Easy and reliable assessment of the prevalence of porcine post-weaning diarrhoea.

Post-weaning diarrhoea is a condition of increasing importance due to recent restrictions and bans on the preventive use of antimicrobials and medicinal zinc oxide. For various purposes, it is valuable to monitor the occurrence of post-weaning diarrhoea. The aim of this paper was to propose a protocol for easy and reliable assessment of the prevalence of post-weaning diarrhoea within a section of pigs as an alternative to clinical examination of a random sample of pigs. Two datasets were collected in two different observational field investigations, including more than 4000 individual clinical examinations of newly weaned pigs. First we identified a clinical marker for post-weaning diarrhoea. Second, we drew samples by simulation from our two dataset using different simplified sampling strategies and compared these to conventional random sampling strategies. The prediction error for estimates of the diarrhoea prevalence within a section was compared for the different sampling strategies. The study showed that pigs with diarrhoea had an increased risk of displaying diarrheic soiling of the hind part as well as dull, long, and bristly hair coat, and possibly also hollow flanks and a suboptimal body condition score. Diarrheic soiling of the hind part was the best clinical predictor of diarrhoea; our best estimate of the diagnostic sensitivity was 76.1% (95% credible interval: 72.0, 79.8), and the specificity was 97.1% (95% credible interval: 96.5 0.97.7). Diarrheic soiling of the hind is only a valid clinical predictor of diarrhoea for the first 14 days after insertion into the nursery unit. The precision and accuracy of prevalence estimates were similar for haphazard sampling compared to random sampling of the pigs. Likewise, sampling from a restricted number of pens produced prediction errors similar to sampling across all pens. Yet, the study had limitations regarding sample sizes, and furthermore it is difficult to provide certainty for absence of effects. We recommend sampling pigs haphazardly within at least three randomly selected pens for post-weaning diarrhoea prevalence surveys in order to easily obtain a reliable prevalence estimate. Based on our findings, we conclude the paper by proposing a simple four-step protocol for surveys of the within-section prevalence of post-weaning diarrhoea.