Analysis and prediction of the BSE incidence in Ireland

Our purpose was to report the statistical methodology that was used to describe the nature of bovine spongiform encephalopathy (BSE) propagation in the Irish cattle population, to predict the number of future cases and to assess the risk to humans in terms of the number of infected animals that were processed. We used a nonlinear Poisson-regression model for the available birth-cohort data and an iterative method to compute the parameter estimates. Standard errors for the estimates were computed from the nonlinear model and these were validated using a bootstrap procedure. We illustrated the use of the model for prediction and risk assessment using the BSE incidence data between 1981 and 2000. The change in case ascertainment or reporting level was a crucial parameter that determined the observed pattern of clinical BSE. Significant propagation risk was detected from 1985 onwards, with peaks in 1986 and 1994. The trough in the propagation risk in 1990 coincided with a ban of the use of meat-and-bone meal for ruminant feed. Excluding the newly adopted active surveillance method in 2001, the predicted and observed data were comparable.