Relating incidence to 'recent infection' prevalence: application to HIV

I present a systematic approach to the derivation of relationships between disease incidence and the prevalence of an experimentally defined state of 'recent infection'. These depend, in general, on details of the population dynamic and epidemiological history, as well as the physiology of early disease progression. The general relations facilitate the evaluation of numerous approximate schemes that could be used for the purpose of estimating incidence from snapshot surveys. Methods for calibrating an incidence/prevalence relation from follow-up studies, for use in subsequent snapshot surveys, are considered. Example data from an ongoing follow-up study are analysed. Statistical power and prospects for practical implementation in southern Africa are also considered.