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

1Despite remarkable progress in scaling up HIV testing and treatment worldwide, 1·7 million new HIV infections occurred in 2018, signalling what UNAIDS has termed a prevention crisis. Oral pre-exposure prophylaxis (PrEP) is highly effective and has the potential to accelerate reductions in HIV incidence as part of combination approaches to HIV prevention. Reduced HIV incidence has been observed in high-income populations of men who have sex with men for whom antiretroviral therapy (ART) and PrEP have been scaled-up., PrEP is recommended in global guidelines and is being introduced across southern and eastern Africa, where the burden of new infections is the greatest.

2In generalised epidemic settings, strategies are needed to identify and engage individuals who might benefit from HIV prevention services, including PrEP. Most studies in sub-Saharan Africa have offered PrEP to prespecified populations who are at high risk, such as young women, sex workers, and serodifferent couples, who either present to clinical care or are selectively recruited from the community. Comprehensive HIV testing and universal access to PrEP for all people without HIV who are at high risk of infection presents an opportunity to offer PrEP to people who might not otherwise engage in health systems, and to assess PrEP engagement at a population level.

3We offered PrEP during population-level HIV testing in an ongoing study in rural Kenya and Uganda, using an inclusive approach to PrEP eligibility, with options for same-day or rapid PrEP start and community-based or facility-based service delivery. We aimed to estimate the proportion of the entire population at elevated HIV risk to establish a denominator for calculations of population-level PrEP uptake; assess programme engagement, medication refill, and adherence (measured by self-report and objective metrics); and identify gaps in the PrEP cascade.

4In this analysis, we report interim observational data from an ongoing study of a PrEP intervention in 16 communities in three regions of eastern Africa with varying HIV prevalence among individuals aged 15 years or older: western Kenya (19%), eastern Uganda (4%), and western Uganda (7%). As part of the second phase of the Sustainable East Africa Research in Community Health (SEARCH) study, we did population-level HIV testing and implemented a PrEP intervention before national PrEP roll-out in Uganda and Kenya. The study population included adults (aged ≥15 years) in the SEARCH study. Here, we report an interim analysis up to 72 weeks of follow-up to address knowledge gaps in the population-level PrEP cascade in the context of roll-out and scale-up in the region.

5Ethical approval for the study was by the institutional review boards of Makerere University (Kampala, Uganda), Kenya Medical Research Institute (Nairobi, Kenya), and University of California San Francisco (San Francisco, CA, USA). All participants provided verbal consent; PrEP participants provided written informed consent in their preferred language.

6We did population-level HIV and multi-disease testing using a hybrid mobile testing approach. We held health fairs at multiple locations across each community over 2 weeks, followed by home-based testing for non-attendees. 1 month before testing events, we conducted community mobilisation and sensitisation on PrEP in collaboration with health officials and government leaders. Study staff disseminated information on PrEP at local events and meetings, including with political and religious leaders; health workers; patients with HIV; adolescents and young adults; parents; and fishing, bar, and transport workers (occupations associated with higher HIV prevalence or transactional sex).,

7Community members received basic group-based education on PrEP upon arrival at health fairs, then attended stations for collection of sociodemographic information, HIV testing, and post-test counselling. During post-test counselling, we offered enhanced individual PrEP counselling to people who tested negative for HIV and had an elevated risk of HIV acquisition: individuals in serodifferent partnerships; individuals classified as being at risk on the basis of an empirical HIV risk-prediction algorithm (risk score); or individuals who were neither in serodifferent partnerships nor identified by risk score, but who otherwise self-identified as being at risk. Serodifferent partnerships were self-identified male-female spouses of differing HIV status or individuals who self-reported having a partner with HIV. As previously described,, the risk score was developed by applying ensemble machine learning to HIV seroconversion and sociodemographic data from the first 2 years of the SEARCH test-and-treat trial (appendix p 1). Data on sexual behaviour were not previously collected and were therefore not available for risk-score development. Among individuals not known to be I n serodifferent partnerships, the score analysed region and socio-demographicf characteristics (age, sex, marital status, polygamy, education, occupation, alcohol consumption, and circumcision) entered in real-time into tablet computers at health fairs before the HIV-testing station. Health-fair staff recorded a dichotomous output (at risk or not at risk) on a paper card for the participant to give to the HIV post-test counsellor.

8Post-test counsellors discussed individuals’ potential HIV risk (eg, knowledge of partner status, concurrent partnerships, condom use, circumcision). Individuals assessed to be at an elevated HIV risk (based on serodifferent partnership, risk score, or otherwise self-identified risk) received enhanced counselling on PrEP, including information on who might benefit, how PrEP is taken, and potential adverse effects. We offered same-day or rapid PrEP initiation through SEARCH at local government HIV clinics (with one-time, study-provided transport). In 14 of 16 communities, on-site PrEP start was also offered at SEARCH health fairs. For individuals who received home-based testing, one study staff member did all of the described steps, including provision of information about PrEP, risk-score assessment, HIV tesing and post-test counselling, and offer of PrEP through local clinics (but not on-site at home).

9During population-level testing, we measured HIV-1 RNA among individuals with HIV. We assessed viral suppression (≤500 RNA copies per mL) among linked HIV-positive spouses in serodifferent partnerships, but real-time results were not available for PrEP counselling. After population-level testing, we offered ongoing PrEP initiation through SEARCH at local clinics, where we did intensive outreach. Patients with HIV were asked to bring their partners who were HIV-negative or with unknown HIV status to the clinic for HIV testing and the offer of PrEP initiation.

10PrEP eligibility criteria included negative HIV testing within 4 weeks, no known hepatitis B infection, and no acute HIV symptoms. Baseline creatinine testing was done but PrEP was provided at enrolment before the receipt of laboratory results. After providing written informed consent, participants were given tenofovir disoproxil fumarate (300 mg) co-formulated with emtricitabine (200 mg) or lamivudine (150 mg). Follow-up visits were scheduled at week 4, week 12, and every 12 weeks thereafter for up to 144 weeks. We provided a supportive delivery system with options for visits at clinic, home, or community locations of the participants’ choice (eg, homes, near schools, trading centres, beaches). Visit procedures included evaluation of adverse events, self-assessed HIV risk (“are you currently at risk for HIV?”), HIV antibody testing, and PrEP refill, plus creatinine testing at week 12, week 24, and every 24 weeks thereafter. Participants who stopped PrEP were offered HIV testing and the opportunity to restart PrEP at each visit. Study drug and procedures were offered at no cost to the participants; no incentives were provided.

11For adherence measurement at follow-up visits, we assessed self-reported adherence using 3-day recall, a feasible-to-collect, short-term measure. Given the limitations of self-reporting (eg, recall and social desirability biases, white-coat dosing) among participants reporting any adherence to PrEP (at least one dose of the past three), we collected small hair samples (around 50–100 strands) at each visit (week 4, week 12, and every 12 weeks thereafter) for adherence measurement via analysis of concentrations of tenofovir. We analysed tenofovir concentrations in hair among a random sample of participants who self-reported PrEP adherence and HIV risk at week 4, and in a random sample of participants who self-reported adherence and risk at week 24; sampling was stratified by age, sex, serodifferent partnership, and occupation (fishing, bar, or transport). 1 cm of hair closest to the scalp (reflecting the most recent 4 weeks of drug exposure) was analysed via liquid chromatography-tandem mass spectrometry using validated methods.

12We calculated the following at the different steps of the PrEP cascade: (1) among all residents without HIV seen during population-level HIV testing, the proportion at elevated HIV risk (based on mutually exclusive categories, defined by serodifferent partnership, risk score without serodifferent partnership, or otherwise self-identified risk); (2) among individuals at elevated risk, the proportion with PrEP uptake, defined as initiation (receipt of pills) within 90 days of HIV testing during population-level testing; (3) among individuals who initiated PrEP within 90 days, programme engagement up to 72 weeks. Programme engagement was defined as follow-up visit attendance during periods defined on the basis of the date of PrEP enrolment: week 4 (−14/+28 days); week 12 (−27/+42 days); and weeks 24–72 (−41/+42 days). We censored at death or study withdrawal.

13At each follow-up visit, we assessed the proportion of individuals who (1) received PrEP medication refills and (2) self-reported adherence to PrEP (at least one dose of the past three scheduled doses, assuming the participants who were not seen were non-adherent), both among all individuals who initiated PrEP and among individuals reporting current HIV risk. We also assessed adherence on the basis of tenofovir concentrations in hair from a sample of individuals to estimate the number of PrEP doses taken per week., Among individuals who stopped PrEP (no refill in visit period or refill ≥30 days late), we assessed the proportion of subsequent visits where HIV testing was done, and the proportion of participants who restarted PrEP.

14We used mixed effects logistic regression with community as random effect and variances adjusted for clustering at the community level to identify factors associated with PrEP uptake (overall and stratified by sex) among participants at elevated HIV risk; week-24 self-reported adherence among individuals who initiated PrEP; and week-4 tenofovir concentrations in the hair of participants reporting risk using inverse weighting to account for sampling for measurement of tenofovir concentrations (appendix p 5) using Stata version 15. The SEARCH study is registered with ClinicalTrials.gov, NCT01864603.

15The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.

16Between June 6, 2016, and June 23, 2017, 70 379 (83%) of 85 047 community residents 15 years or older and not previously diagnosed with HIV were tested during population-level HIV testing. Of the individuals who were tested, 69 121 residents tested negative for HIV, of whom 12 935 were assessed as having an elevated risk of HIV acquisition on the basis of serodifferent partnership (1353 [10%]), risk score (6938 [54%]), or otherwise self-identified risk (4644 [36%]) and were targeted for enhanced individual counselling on PrEP (figure 1). 820 (61%) of 1353 individuals in serodifferent partnerships and 2316 (32%) of all 7256 individuals identified by risk score also self-identified as being at risk (appendix p 2).

17We measured PrEP uptake among the 12 935 people testing negative for HIV who were assessed to be at elevated HIV risk: 3489 (27%) initiated PrEP within 90 days of population-level testing; 2865 (82%) of whom started PrEP on the same day as testing. Among individuals who initiated PrEP, 1733 (50%) were men, 2175 (62%) were younger than 35 years, and 3395 (97%) were tested at health fairs rather than at home (table 1). Compared with women who initiated PrEP, men were more likely to be unmarried, use alcohol, and work in fishing, bar, or transport occupations (appendix p 3). Among individuals who were assessed to be at elevated HIV risk, 446 (19%) of 2376 adolescent girls and young women aged 15–24 years, 1226 (17%) of the 7256 individuals identified by risk score, and 3374 (45%) of 7581 individuals with self-identified HIV risk initiated PrEP (appendix p 4). Of 1353 serodifferent partners, 603 (45%) initiated PrEP (235 [39%] of 603 men; 368 [49%] of 750 women); 192 (26%) of 750 who did not initially start PrEP started after 90 days. Of 1014 HIV-negative residents linked to a serodifferent spouse, 729 (72%) were in a partnership for at least 3 years with a spouse living with HIV. Among 811 spouses living with HIV with HIV-RNA measured during population-level testing, 246 (82%) of 299 spouses of individuals who initiated PrEP had viral suppression (≤500 copies per mL) compared with 410 (80%) of 512 spouses of participants who had not initiated PrEP. Among individuals who initiated PrEP and were linked to a serodifferent spouse, 238 (68%) of 351 (165 [71%] of 231 women; 73 [61%] of 120 men) reported a serodifferent partnership at PrEP enrolment, suggesting that disclosure of HIV status might have occurred.

18We analysed predictors of PrEP uptake among 12 850 individuals assessed to have elevated HIV risk with complete covariate data (table 2). The adjusted odds of PrEP uptake were significantly higher among individuals in serodifferent partnerships, polygamous marriages, or who were divorced, separated, or widowed, and were significantly lower among participants aged 15–24 years and 25–34 years, and among mobile individuals (mobility was defined as migration out of the community for at least 1 month, or moving residence within the past 12 months). In analyses stratified by sex, non-single marital status was associated with PrEP uptake for both men and women (table 3). Among women, the adjusted odds of PrEP uptake were significantly lower for individuals with fishing, bar, or transport work, alcohol consumption, or mobility, and were higher among serodifferent partners.

19We measured PrEP programme engagement, refills, and adherence up to June 20, 2019, among 3466 individuals (excludes participants who were withdrawn or died before the first visit) who had initiated PrEP; 2693 (78%) of 3466 attended at least one follow-up visit. At week 4, 2215 (64%) individuals who initiated PrEP were engaged in the PrEP programme; 1701 (49%) received PrEP refills; and 1388 (40%) self-reported adherence to PrEP (at least one dose of the past three), of whom 1268 (91%) reported that all of the past three doses were taken (figure 2A). Programme engagement, refills, and self-reported adherence declined until week 24, then stabilised up to week 72. At week 72, 1832 (56%) of 3274 were engaged, 1070 (33%) received a refill, and 900 (27%) self-reported adherence. Over time, more follow-up visits occurred at home and community sites (vs at facilities): 1311 (59%) of 2215 at week 4, 1443 (76%) of 1912 at week 24, and 1459 (80%) of 1832 at week 72. In multivariable analyses of factors associated with self-reported PrEP adherence at week 24, self-assessed current HIV risk was associated with the greatest odds of adherence; serodifferent partnership, or being divorced, separated, or widowed were positively associated with adherence; and being aged 15–24 years was negatively associated with adherence (table 4).

20Among the subset of individuals who initiated PrEP and reported current HIV risk during follow-up, a high proportion received PrEP refills and self-reported adherence. At weeks 4–72, 89–93% received refills and 70–76% self-reported adherence (figure 2B). Self-assessed risk also changed over time: 1462 (54%) of 2693 participants seen in follow-up reported no risk at at least one visit. Of 1699 participants reporting current HIV risk at week 4, 330 (19%) reported no risk at week 24; conversely, 96 (20%) of 471 participants who reported no risk at week 4 reported risk at week 24.

21Among subpopulations considered to be at higher risk of HIV, we examined cascade outcomes at week 24. adults reported current HIV risk versus 219 (86%) of 256 women with serodifferent partners (figure 3B).

22We analysed concentrations oftenofovir in hair samples from a sample of individuals who initiated PrEP and who reported current risk and any adherence. After accounting for sampling weights for 120 participants with tenofovir concentrations measured at week 4, 57% (95% CI 47– 67) had tenofovir concentrations consistent with taking at least four PrEP doses per week; 40% (31–51) had tenofovir concentrations consistent with taking seven PrEP doses per week (figure 4). Adherence of at least four doses per week was highest among men aged 25 years and older (77%, 58–90) and lowest among young women aged 15–24 years (27%, 12–46). Participants sampled at week 24 had concentrations of tenofovir in hair samples consistent with similar or higher numbers of doses taken per week compared with those sampled at week 4, both overall and in most subgroups, including young women. After accounting for sampling weights for 116 participants with concentrations of tenofovir in hair samples measured at week 24, 66% (55–76) had tenofovir concentrations consistent with taking at least four doses per week, and 44% (33–55) had tenofovir concentrations consistent with taking seven doses per week. However, one-third (34%, 24–45) of individuals who initiated PrEP had drug concentrations that suggest poor adherence (fewer than four doses per week). Among individuals who initiated PrEP and reported risk at week 4, female sex and being aged 15–24 years were associated with lower odds of having concentrations of tenofovir consistent with seven doses per week; serodifferent partnership was associated with the greatest odds (appendix p 5). When analysed as a continuous outcome, higher tenofovir concentrations were associated with serodifferent partnership, whereas younger age was associated with lower tenofovir concentrations (appendix p 6).

23Many individuals who initiated PrEP stopped the drug and later restarted. Among 2693 participants seen in follow-up, 2205 (82%) stopped PrEP at least once; 1096 (50%) restarted by week 72. Among participants who restarted, 201 (18%) stopped and restarted more than once and the median time off PrEP was 68 days (IQR 37–188). HIV testing was done at 4274 (83%) of 5140 visits where participants had stopped PrEP before the visit or did not receive a refill.

2429 of 3489 participants who initiated PrEP had serious adverse events, including seven deaths (table 5, appendix p 7). Five adverse events (all grade 3) were assessed as being possibly related to the study drug. One grade 3 creatinine elevation occurred in a 71-year-old man who was treated in hospital for urinary retention and hydronephrosis. Creatinine returned to baseline following relief of urinary obstruction and cessation of the study drug.

25During HIV testing of more than 70000 individuals across three regions of rural Kenya and Uganda, we evaluated a population-based approach to engage people in PrEP that included community sensitisation, group-based education, individual counselling, real-time risk-score assessment, and universal PrEP access with flexibility in refill location. More than 12 000 individuals were assessed to be at elevated risk of HIV, of whom nearly 3500 initiated PrEP within 90 days of population-level HIV testing. The greatest gap in the PrEP cascade was initial uptake, with only 27% starting PrEP; adolescents, young adults, and mobile individuals had the lowest uptake. Programme engagement up to 72 weeks was high (56%) among individuals who initiated PrEP, with 78% of these individuals having at least one follow-up visit and most engaging in visits for HIV testing. Receipt of PrEP refills and self-reported adherence were much higher among participants who perceived themselves to be at ongoing risk at follow-up visits. However, one-third of participants who self-reported adherence had drug concentrations that indicated low adherence.

26We posited that universal access to PrEP and inclusive risk assessment could provide an opportunity for PrEP initiation for individuals who might not otherwise engage with health systems. Moreover, given concerns that targeting PrEP towards specific risk groups could stigmatise use, offering PrEP more broadly could foster uptake. Although we have no comparator for other population-level approaches, only one-quarter of individuals assessed as at elevated HIV risk initiated PrEP within 90 days. Our study rapidly introduced PrEP in communities before national roll-out in Kenya and Uganda, and the novelty of PrEP might have mitigated against higher uptake. We have previously reported on barriers to PrEP uptake in study communities, including myths and misconceptions about this new drug, unsupportive partners, daily pilltaking, fear of side-effects, and anticipated stigma, similar to previous studies. We have also found that distance to clinic is a barrier to PrEP uptake and engagement. Moreover, because an individual’s HIV risk is potentially dynamic, some individuals assessed as being at elevated HIV risk at screening might not have had current sexual risk factors or partners at the time of PrEP screening. Finally, although we offered universal PrEP access, our approach focused on risk assessment. Wellness framing of PrEP (as fostering protection and sexual health and wellbeing) might be more successful than risk framing and could be evaluated in future studies.

27Our study provides insights into PrEP uptake in several subpopulations, including adolescents and young adults, men, and serodifferent partners, using a population-level denominator. In our study communities and across sub-Saharan Africa, adolescent girls and young women have among the highest HIV risk, yet they were less likel than women aged 25 years or older to initiate PrEP. A barrier for adolescents and young adults in our study was that these individuals were more likely to be mobile than older people. In our qualitative work, adolescents and young adults who initiated PrEP were motivated by high perceived HIV risk and beliefs that PrEP use supported pursuing life goals, but some wanted “proof” of PrEP efficacy. Further mixed-methods investigation of these barriers is warranted.

28Men have largely not been prioritised for PrEP in sub-Saharan Africa, except for men in serodifferent couples and men who have sex with men. Yet, one-third of new HIV diagnoses in our study region are among men. We found substantial PrEP uptake among men, who comprised half of the individuals who initiated PrEP. In our qualitative work, we have found high levels of interest in PrEP among men, particularly among those in concurrent partnerships, and found that men typically face fewer barriers to PrEP adoption (eg, partner permission) than women. Our approach to PrEP service delivery, with out-of-facility options for PrEP initiation and follow-up, might have fostered uptake among men, for whom clinic attendance might pose a barrier to HIV testing and prevention services. Men are a crucial link in transmission networks, and expanding their prevention options could accelerate incidence reductions in generalised epidemics.

29PrEP uptake was higher among serodifferent partners than among other groups in this study, yet fewer than half of individuals in serodifferent partnerships started PrEP within 90 days of population-level HIV testing. Our study population contrasts with previous studies that have reported high levels of PrEP initiation among mutually disclosed serodifferent couples presenting to clinics before ART eligibility or initiation., We approached serodifferent partners at a population level, who might not have tested together or disclosed their HIV status. We did intensive outreach to serodifferent couples during population-level testing and at clinics, where couples’ counselling was routinely offered, but some partners were not residing in or in care within the study communities. We also found that, overall, 80% of spouses with HIV were virally suppressed and 70% were in serodifferent partnerships for at least 3 years. These factors could have reduced the perceived need for couples to initiate PrEP. Finally, we observed sex differences in PrEP uptake among individuals without HIV in serodifferent partnerships: 49% of women and 39% of men initiated PrEP. This difference could have been related to the interest of women in discreet, female-controlled HIV prevention, or to barriers to disclosure among women with HIV. As ART and viral-load monitoring are scaled up, counselling messages for serodifferent partners should be tailored on the basis of the durability of viral suppression and barriers to ART adherence for the partner who has HIV, and whether the partner without HIV infection has concurrent partners who might be viraemic or of unknown HIV status.

30Among participants in our study who initiated PrEP, engagement at follow-up visits was among the highest reported to date in Kenya and Uganda in studies not solely enrolling serodifferent couples.,,, Three-quarters of ndividuals who initiated PrEP were seen at least once during follow-up, and more than half remained engaged in the PrEP programme up to 72 weeks. Medication refill (≥89%) and self-reported adherence (≥70%) were high among participants reporting HIV risk during follow-up, but more than half of participants who initiated PrEP reported no current HIV risk at least once, and many had breaks in PrEP use. Barriers to PrEP engagement and adherence were similar to barriers to uptake, but reasons for stopping also included changes in partnerships and risk perception. Among individuals who stopped PrEP, ongoing programme engagement provided opportunities for repeat HIV testing and PrEP access, and half of the participants who had stopped PrEP later restarted the drug. As prevention options expand to other modalities, such as long-acting PrEP, follow-up visits could serve as platforms for frequent HIV testing and delivery of a variety of HIV prevention options.

31Although at least 70% of individuals reporting current HIV risk at each follow-up visit self-reported adherence to PrEP, concentrations of tenofovir in hair samples showed poor adherence (fewer than four doses per week) in one-third of those individuals. Of particular concern is that women and participants younger than 25 years had lower odds of having tenofovir concentrations consistent with daily dosing (which might be required for protection from vaginal exposure to HIV) at week 4. We did find higher adherence among young women sampled at week 24, but only slightly more than half had tenofovir concentrations consistent with taking at least four doses per week. Given that many participants had breaks in PrEP use, concentrations of tenofovir in hair could have underestimated adherence among individuals restarting PrEP in the preceding 30 days. Although data on cumulative tenofovir concentrations are limited in eastern Africa, studies among young women in southern Africa, found that half or fewer had tenofovir-diphosphate concentrations in dried blood spots consistent with at least four doses per week, and a minority of participants had concentrations consistent with daily PrEP use.

32To our knowledge, this study is the first in sub-Saharan Africa to implement a machine learning-based risk score to identify individuals for PrEP initiation. We found that real-time entry of demographic data and score assessment at the point-of-contact during community-based HIV testing was feasible. In generalised epidemic settings, machine learning-based scores have shown the potential to improve the efficiency of identifying individuals for enhanced prevention compared with model-based scores or targeting traditional risk groups. Further research is needed on optimal approaches for implementing risk scores (and results counselling), and comparing risk scores to risk self-assessment and guidelines-based eligibility. Importantly, our risk score was not used to exclude individuals from PrEP eligibility but to foster conversations about PrEP. Individuals not identified by risk score who self-identified as being at risk (regardless of serodiscordance or risk score) could access PrEP, in line with WHO guidance to offer PrEP to those requesting it.

33A major strength of this study is the use of population-based HIV testing and risk assessment, which enabled us to establish a population-level denominator for PrEP cascade analyses. Our approach achieved high testing coverage, allowing individuals to rapidly learn their status and seek ART or PrEP. As ART coverage and viral suppression increase, achieving epidemic control will require intensified HIV prevention for individuals who remain at risk. Questions remain, however, about the impact and cost-effectiveness of population-level PrEP roll-out, and the levels of uptake and adherence needed to reduce HIV incidence.

34This study has several limitations. During the population-level testing window, 17% of community members did not have HIV testing or a risk assessment. The risk score did not include questions about sexual behaviour, sex work, or same-sex partnerships, although we also allowed for risk self-identification. Future research should evaluate whether sexual behaviour data improve risk score performance (and optimal methods to facilitate disclosure and discussions on sexual health). Additionally, written consent was required to initiate PrEP, which might limit generalisability to settings now offering PrEP within routine care.

35Rapid introduction of universal PrEP access during population-level HIV testing resulted in nearly 3500 PrEP starts within 90 days in rural Kenya and Uganda. We found that a substantial proportion of the population want PrEP and can take it, including many who would not have otherwise had access to PrEP without inclusive eligibility. As PrEP roll-out accelerates, our study provides insights into gaps in the PrEP cascade that can be addressed to optimise the impact of this powerful prevention tool. Combination approaches to HIV prevention (including oral PrEP and, ultimately, on-demand and long-acting methods) that are inclusive and offer flexible delivery, hold promise to ultimately reduce HIV incidence in generalised epidemic settings.