Comparing the test performance of dried-blood-spot and plasma HIV recent infection testing samples in a nationally scaled sex worker programme in Zimbabwe

Recency testing can provide strategic insights as to whether a person newly diagnosed with HIV recently acquired their infection or not. To understand potential biases associated with HIV recency testing, we explored the extent sample type influences whether a person is assigned as being recent. Implementing a laboratory-based Recent Infection Testing Algorithm (RITA) across the Centre for Sexual Health and HIV/AIDS Research (CeSHHAR) key populations programme in Zimbabwe between October 2021 and January 2023, we compared plasma-based and dried-bloodspot (DBS) HIV recency samples. Over the study period, 24,976 individual female sex workers HIV tested, of whom 9.5% (2,363/24,979) newly tested HIV positive. Of these 2,363 women, 55.5% (1311/2,363) were offered and gave consent for a sample to be taken for DBS recency and viral load testing, among whom 11.7% (153/1,311) were classified as having a recent infection. A subset of 464 women were offered and consented to paired sample collection, among whom 10.1% (47) and 12.3% (57) of plasma and DBS samples, respectively, were classified as recent. Overall, categorical determination was good, with 97% of results concordant. Of 58 women with paired sample collection who had a test result classified as recent, 46 (79.3%) were concordant recent on both DBS and plasma, with 12 (20.7%) being discordant. Of these 12 women's samples, 11 were deemed long-standing by the plasma assay but recent by the paired DBS, and one deemed long-standing by DBS but recent by the paired plasma sample. On average, plasma samples had a higher normalised optical density than DBS samples (mean difference of 0.53). Depending on use-case and setting, there are trade-offs when considering DBS or plasma-based samples between test performance and ease of implementation. Our data can help inform statistical adjustments to harmonise cut-offs on DBS and plasma assays, thereby improving the use and interpretation of recency assays in population-level HIV surveillance activities.