Home-based oral self-testing for absent and declining individuals during a door-to-door HIV testing campaign in rural Lesotho (HOSENG): a cluster-randomised trial

In sub-Saharan Africa, home-based HIV testing is validated and accepted, but coverage is low because household members are often absent during home-based testing campaigns. We aimed to measure the effect of a secondary distribution of oral-fluid HIV self-tests on coverage during home-based testing in rural Lesotho.; The Home-Based Self-Testing (HOSENG) trial was a cluster-randomised, non-blinded superiority trial in rural villages in the catchment area of 20 health facilities of two districts in Lesotho (Butha-Buthe and Mokhotlong). Eligible villages had a consenting village chief and at least one registered village health worker; eligible households had a consenting representative aged 18 years or older. The HOSENG trial provided a recruitment platform for the interlinked Village-Based Refill of Antiretroviral Therapy (VIBRA) trial. Villages were randomly assigned 1:1:1:1 with block sizes of four to one of four groups: VIBRA control and HOSENG control; VIBRA control and HOSENG intervention; VIBRA intervention and HOSENG control; and VIBRA intervention and HOSENG intervention. Randomisation was stratified by district, village size, and access to the nearest health facility. An independent statistician was responsible for the computer-generated randomisation list. In the intervention group, oral-fluid HIV self-tests were left for absent or declining household members (aged ≥12 years) during a home visit from the HIV testing campaign team. One present household member was trained on self-test use. Distributed self-tests were followed up by village health workers. In control village clusters, absent or declining household members were referred to the clinic for HIV testing. The primary outcome was HIV testing coverage among all household members aged 12 years or older within 120 days, defined as a confirmed HIV test result or known status, reported in testing registers at the health facilities or on the follow-up forms of the village health worker. Adjusted random-effects logistic regression with individuals as the unit of analysis was used. This trial is registered with ClinicalTrials.gov, NCT03598686.; Between July 26, 2018, and Dec 12, 2018, 3091 consenting households with 7816 household members aged 12 years or older were enrolled and randomly assigned (intervention: 57 village clusters, 1620 households, 4174 household members; control: 49 village clusters, 1471 households, 3642 household members). In the control group, 38 (3%) of 1455 initially absent or declining household members tested at a clinic within 120 days. In the intervention group, 841 (53%) of 1601 initially absent or declining household members had a confirmed status within 120 days; 12 (1%) of 841 tested at the clinic and 829 (99%) used their self-test kit. This resulted in a testing coverage of 2201 (60%) of 3642 in the control group versus 3386 (81%) of 4174 in the intervention group (odds ratio 3·00 [95% CI 2·52-3·59]; p