The statistical analysis of data from immunoepidemiological studies.

Immune responses encountered in human populations display a higher level of heterogeneity than is typically seen with laboratory animals, and the expected clear distinction between responders and non-responders to particular antigens may not be apparent. In such a situation, the methods of statistical analysis traditionally used in laboratory studies may no longer be appropriate, and there is no consensus among immunologists about how such data should be analysed. Using data from a large study of lymphocyte proliferation responses to candidate antigens for a vaccine against Plasmodium falciparum malaria, we discuss questions such as the appropriateness of logarithmic transformation of the data, the use of the stimulation index versus delta--cpm, the use of the proportion of responders for comparing two groups of individuals, and the choice of threshold for defining such a responder.