What do people actually learn from public health education campaigns? Incorrect inferences about male circumcision and female HIV infection risk in a cluster randomized trial in Malawi

Objective: To examine whether individuals who learn that voluntary medical male circumcision (VMMC) partially reduces female-to-male HIV transmission erroneously infer a reduction in male-to-female HIV transmission risk. Design: Cluster randomised controlled trial. Methods: In 2008, information that VMMC reduces female-to-male HIV transmission risk was randomly disseminated to men in rural Malawi, with follow-up in 2009 (n=917). Data was collected on perceived male and female HIV-transmission risks. We assessed whether beliefs about male circumcision and female HIV-risk varied by receipt of VMMC information and by whether or not individuals believed that VMMC partially protects men from HIV-infection. Results: Men informed about VMMC were more likely to believe that sex with a circumcised male would confer lower transmission risk for women vis-à-vis sex with an uncircumcised male (38% versus 50%, p0.01). Multivariate regression analyses showed that incorrect inferences were most likely to be made by those who believed that circumcised men were partially protected from contracting HIV. Consistent with this, instrumental variable analyses indicated that those individuals who received information about VMMC, and consequently believed it, were 82 percentage points more likely to believe that male circumcision also protects women (p0.01). The inferred reduction in direct HIV infection risk for women due to male circumcision was approximately 50%. Conclusions: Our results suggest the need for VMMC campaigns to make explicit that male circumcision does not directly protect women from HIV-infection. It is also important to assess whether incorrect inferences lead to updated self-perceived HIV-risk and the adoption of riskier sexual behaviours.