Refusal to Be Tested for HIV and the Likelihood of Infection: A Potential, but Unlikely Source of Bias in Community-Based Studies of HIV Prevalence

Georges Reniers, University of Pennsylvania
Tekebash Araya, Addis Ababa University
Yemane Berhane, Addis Ababa University
Eduard J. Sanders, Wellcome Trust
Gail Davey, Addis Ababa University

Population-based studies involving HIV serostatus testing are regularly used to evaluate HIV prevalence estimates derived from ANC sentinel surveillance sites or to provide alternative estimates altogether. Bias in community-based estimates is, however, also plausible because of shortcomings in the sampling frame and non-response due to population mobility and/or refusal. In this paper, we investigate the association between refusal and HIV infection in a large governmental hospital in Addis Ababa, and via regression models that account for sample selection, we quantify the ensuing bias in HIV prevalence estimates. We find that refusal is indeed correlated with the likelihood of infection, but the resulting bias in HIV prevalence estimates –in our study population as well as in community-based studies– is likely to be negligible. The latter will depend in great part on the study protocol and informed consent procedures. We also find that consent for testing increased since the introduction of antiretroviral treatment.