When did bovine spongiform encephalopathy (BSE) start? Implications on theprediction of a new variant of Creutzfeldt-Jakob disease (nvCJD) epidemic

Background Knowing the starting date of the BSE epidemic and its size at th e very beginning is crucial to interpret the timing of the nvCJD cases and to forecast the nvCJD epidemic. The first cases occurred in 1985. The model s devised by Anderson (backcalculation) and Dealler (age-period-cohort) led to an estimate of less than 50 cases in 1983, and none earlier. Here, we a pplied age-cohort models to the BSE data in order to estimate the earliest possible date of the first unrecognized BSE cases. Methods The numbers of confirmed BSE cases in the UK, by age group and by c alendar year from 1988 to 1996, were analysed by Poisson regression. The ca ses' age distribution was considered as constant between the different birt h cohorts. The herd's age structure was taken into account. Results According to the models, BSE cases may have occurred as early as 19 80. The expected number of cases before 1990 is almost twice the number of confirmed cases and exceeds by more than 20% the expected value of Anderson 's model. The scenario of first human exposure in 1980 leads to fewer futur e nvCJD cases than predicted by Cousens with exposure patterns starting in 1983 or 1985. Conclusions The first birth cohort available, consisting of two cases older than 10 in 1988, does not allow any projections before 1980. Moreover conf idence intervals are wide and the power of the study is limited by the grea t dispersion of the data; the precision of the estimations would be improve d by considering geographical incidence. Nevertheless, our projections are consistent with Wilesmith's survey of rendering plants relating the emergen ce of BSE to the dramatic fall in the proportion of meat and bone meal foll owing-solvent extraction, initiated in the late 1970s (65% in 1977 to 10% i n 1983).