Background:  Emerging data from Sub-Saharan Africa point to possible declines in HIV prevalence in Uganda, Zambia, and preliminarily Zimbabwe. However, estimates are based on antenatal clinic attendees and therefore do not directly measure HIV prevalence in men. Since male sexual risk behaviors drive the epidemic in Sub-Saharan Africa, it is important to establish methods to assess HIV prevalence and sexual risk behavior trends in men. Beerhalls are valued social institutions, well attended by a broad section of Zimbabwean male society and may serve as potential sentinel sites for monitoring the HIV epidemic among men. From 2000 to 2004, we conducted three surveys of male beerhall patrons in Harare, Zimbabwe.

Methods:  Randomized, time-location cross-sectional surveys and serological HIV testing were conducted in 2000 (n = 324), 2002 (n = 1313), and 2004 (n = 1236) among male attendees of 43 beerhalls. Analysis of the trend in HIV prevalence across the 3 surveys used the Cochrane-Armitage test. The change in risk behaviors between the 2002 and 2004 surveys, which collected comparable measures, was assessed using the c² test.

Results:  HIV prevalence remained stable across the 3 surveys at 30% in 2000, 28% in 2002, and 29% in 2004 (p = 0.92). HIV prevalence was highest among men aged 35 to 39 years. Risk behavior did not change between the 2002 and 2004 surveys with respect to having an additional partner aside from the regular partner (50% vs 49%; p = 0.46), unprotected sex with a non-wife partner (32% vs 33%; p = 0.40), sex with a non-wife partner while drunk (39% vs 39%; p = 0.80), and meeting a sex partner at a beerhall (27% vs 26%; p = 0.34). HIV infection was associated with older age (p < 0.001) and a self-report of penile sores/ulcers (p < 0.01) in all 3 surveys. HIV infection was also associated with meeting a sex partner at a beerhall in the 2000 and 2002 surveys (p < 0.04), with unprotected sex with any non-wife, casual, or anonymous partner in the 2002 survey (p = 0.03), and sex with a non-wife partner while drunk in the 2000 (p = 0.04) and 2004 (p = 0.03) surveys.

Conclusions:  HIV prevalence among male beerhall patrons in Harare has remained stable across the 2000, 2002, and 2004 surveys, corroborated by the lack of change in sexual risk behaviors from 2002 to 2004. While the apparent leveling off in the epidemic is hopeful, HIV prevalence remains high and underscores the urgent need for continued risk reduction efforts.

Keywords: HIV prevalence; sexual risk behavior; developing countries