Antenatal clinic HIV data found to underestimate actual prevalence declines: evidence from Zambia.

OBJECTIVE To determine to what extent antenatal clinic (ANC)-based estimates reflect HIV prevalence trends among men and women in a high prevalence urban population. METHODS Examination of data from serial population-based HIV surveys in 1995 (n = 2115), 1999 (n = 1962) and 2003 (n = 2692), and ANC-based surveillance in 1994 (n = 450), 1998 (n = 810) and 2002 (n = 786) in the same site in Lusaka, Zambia. The population-based surveys recorded refusal rates between 6% and 10% during the three rounds. RESULTS Among ANC attendees, prevalence declined by 20% (25.0% to 19.9%; P = 0.101) in the age group 15-24 years and was stable overall. In the general population, the prevalence declined by 49% (P < 0.001) and by 32% (P < 0.001) in age group 15-24 and 15-49, respectively. Among women only, HIV prevalence declined by 44% (22.5% to 12.5%; P < 0.001) and by 27% (29.6% to 21.7%; P < 0.001) in age group 15-24 and 15-49 years, respectively. In addition, prevalence substantially declined in higher educated women aged 15-24 years (20.7% to 8.5%, P < 0.001). CONCLUSION ANC-based estimates substantially underestimated declines in HIV prevalence in the general population. This seemed to be partially explained by a combination of marked differentials in prevalence change by educational attainment and changes in fertility-related behaviours among young women. These results have important implications for the interpretation of ANC-based HIV estimates and underscore the importance of population-based surveys.