Estimation of the relative risk of developing clinical scrapie: the role of prion protein (PrP) genotype and selection bias.

Prion protein (PrP) genotype data from statutory confirmed cases and from three non-case datasets have been used to calculate the odds ratio (or) for the development of clinical scrapie for an individual sheep of a given PrP genotype, compared with one possessing the "wild-type" ARQ/ARQ genotype. Logistic regression has been used to estimate the ors, and a multiple-test procedure has been used to evaluate the statistical significance of each comparison. The results are similar to those observed in other studies: the VRQ/VRQ genotype has or point estimates greater than 20; the ARQ/VRQ and ARH/VRQ genotypes have or point estimates between 5 and 20; AHQ/VRQ between 0.03 and 0.1; ARR/VRQ 0.4 and 0.5; all the other PrP genotypes, excluding ARR/ARR, ARR/ARH and AHQ/ARH for which no clinical cases have been recorded have or point estimates of less than 0.3. The estimates derived from each dataset are comparable, but not identical. This can be explained by plausible biases inherent in the sampling of the non-case populations.