Survey and surveillance development in settings with low human immunodeficiency virus prevalence

In most countries, during the early phases of a human immunodeficiency virus epidemic, independently initiated surveys of perceived high-risk groups tend to precede the development of formal surveillance systems. Unfortunately, in low-prevalence settings, small sample sizes produce unreliable estimates of prevalence and trends, with an inevitable tendency towards positive results. In our study, we present sample size calculations and typical samples used in actual surveys, with Pakistan as our example. More useful data on risk behaviour and potential for spread can be derived from the study of commoner sexually transmitted diseases and associated risk behaviours, including assessments of knowledge, attitudes, beliefs and practices.