Collecting data on HIV/AIDS mortality during household surveys: A randomized validation study in Malawi

Background: In countries with limited civil registration, household surveys do not include questions on HIV-related mortality. Objective: We aim to test whether accurate data on the HIV status of the deceased siblings of survey respondents could be collected during household surveys conducted in low- and middle-income countries. Methods: We added HIV questions to the standard module on adult and maternal mortality used in household surveys. We compared the data generated by these questions to data collected by the Karonga Health and Demographic Surveillance System in Malawi. We randomly assigned men and women aged 15–59 years to face-to-face (FTF) interviewing or the more confidential method of audio computer-assisted self-interviewing (ACASI). We compared survey reports of adult deaths to prospective reference data on mortality and HIV status. Results: The added HIV questions yielded nearly complete data on the HIV status of deceased siblings, particularly those who had died recently. The sensitivity of survey data on the HIV status of the deceased was high in both study groups, even after accounting for partial verification bias (>0.75). There were few false negative reports of the HIV status of deceased siblings (specificity > 0.95). ACASI did not improve the accuracy of survey data, but it required extensive training of respondents, and more time to collect mortality data. Conclusions: Adding HIV questions to instruments used to measure mortality in LMICs is feasible and generates accurate data on the HIV status of deceased adults. Contribution: The data generated by these new HIV-related questions might help to better track progress towards global HIV elimination targets.