Sources of variability in foot and mouth disease vaccine potency estimates based on serum neutralizing antibody assay

Some vaccines can be assayed for potency by measuring the serum antibody response they produce in vaccinated test animals. Using data obtained from potency assays on batches of foot and mouth disease vaccine, the sources of variability in such a method were examined. A linear model is proposed for the analysis of replicate serum neutralizing antibody assays, which represents an improvement on the usual approach of working with only a mean serum assay value for each test animal. Components of variance were calculated, allowing the relative importance of the numbers of test animals, or the numbers of serum assays per test animal, to be estimated in terms of the variability of the overall group mean antibody response. A method is described for calculating fiducial intervals for the serological potency estimates, and it is shown that these intervals are no larger than, and are in fact probably smaller than, those obtained from quantal challenge tests. The results have important implications for the design and analysis of similar biological tests used for other products.