistance in urinary Escherichia coli from dogs in the United States (2018-2024).

backgroundBacterial cystitis, predominantly caused by Escherichia coli, is a leading cause for antimicrobial prescriptions in dogs. The use of antimicrobials contributes to selection pressure for antimicrobial resistance (AMR), a growing concern for veterinary and public health. This study examined risk factors for resistance in urinary E. coli isolates from dogs in the United States, focusing on prior positive cultures and other host, socioeconomic, and geographic factors using data from a national commercial diagnostic laboratory. Associations between selected exposures of interest and AMR were assessed using multilevel logistic regression models, accounting for clustering at the patient, county, and state levels.

resultsA total of 393,972 clinical urinary E. coli isolates from 339,977 dogs across the United States, submitted between 2018 and 2024, were tested for antimicrobial susceptibility. Resistance was highest to amoxicillin (27.39%; 95% CI: 27.25-27.53). When examined in multilevel logistic regression models, and across all antimicrobials, the greatest proportion of variance occurred at the patient level (range: 22.2-46.1%). The log odds of detecting resistance increased linearly with the number of prior positive cultures in the previous six months. After adjusting for all other covariates, every 1% increase in the annual county-level percentage of dogs with positive culture and antimicrobial susceptibility test was associated with a 1-2% decrease in the odds of resistance across all antimicrobials (OR range: 0.98-0.99; p < 0.001). States with 1,500-1,999 housing units per veterinarian had higher odds of marbofloxacin resistance (OR: 1.85, 95% CI: 1.20-2.85, adjusted p = 0.027) than states with 500-999 units, a pattern consistent across other antimicrobials.

conclusionPatient-level factors, particularly recent repeated positive cultures examined in this study, are associated with AMR in canine urinary E. coli, likely reflecting cumulative antimicrobial exposure and/or chronic resistant infections. Additionally, lower odds of detecting AMR were observed in counties with higher microbiological testing frequency and in states with higher veterinarian density, which may reflect differences in access to veterinary care or diagnostic practices. However, because these variables are measured at the group level, they do not capture individual-level veterinary care. Further research incorporating patient-level measures of care access and prescribing practices will be crucial.