Systematic Difference in HIV Prevalence Rates between Pregnant Women and the General Female Population for Kisesa, Tanzania

Kristin Bietsch, Princeton University
Mark Urassa, National Institute for Medical Research, Tanzania

HIV prevalence rates and trends are often constructed using data collected from pregnant women who visit ante-natal clinics under the assumption that pregnant women are representative of the whole female population. However, women who are HIV positive have lower fertility than women who are HIV negative, thereby downward biasing HIV prevalence estimates from clinic based samples. This article examines the scale of underestimation of HIV prevalence based on pregnant women compared to all women in a rural area in Tanzania. We find that the ratio of prevalence for pregnant women to women in the whole population varies dramatically. Adjusting for age and marital status does not reduce the difference between pregnant women and the whole population. Adjusting for the nulliparous and parous status of women offers significant improvement. Estimates based on antenatal clinic data cannot be taken as representative and adjustments are necessary for estimating HIV prevalence in these populations.