Abstract
Monitoring progress towards the UNAIDS 'first 90' target requires accurate estimates of levels of diagnosis among people living with HIV (PLHIV), which is often estimated using self-report. We conducted a systematic review and meta-analysis quantifying under-reporting of known HIV-positive status using objective knowledge proxies. Databases were searched for studies providing self-reported and biological/clinical markers of prior knowledge of HIV-positive status among PLHIV. Random-effects models were used to derive pooled estimates of levels of under-reporting. Thirty-two estimates from 26 studies were included (41,465 PLHIV). The pooled proportion under-reporting known HIV-positive status was 20% (95% confidence interval 13-26%, I(2) = 99%). In sub-group analysis, under-reporting was higher among men who have sex with men (32%, number of estimates [N(e)] = 10) compared to the general population (9%, N(e) = 10) and among Black (18%, N(e) = 5) than non-Black (3%, N(e) = 3) individuals. Supplementing self-reported data with biological/clinical proxies may improve the validity of the 'first 90' estimates.