Objectives: To describe the methodology used in a recent survey of HIV/AIDS in South Africa and to present the response rates.
Methods: A cross-sectional, national household-based survey was conducted using second-generation surveillance procedures. A complex multistage sampling technique was used to create a master sample of 1,000 census enumerator areas out of a total of 86,000 nationally. Aerial photographs were taken and used to randomly select more than 10,197 households and ultimately 13,518 individuals from a sampling frame of 31,321 people. Phase 1 of the study involved notifying the household residents about the study and collecting key demographic information on respondents aged 2 years and older. This information was used to randomly select up to 3 respondents from each household: 1 adult (25 years and older), 1 youth (15-24 years), and 1 child (2-14 years). In phase 2 nurses interviewed respondents and collected oral fluid specimens for HIV testing. In the case of children aged 2-11 years, parents or guardians were interviewed, but HIV testing was performed on the selected children. Questionnaire data were anonymously linked with HIV test results.
Results: A total of 9,963 persons agreed to be interviewed and 8,840 were tested for HIV, yielding a response rate of 73.7% and 65.4% respectively. However, only 8,428 (62.3%) HIV test results were correctly matched with behavioural data. The results showed that those tested for HIV did not differ from those not tested in terms of key determinants.
Conclusion: It is possible to use community-based surveys to study the prevalence of HIV in the general population.