Study of bias in antenatal clinic HIV-1 surveillance data in a high contraceptive prevalence population in sub-Saharan Africa Simon Gregson a,b , Nicola Terceira c , Memory Kakowa b , Peter R. Mason b , Roy M. Anderson a , Stephen K. Chandiwana b and Michel Carae Èl d Objective: To describe patterns, sources and consequences of bias in antenatal clinic (ANC) HIV prevalence estimates in a high contraceptive prevalence population. Background: HIV surveillance in Africa relies on data from pregnant women attend- ing ANCs. HIV estimates from pregnant women understate female infection levels in low income, high fertility populations. Bias in high contraceptive use, delayed sexual debut populations remains undescribed. Design and method: Comparison of parallel cross-sectional population and antenatal survey data from rural Zimbabwe, where 60% of women are recent contraceptive users. Results: HIV prevalence in recently pregnant women (25.7%; n  576) and all women (25.5%; n  5138) is similar over the age-range 15±44 years. As in high fertility populations, HIV prevalence is higher in pregnant women at young ages and lower at older ages but the crossover point occurs later due to delayed sexual activity. HIV understatement at older ages due to HIV-associated infertility is mitigated by less HIV infection and less frequent ANC attendance in contraceptive users. The local ANC HIV prevalence estimate is lower [21.2%; n  1215; risk ratio versus pregnant women in the general population, 0.8; 95% con®dence interval (CI), 0.7±1.0], possibly because women from more remote areas are included. ANC estimates overstate the relative risk of HIV in more educated women (age-adjusted odds ratio, 1.1; 95% CI, 0.8 ± 1.4 versus 0.7; 95% CI, 0.6 ± 0.9). Conclusions: ANC estimates understate female HIV prevalence in this low fertility population but, here, the primary cause is not selection of pregnant women. ANC estimate adjustment procedures that control for contraceptive use and age at ®rst sex are needed. & 2002 Lippincott Williams & Wilkins AIDS 2002, 16:643±652 Keywords: HIV, surveillance, family planning, bias, fertility, Zimbabwe Introduction Sentinel surveillance systems in sub-Saharan Africa are essential for monitoring trends in HIV infection and for assessing future socio-demographic impact of local and global HIV epidemics [1]. For reasons of convenience, these systems rely almost exclusively on data from pregnant women attending antenatal clinics (ANC) From the a Department of Infectious Disease Epidemiology, Imperial College School of Medicine, London, UK, b Biomedical Research and Training Institute, Harare, Zimbabwe, the c Wellcome Trust Centre for the Epidemiology of Infectious Disease, Zoology Department, University of Oxford, UK, and d UNAIDS, Geneva. Requests for reprints to: S. Gregson, Department of Infectious Disease Epidemiology, Imperial College School of Medicine, Norfolk Place, London W2 1PG, UK. Received: 23 February 2001; revised: 5 September 2001; accepted: 13 September 2001. ISSN 0269-9370 & 2002 Lippincott Williams & Wilkins 643