Using mathmatical modeling to assess the representativeness of antimicrobial resistance patterns in salmonella isolates from retail meat

Antibiotic resistance is a growing issue in the clinical treatment of severe Salmonella spp infections in the United States. A large proportion of Salmonella infections are contracted through contaminated meat products. A recent study revealed that there exist a large variety of serotypes and antimicrobial resistance patterns in Salmonella isolates recovered from retail meat; among the 365 isolates there were 175 distinct PFGE patterns. This project used data from the Iowa retail meat study to assess the representativness of the sampling scheme currently employed by the FDA.

Data for this study have been provided by the US Food and Drug Administration; the data is a subset of data collected during the Iowa retail meat study. An iterative Markov Chain Monte Carlo method is used to model the various antimicrobial resistance patterns found among the isolates and the variety of patterns seen across various cultivation media used in the laboratory. Several models are tested that attempt to discern the role that different media play in the manifestation of resistance patterns seen in the laboratory.

This analysis helps to determine whether federal antimicrobial resistance monitoring programs are testing enough meat samples and Salmonella isolates per meat sample to adequately predict the antibiotic resistance patterns seen in clinical salmonellosis. This analysis also contributes to the growing body of knowledge being used to conduct quantitative microbial risk assessments.