Antimicrobial resistance surveillance: bacterial prevalence estimates are not enough

Surveillance programs for monitoring antimicrobial resistance in foodborne (enteric) bacteria largely focus on estimating and comparing phenotypic prevalence among different livestock species, their farm environments and food products, and assessing trends over many years. Often, these monitoring systems have a high threshold for detection of rare resistance geno-and pheno-types owing to: 1) relatively small sample sizes that are further diluted by multiple species/strain/antimicrobial combination endpoints, and 2) the use of non-selective media that, while assuring unbiased resistance prevalence estimates at the bacterial population level, provide little confidence in freedom from the presence of bacterial resistance at the sample / animal / pen / farm / industry / country level, particularly for commensal (indicator) organisms. Much of the latter problem lies with the enormous size of the bacterial population in a single gram of feces, and therefore the relatively small probability of culturing a strain unless it is present in relatively high numbers. Very few surveillance programs specifically target early detection and characterization of emergent strains of resistant bacteria, particularly in the years immediately following the introduction of new classes, variants or formulations of antimicrobial. This has serious implications for the utility of these surveillance programs in ensuring food safety and protecting public health. Using the relatively recently introduced 3rd generation cephalosporin class of antimicrobial and the more longstanding tetracycline class as contrasting examples, we illustrate an alternative approach to monitoring and surveillance that blends elements of more traditional resistance prevalence estimation with targeted detection and strain characterization; this, in order to better understand novel threats to food safety and potentially mitigate emerging problems before it is too late.