Improving antimicrobial residue surveillance in finishing pigs by risk-based sampling designs

EU Member States are obliged by legislation to implement residue surveillance programs to detect illegal use or misuse of veterinary medicines in food producing animals and investigate the reasons for residue violations. According to EU legislation, these programs should be (partly) risk-based, meaning targeted towards groups of animals, where the probability of finding residues is the highest. There is however no default surveillance procedure describing the most efficient way to do so. In this study, a quantitative analysis was conducted to quantify the effectiveness of detecting antimicrobial residues in finishing pigs via risk-based sampling of carcasses. A stochastic scenario tree analysis was applied to estimate the sensitivity of random and risk-based sampling strategies to detect a contaminated carcass. In these models, the probability was calculated that a single carcass will yield a positive outcome when subjected to the testing protocol laid down in the design, given that contamination with antimicrobial residues is prevalent in the herd of origin at the level of the design prevalence. Two design prevalences were used: 0.01% (the assumed true prevalence of residue-positive carcasses) and 0.22% (the prevalence that can be detected using the sample size laid down in EU legislation). In the random design, it was assumed that the carcasses examined for presence of antimicrobial residues were selected randomly from all finishing pigs slaughtered in a year. In the risk-based design, two risk factors were taken into account. First, a high prevalence of chronic pleuritis and pneumonia in the herd of origin was assessed. Secondly, the route of administration of antimicrobials (oral/parenteral) via visual inspection of skin lesions indicative of injectables was used as an additional risk factor. Results showed that the probability of detecting a residue-positive carcass doubled when surveillance was targeted at pigs originating from herds with a high prevalence of chronic pleuritis and pneumonia (compared to random sampling), at similar costs of testing. Including administration route as an additional risk factor led to a negligible increase in sensitivity. Nevertheless, sensitivity values at unit level remained extremely low due to the very low prevalence of antimicrobial residues in pigs. In this study, risk-based alternatives to random sampling improved the cost-effectiveness of residue surveillance in slaughter pigs in the Netherlands, which could be used to enhance current programs and to increase awareness in food business operators.