Abstract Body

Background: Since 2001, a large number of national surveys with measurement of HIV status have been conducted, most of them in sub-Saharan Africa (35 countries), but also in selected countries in Asia (2 countries) and in the Caribbean (2 countries). Many of these surveys are conducted within the Demographic and Health Survey programme, others are conducted as AIDS Indicator Surveys supported by international or national organisations. Typical surveys have been limited to the 15-49 year age range for women and 15-54 or 15-59 for men. Conclusions: Population surveys provide many advantages over traditional surveillance systems. These include that they include men as well as women, and that they are nationally representative and in recent years also designed to inform at subnational level (typically state or province). Their representative design has been critical for adjusting biases in estimated prevalence from antenatal clinic data. For example Ethiopia and the Democratic Republic of the Congo revised their national prevalence dramatically down on the basis of national surveys which showed much lower prevalence than the ANC surveillance data. The HIV prevalence information from surveys has informed epidemiological patterns used in the estimation of national epidemics, including for the sex and age pattern of prevalence, and to adjust prevalence levels from antenatal clinics in countries without survey. Another important benefit from population-based surveys is the ability to link HIV status with socio-demographics and behaviours, allowing cross-sectional bivariate and multivariate analyses of determinants and impacts. Examples of how this has changed our understanding of HIV include the ability to map HIV prevalence to wealth status, geographic location, and behaviours. Limitations include the possibility of biased estimates of HIV prevalence related to survey participation, their infrequent conduct (often once every 5 years), and their relatively high cost (several million USD). Recently, the inclusion of novel laboratory methodologies is starting to enable direct estimates of HIV incidence (while validation of the optimal assay algorithms is ongoing, and confidence intervals are large) and exposure to antiretroviral treatment (ART). In the future, surveys may be also be used to assess community-based viral load and to inform the epidemiology of small geographic areas of particular interest. Expanding the age range of the sampled population, surveys can provide additional information on the HIV epidemiology among children (to date few surveys have included children) and among people over 50 years of age.