Time at risk: use of data on transitions between risk categories to explain individual HIV status and community HIV prevalence levels

Basia Zaba, London School of Hygiene and Tropical Medicine
Milly Marston, London School of Hygiene and Tropical Medicine
Emma Slaymaker, London School of Hygiene and Tropical Medicine
Mark Urassa, National Institute for Medical Research of Tanzania
Andreas Jahn, Karonga Prevention Study
Mia Crampin, London School of Hygiene and Tropical Medicine
Raphael Isingo, National Institute for Medical Research of Tanzania

Attempts to measure the influence of sexual behaviour on HIV, generally focus on indicators of recent sexual behaviour such as number and type of partners in the last year, and condom use with these partners. Since HIV is an incurable disease with long duration from infection to death, past behaviour is also important, but difficulties in obtaining accurate data on earlier partnerships have limited their usefulness. An alternative approach to sexual risk measurement involves identifying critical events (such as first sex, first cohabiting sexual relationship, break-up and re-formation of cohabiting relationships), categorising the various states marked by these transitions and estimating time spent in each. This paper reviews the potential of data from a wide variety of sources (sentinel surveillance in ANC, retrospective enquiries in community-based cross-sectional surveys such as DHS, and longitudinal demographic surveillance systems) to yield useful transition measures, and illustrates methods for individual and community level analyses.