Uptake, engagement, and adherence to pre-exposure prophylaxis offered after population HIV testing in rural Kenya and Uganda: 72-week interim analysis of observational data from the SEARCH study

Abstract

Background: Optimal strategies for pre-exposure prophylaxis (PrEP) engagement in generalised HIV epidemics are unknown. We aimed to assess PrEP uptake and engagement after population-level HIV testing and universal PrEP access to characterise gaps in the PrEP cascade in rural Kenya and Uganda.

Methods: We did a 72-week interim analysis of observational data from the ongoing SEARCH (Sustainable East Africa Research in Community Health) study. Following community sensitisation and PrEP education, we did HIV testing and offered PrEP at health fairs and facilities in 16 rural communities in western Kenya, eastern Uganda, and western Uganda. We provided enhanced PrEP counselling to individuals 15 years and older who were assessed as having an elevated HIV risk on the basis of serodifferent partnership or empirical risk score, or who otherwise self-identified as being at high risk but were not in serodifferent partnerships or identified by the risk score. PrEP follow-up visits were done at facilities, homes, or community locations. We assessed PrEP uptake within 90 days of HIV testing, programme engagement (follow-up visit attendance at week 4, week 12, and every 12 weeks thereafter), refills, self-reported adherence up to 72 weeks, and concentrations of tenofovir in hair samples from individuals reporting HIV risk and adherence during follow-up, and analysed factors associated with uptake and adherence. This study is registered with ClinicalTrials.gov, NCT01864603.

Findings: Between June 6, 2016, and June 23, 2017, 70 379 community residents 15 years or older who had not previously been diagnosed with HIV were tested during population-level HIV testing. Of these individuals, 69 121 tested HIV-negative, 12 935 of whom had elevated HIV risk (1353 [10%] serodifferent partnership, 6938 [54%] risk score, 4644 [36%] otherwise self-identified risk). 3489 (27%) initiated PrEP, 2865 (82%) of whom did so on the same day as HIV testing and 1733 (50%) of whom were men. PrEP uptake was lower among individuals aged 15-24 years (adjusted odds ratio 0·55, 95% CI 0·45-0·68) and mobile individuals (0·61, 0·41-0·91). At week 4, among 3466 individuals who initiated PrEP and did not withdraw or die before the first visit, 2215 (64%) were engaged in the programme, 1701 (49%) received medication refills, and 1388 (40%) self-reported adherence. At week 72, 1832 (56%) of 3274 were engaged, 1070 (33%) received a refill, and 900 (27%) self-reported adherence. Among participants reporting HIV risk at weeks 4-72, refills (89-93%) and self-reported adherence (70-76%) were high. Among sampled participants self-reporting adherence at week 24, the proportion with tenofovir concentrations in the hair reflecting at least four doses taken per week was 66%, and reflecting seven doses per week was 44%. Participants who stopped PrEP accepted HIV testing at 4274 (83%) of 5140 subsequent visits; half of these participants later restarted PrEP. 29 participants of 3489 who initiated PrEP had serious adverse events, including seven deaths. Five adverse events (all grade 3) were assessed as being possibly related to the study drug.

Interpretation: During population-level HIV testing, inclusive risk assessment (combining serodifferent partnership, an empirical risk score, and self-identification of HIV risk) was feasible and identified individuals who could benefit from PrEP. The biggest gap in the PrEP cascade was PrEP uptake, particularly for young and mobile individuals. Participants who initiated PrEP and had perceived HIV risk during follow-up reported taking PrEP, but one-third had drug concentrations consistent with poor adherence, highlighting the need for novel approaches and long-acting formulations as PrEP roll-out expands.

Funding: National Institutes of Health, President's Emergency Plan for AIDS Relief, Bill & Melinda Gates Foundation, and Gilead Sciences.

Figure 1

PrEP uptake after population-level HIV testing in 16 communities in rural Kenya and Uganda PrEP=pre-exposure prophylaxis. SEARCH=Sustainable East Africa Research in Community Health. *Empirical risk score developed on the basis of applying ensemble supervised machine learning methods to HIV seroconversion data from the first 2 years of the SEARCH test-and-treat trial, with a threshold selected to correctly classify 50% of seroconversions as at elevated risk across the three regions and minimise the number of individuals classified as at risk. Variables: age, sex, marital status, polygamy, education, occupation, alcohol, and circumcision. †Individuals neither in serodifferent partnerships nor identified by the risk score could self-identify as at risk of HIV acquisition. ‡82% initiated PrEP on the same day as seen during population-level HIV testing.

Figure 2

PrEP programme engagement, refill, and adherence overall and by self-assessed current HIV risk up to week 72 (A) PrEP programme engagement, refill, and self-reported adherence among PrEP initiators. (B) Refill and self-reported adherence among PrEP participants reporting self-assessed current HIV risk. PrEP=pre-exposure prophylaxis. *Programme engagement is defined as attendence at a PrEP follow-up visit during scheduled visit weeks; eligibility for visit excludes participants who were withdrawn or died before the visit. †Self assessed current HIV risk was evaluated at each follow-up visit among individuals engaged in the PrEP programme.

Figure 3

PrEP programme engagement, refill, and adherence at week 24 among demographic subgroups of participants who initiated PrEP (A) PrEP programme engagement, refill, and self-reported adherence among risk groups of participants who initiated PrEP. (B) Refill and self-reported adherence among PrEP participants reporting self-assessed current HIV risk. PrEP=pre-exposure prophylaxis. *Programme engagement is defined as attendence at a PrEP follow-up visit during scheduled visit weeks; eligibility for visit excludes participants who were withdrawn or died before the visit. †Self-assessed current HIV risk was evaluated at each follow-up visit among individuals engaged in the PrEP programme. ‡Mobile individuals could be in any age group.

Figure 4

Adherence to PrEP estimated from the concentration of tenofovir in hair samples Adherence at (A) week 4 (n=166) and (B) week 24 (n=152) among subgroups of sampled participants reporting self-assessed current HIV risk and any PrEP adherence (at least one dose taken of the past three). BLQ=below the limit of quantification.