People Who Self-Reported Testing HIV-Positive but Tested HIV-Negative: A Multi-Country Puzzle of Data, Serology, and Ethics, 2015-2021

Abstract

During population-based HIV impact assessments (PHIAs), some participants who self-reported testing HIV-positive (PSRP) tested negative in one or more subsequent survey HIV tests. These unexpected discrepancies between their self-reported results and the survey results draw into question the validity of either the self-reported status or the test results. We analyzed PSRP with negative test results aged 15-59 years old using data collected from 2015 to 2021 in 13 countries, assessing prevalence, self-report status, survey HIV status, viral load, rapid tests and confirmatory tests, and answers to follow-up questions (such as years on treatment). Across these surveys, 19,026 participants were PSRP, and 256 (1.3%) of these were concluded to be HIV-negative after additional survey-based testing and review. PSRP determined to be HIV-negative trended higher in countries with a higher HIV prevalence, but their number was small enough that accepting self-reported HIV-positive status without testing would not have significantly affected the prevalence estimates for HIV or viral load suppression. Additionally, using more detailed information for Uganda, we examined 107 PSRP with any negative test results and found no significant correlation with years on treatment or age. Using these details, we examined support for the possible reasons for these discrepancies beyond misdiagnosis and false reporting. These findings suggest that those conducting surveys would benefit from a nuanced understanding of HIV testing among PSRP to conduct surveys ethically and produce high-quality results.

Keywords: HIV; HIV rapid tests; PHIA; Uganda; algorithms; diagnostic markers; misdiagnosis; population-based surveys; self-report status.

Figure 1

Thirteen countries with public data for population-based HIV impact assessments; 2015: Malawi and Zimbabwe; 2016: Eswatini, Lesotho, Tanzania, and Zambia; 2017: Cameroon, Côte d’Ivoire, Ethiopia, and Namibia; 2018: Kenya and Rwanda; 2016, 2020, and 2021: Uganda, for which pre-published data from the second survey and the refugee survey are used.

Figure 2

National HIV testing algorithm in Uganda, as used for population-based HIV impact assessment surveys in 2016, 2020, and 2021.

Figure 3

Examples of retesting scenarios for self-report discrepancies during population-based HIV impact assessments in multiple countries: (a) all tests negative; (b) possible household testing or recording errors; (c) inconsistent rapid test results; (d) inconsistent Geenius results. RT: rapid test; SR: self-report; TNA PCR: total nucleic acid polymerase chain reaction; TND: Target Not Detected.

Figure 4

Frequency of population-based HIV impact assessment participants self-reporting HIV-positive results with a final survey status of HIV-negative among the participants versus country survey prevalence, with trendline; points are on a log–log scale for visibility. (1) Uganda Refugees survey is an outlier and not included in the trendline.

Figure 5

(a) Age range and (b) years since ARVs were first taken (only asked of PSRP) (for simplicity, only the year is used) among population-based HIV impact assessment participants who self-reported HIV-positive results (PSRP) aged 15–59 years old. Unweighted. Blue line: median; blue box: first quartile (Q1) through third quartile (Q3); whiskers: minimum and maximum. Across three Ugandan surveys from 2016 to 2021. PSRD: PSRP with discrepant (non-positive) test results; PSRP-CN: PSRP concluded to be negative.