Erratum to “Estimation of infection prevalence from correlated binomial samples” [Prev. Vet. Med. 64 (2004) 1–14]

Campylobacter in housed broiler chickens: a longitudinal study of risk factors

Infections by Campylobacter spp. are a major cause of gastrointestinal disease in the United Kingdom. Most cases are associated with the consumption of chicken that has become contaminated during production. We investigated the epidemiology of Campylobacter spp. in chickens in a 3-year longitudinal study of flocks reared on 30 farms in the United Kingdom. We used Generalized Linear Mixed Effect Models (GLMM) to investigate putative risk factors associated with incidence and prevalence of flock infection arising from farm and flock management and local environmental conditions during rearing. We used survival analysis to investigate infection events and associated risk factors over the course of the study using two marginal models - the independent increment approach, which assumed that individual infection events were independent; and a conditional approach, which assumed that events were conditional on those preceding. Models of flock prevalence were highly overdispersed suggesting that infection within flocks was aggregated. The key predictors of flock infection identified from the GLMM analyses were mean temperature and mean rainfall in the month of slaughter and also the presence of natural ventilation. Mean temperature in the month of slaughter was also a significant predictor of flock infection, although the analyses suggested that the risk in flocks increased in a unimodal way in relation to temperature, peaking at 12 degrees C. The extent of pad burn was also identified as a predictor in these analyses. We conclude that predicting prevalence within flocks with linear modelling approaches is likely to be difficult, but that it may be possible to predict when flocks are at risk of Campylobacter infection. This is a key first step in managing disease and reducing the risks posed to the human food chain.