Bias in HIV Prevalence Estimates from Refusals to Be Tested in Seroprevalence Surveys

Jeff Eaton, University of Washington

Nationally-representative HIV seroprevalence surveys are increasingly being relied upon for HIV prevalence estimates. We explore the potential for bias in these estimates because of non-response due to the refusal to be tested. The few studies on this topic have failed to identify any substantial bias, but they typically ignore bias due to refusals that are informed by prior knowledge about one’s HIV status. In a sample of respondents from Malawi that had been tested before, we find that HIV positives are five times more likely to refuse a subsequent test than HIV negatives. We use this parameter in simulations that further rely on empirical data from the Demographic and Health Surveys and demonstrate that this factor alone may lead to significant bias in HIV prevalence estimates; particularly in urban areas where HIV prevalence, refusal rates, and coverage of VCT are often higher.