Table 3.
Effect size estimates for HIV testing, PrEP refilling, and PrEP adherence, by study population and un-pooled intervention groups
| Oral-fluid HIVST1 vs. Standard of care | Blood-based HIVST1 vs. Standard of care | Oral-fluid HIVST1 vs. Blood-based HIVST1 | |
|---|---|---|---|
| All participants (N=495) | RD (1-sided 95% CI lower bound) 4 | RD (1-sided 95% CI lower bound) 4 | RD (1-sided 95% CI lower bound) 4 |
| Returned to clinic2 | |||
| Tested for HIV | 0·72% (−5·74%) | −3·24% (−10·03%) | 3·96% (−2·78%) |
| Refilled PrEP | −1·18% (−8·39%) | −4·31% (−11·70%) | 3·07% (−4·39%) |
| Adherent (any TFV-DP detected) | −0·62% (−9·01%) | 4·67% (−3·78%) | −5·74% (−14·45%) |
| HIV serodifferent couples3 (N=295) | RD (1-sided 95% CI lower bound) 4 | RD (1-sided 95% CI lower bound) 4 | RD (1-sided 95% CI lower bound) 4 |
| Returned to clinic2 | |||
| Tested for HIV | 0·42% (−7·66%) | −6·43% (−15·18%) | 6·79% (−1·93%) |
| Refilled PrEP | 0·28% (−9·07%) | −3·13% (−12·76%) | 3·33% (−6·25%) |
| Adherent (any TFV-DP detected) | −8·37% (−18·88%) | −7·84% (−18·28%) | −0·63% (−11·65%) |
| All women3 (N=330) | RD (1-sided 95% CI lower bound) 4 | RD (1-sided 95% CI lower bound) 4 | RD (1-sided 95% CI lower bound) 4 |
| Returned to clinic2 | |||
| Tested for HIV | 3·51% (−4·65%) | 0·76% (−7·67%) | 2·74% (−5·39%) |
| Refilled PrEP | 0·35% (−8·59%) | −1·21% (−10·26%) | 1·53% (−7·49%) |
| Adherent (any TFV-DP detected) | 7·19% (−3·45%) | 16·29% (5·66%) | −9·26% (−20·09%) |
| Women singly enrolled3 (N=200) | RD (2-sided 95% CI) 5 | RD (2-sided 95% CI) 5 | RD (2-sided 95% CI) 5 |
| Returned to clinic2 | |||
| Tested for HIV | 1·24% (−11·61%, 14·09%) | 1·49% (−11·27%, 14·26%) | −0·25% (−12·88%, 12·38%) |
| Refilled PrEP | −3·30% (−16·78%, 10·17%) | −5·97% (−19·70%, 7·76%) | 2·67% (−11·53%, 16·86%) |
| Adherent (any TFV-DP detected) | 1·26% (−3·74%, 28·93%) | 26·87% (10·65%, 43·08%) | −14·27% (−31·09%, 2·55%) |
Risk differences (RDs) measured using binomial regression models with identity links, adjusted for study population at enrollment (e.g., men in HIV serodifferent couples, women in HIV serodifferent couples, and women singly enrolled).
Participants randomized to these intervention arms received six-month PrEP dispensing + interim HIVST (either oral-fluid or blood-based) with biannual clinic visits.
We included all follow-up visits assigned as six-month visits by study staff; the window for these visits closed 2 weeks prior to the next scheduled visit date, which as at nine months for participants randomized to the SOC arm and 12 months for participants randomized to one of the intervention arms.
The sub-group HIV serodifferent couples included all men and women enrolled in HIV serodifferent couples (N=295). The sub-group all women included both women enrolled in HIV serodifferent couples and women singly enrolled (N=330). The sub-group women singly enrolled just included these women and not those enrolled in HIV serodifferent couples (N=200).
In our non-inferiority analyses, we interpreted one-sided 95% CIs above −10% as non-inferior.
In our superiority analyses, we interpreted two-sided 95% CIs that did not include zero as significant.
CI: confidence interval; HIVST: HIV self-testing; PrEP: pre-exposure prophylaxis; RD: risk difference; TFV-DP: tenofovir-diphosphate