Articles

A multilevel health system intervention for virological

suppression in adolescents and young adults living with HIV
in rural Kenya and Uganda (SEARCH-Youth): a cluster
randomised trial
Theodore Ruel*, Florence Mwangwa*, Laura B Balzer, James Ayieko, Marilyn Nyabuti, Wafula Erick Mugoma, Jane Kabami, Brian Kamugisha,
Douglas Black, Bridget Nzarubara, Fred Opel, John Schrom, George Agengo, Janet Nakigudde, Hellen N Atuhaire, Josh Schwab, James Peng,
Carol Camlin, Starley B Shade, Elizabeth Bukusi, Bill G Kapogiannis, Edwin Charlebois, Moses R Kamya, Diane Havlir

Summary
Lancet HIV 2023; 10: e518–27 Background Social and cognitive developmental events can disrupt care and medication adherence among adolescents
See Comments page e489 and young adults living with HIV in sub-Saharan Africa. We hypothesised that a dynamic multilevel health system
*Joint first authors intervention helping adolescents and young adults and their providers navigate life-stage related events would
Department of Pediatrics increase virological suppression compared with standard care.
(Prof T Ruel MD), Division of
HIV, Infectious Diseases and Methods We did a cluster randomised, open-label trial of young individuals aged 15–24 years with HIV and receiving
Global Medicine, Department
of Medicine (D Black BA,
care in eligible clinics (operated by the government and with ≥25 young people receiving care) in rural Kenya and
J Schrom MPH, Prof D Havlir MD), Uganda. After clinic randomisation stratified by region, patient population, and previous participation in the SEARCH
Department of Epidemiology trial, participants in intervention clinics received life-stage-based assessment at routine visits, flexible clinic access,
and Biostatistics and rapid viral load feedback. Providers had a secure mobile platform for interprovider consultation. The control
(Prof E Charlebois PhD,
S B Shade PhD), and
clinics followed standard practice. The primary, prespecified endpoint was virological suppression (HIV RNA
Department of Obstetrics, <400 copies per mL) at 2 years of follow-up among participants who enrolled before Dec 1, 2019, and received care at
Gynecology & Reproductive the study clinics. This trial is registered with ClinicalTrials.gov, NCT03848728, and is closed to recruitment.
Sciences (Prof C Camlin PhD),
University of California,
San Francisco, San Francisco,
Findings 28 clinics were enrolled and randomly assigned (14 control, 14 intervention) in January, 2019. Between
CA, USA; Infectious Diseases March 14, 2019, and Nov 26, 2020, we recruited 1988 participants at the clinics, of whom 1549 were included in the
Research Collaboration, analysis (785 at intervention clinics and 764 at control clinics). The median participant age was 21 years (IQR 19–23)
Kampala, Uganda
and 1248 (80·6%) of 1549 participants were female. The mean proportion of participants with virological suppression
(F Mwangwa MSc,
B Nzarubara MSc, at 2 years was 88% (95% CI 85–92) for participants in intervention clinics and 80% (77–84) for participants in control
B Kamugisha MSc, clinics, equivalent to a 10% beneficial effect of the intervention (risk ratio [RR] 1·10, 95% CI 1·03–1·16; p=0·0019).
J Kabami MPH, The intervention resulted in increased virological suppression within all subgroups of sex, age, and care status at
H N Atuhaire BMLS); Division of
baseline, with greatest improvement among those re-engaging in care (RR 1·60, 95% CI 1·00–2·55; p=0·025).
Biostatistics, School of Public
Health, University of California,
Berkeley, CA, USA Interpretation Routine and systematic life-stage-based assessment, prompt adherence support with rapid viral load
(L B Balzer PhD, J Schwab MSc); testing, and patient-centred, flexible clinic access could help bring adolescents and young adults living with HIV
Kenya Medical Research
closer towards a goal of universal virological suppression.
Institute, Kisumu, Kenya
(J Ayieko PhD, M Nyabuti MBchB,
W E Mugoma MSc, F Opel BS, Funding Eunice Kennedy Shriver National Institute of Child Health and Human Development, US National Institutes
G Agengo BSc, of Health.
Prof E Bukusi PhD); Department
of Psychiatry (J Nakigudde PhD)
and Department of Medicine Copyright 2023 The Author(s). Published by Elsevier Ltd. This is an Open Access article under the CC BY 4.0 license.
(Prof M R Kamya PhD), Makerere
University College of Health Introduction behaviours of adolescents and young adults living with
Sciences, Kampala, Uganda;
Department of Biostatistics,
Adolescents and young adults bear a large burden of the HIV. We developed the SEARCH-Youth intervention
University of Washington, HIV epidemic. In 2020, 1·75 million adolescents (aged using the empirically validated PRECEDE model to
Seattle, WA, USA (J Peng MS); 10–19 years) were estimated to be living with HIV globally.1 identify and accommodate the dynamic challenges faced
Eunice Kennedy Shriver As they age, adolescents and young adults with HIV in by adolescents and young adults living with HIV.5 The
National Institute of Child
Health and Human
sub-Saharan Africa face highly dynamic pressures that PRECEDE model is based on the idea that health
Development, Bethesda, MD, lead to disruptions in care and poor adherence to promotion strategies are most effective when they are
USA (B G Kapogiannis MD) treatment, resulting in lower rates of suppression created with the people affected and when they address
compared with older adults.2 As described in the predisposing factors, including knowledge, attitudes, or
conceptual framework of Sawyer and colleagues3 and as beliefs that affect behaviour; enabling factors, which
recognised by UNAIDS,4 cognitive and social role changes facilitate change by making the behaviour easier;
interact with social determinants to impact the health and and reinforcing factors, which include anticipated

e518 www.thelancet.com/hiv Vol 10 August 2023

 Articles

Correspondence to:
Research in context Prof Theodore Ruel, Department
of Pediatrics, University of
Evidence before this study isolated rural providers. In a pragmatic cluster randomised trial, California, San Francisco,
Adolescents and young adults living with HIV have lower rates the SEARCH-Youth intervention resulted in higher rates of CA 94158-0434, USA
of virological suppression than older adults, and few virological suppression and retention in care at 2 years of [email protected]

interventions have shown evidence of improving outcomes. follow-up. During the study period, the study regions
We searched PubMed for articles published in English from transitioned young individuals to dolutegravir-based treatment
database inception to Sept 28, 2022, using the terms: and were subject to mandated closures from the COVID-19
“(HIV) AND ((adolescents) OR (youth) OR (young adults)) AND pandemic. Our results represent rigorous evidence in support of
((virologic) or (viral)) and (suppression) and (Africa)”. We a new, multilevel service delivery model, including
identified 834 articles. Before this study, a systematic review of interventions at patient, provider, and clinic levels, which
articles published from 2015 to 2019 about interventions to improves the virological outcomes of adolescents and young
improve antiretroviral therapy adherence among adolescents adults living with HIV, generalisable to the current treatment
and young people in low-income and middle-income countries context in rural sub-Saharan Africa.
identified only three patient-level interventions and four health
Implications of all the available evidence
services interventions, none of which increased rates of
Our study adds to a growing evidence base that we can increase
virological suppression.
virological suppression with multicomponent interventions in
Added value of this study young people living with HIV. Recent trials have shown that
The SEARCH-Youth intervention was designed to provide combinations of peer mentoring with facilitated transitions to
dynamic support to adolescents and young adults with HIV in adult care, and peer-led differentiated service delivery with
rural sub-Saharan Africa as they developed into adults, facing adherence support can result in increased virological
potentially disruptive events. The intervention combined a life- suppression. Interventions such as SEARCH-Youth, applying
stage-based assessment to help providers and patients identify life-event informed multilevel dynamic support to adolescents
potential challenges to retention and adherence, with and young adults with HIV and providers, should be used to
alternative options for clinic access to address barriers, rapid achieve the ultimate goal of universal virological suppression
viral load testing to provide timely feedback about adherence, across all ages.
and a WhatsApp platform to promote collaboration among

consequences after a behaviour. Engaging with young Methods

people and clinic providers, we developed the SEARCH- Study design and participants
Youth intervention, with feature components at patient, This study was a cluster randomised, unblinded,
provider, and clinic levels. First, a life-stage-based controlled trial, with clinics as the unit of randomisation.
assessment tool aims to change the nature of the clinical We selected clinics in rural regions (approximately
interaction and help providers and young people identify 9000–11 000 people) in western Kenya and southwestern
facilitators and barriers to care and adherence. Second, Uganda that were operated by the government, provided
young people are offered alternative access options to antiretroviral therapy to similar numbers of young
address barriers to coming to the clinic, including having people with HIV (approximately 25–400 people), and
visits by phone, after hours, or offsite. Third, viral load were geographically distant from other potential trial
testing is done with rapid turnaround time, so that clinics. We excluded clinics with small youth clinic
providers and young people can reinforce good adherence patient population sizes (<25 people). Female and male
or identify challenges in a timely manner. Finally, an individuals aged 15–24 years with HIV were recruited
electronic platform is intended to facilitate communication from local health centres and were eligible if they had a
(e-collaboratives) and shared problem solving among confirmed HIV diagnosis, and had received or were
providers who are often isolated in rural clinics. initiating care at a study clinic. Written informed consent
To evaluate the effect of the SEARCH-Youth was obtained before enrolment from all participants.
intervention, we used a cluster randomised clinical trial Minors aged 15–17 years independently provided
design, with clinics as the unit of randomisation. We informed consent in accordance with Ugandan and
chose this design because the implementation of the Kenyan guidelines for research related to care for
SEARCH-Youth intervention occurs at a clinic level and patients with a sexually transmitted infection. The
the risk of contamination of the control condition at the ethical and institutional review boards of Makerere
provider and participant levels is minimised. The University (Kampala, Uganda), the Uganda National
primary objective of the trial was to establish the Council for Science and Technology, the Kenya Medical
effectiveness of the SEARCH-Youth intervention in Research Institute, and the University of California
increasing the proportion of participants with virological San Francisco (San Francisco, CA, USA) approved this
suppression. study.

www.thelancet.com/hiv Vol 10 August 2023 e519

 Articles

Randomisation and masking of both countries advised HIV services to institute and
We randomised the eligible clinics within strata of total rapidly scale up alternative care delivery approaches:
youth clinic patient population size (<300 or ≥300 people), home delivery of drugs, offsite visits, phone-based
region (Kenya or Uganda), and previous participation in counselling, and 6 months’ drug refills; all of these
the SEARCH trial.6 Randomisation was done by an services were already being implemented at intervention
independent statistician using a random number sites.
generator. Clinics were not masked to the randomisation We trained clinicians in both control and intervention
group, but the study statistician (LBB) was masked until sites on the development of young individuals from age
trial completion. 15 years through adolescence and into adulthood and
their associated medical issues (eg, contraception and
Procedures alcohol use) using country-approved curricula. We
Participants at both intervention and control clinics additionally trained the clinicians at intervention sites in
received the local standard clinical care, including any the components of the SEARCH-Youth intervention (life-
youth-targeted programmes being implemented by the stage tool, clinic access choice, rapid viral load feedback,
countries’ ministries of health or other organisations and e-collaboratives) 2 weeks before the study start.
See Online for appendix (appendix p 2). Caregivers were included in care at the Data collection for study endpoints occurred through
discretion of youth participants, assisting the study several mechanisms. Patient satisfaction was assessed
teams in contacting youth participants and supporting using a Likert-type scale, ranging from strongly disagree
adherence at home. In control clinics, routine viral load to strongly agree. Care engagement, clinic transfers, and
monitoring was implemented every 6 months according treatment changes were ascertained from clinical charts.
to Ugandan and Kenyan guidelines and using ministry Use of each component of the SEARCH-Youth
of health laboratory facilities.7,8 In the current standard of intervention was measured using the tablet-based study
care, results are generally reported at the next routine form data.
visit, with adherence counselling for patients with a viral For the costing analyses, we did 2-week site visits
load higher than 1000 copies per mL and retesting during during enrolment and follow-up at both intervention and
a 3-month period. control facilities. During these visits, clinic and study
In intervention clinics, the SEARCH-Youth staff were interviewed to identify resources expended
intervention was added to standard practices and during care for young people with HIV; implementation
comprised four components (appendix p 2). The life- staff completed self-administered time-and-motion
stage assessment was done by clinicians at intervention surveys to assess the proportion of their effort required to
sites using devoted tablet-based software developed by provide HIV care and implement study activities. We
our team (appendix pp 3–4) at every routine visit. have previously used these methods to assess the costs of
Alternative clinic access was offered as needed to address streamlined HIV care for individuals aged 15 years or
perceived barriers. In intervention clinics, routine viral older in this setting.9
load testing occurred at the same frequency as in control
clinics. However, the mechanism for testing and results Outcomes
communication differed in intervention clinics. Plasma The primary outcome was the clinic-level proportion of
was transported the same or next day to one of the hubs, study participants with virological suppression (HIV
tested using the Xpert HIV-1 Viral Load assay (Cepheid, RNA <400 copies per mL) at 2 years of follow-up. We
Sunnyvale, CA, USA) with results communicated to the prespecified that the primary endpoint analysis would
clinicians by phone or electronically, who then reported include data from participants enrolled before
results to participants by phone or in person (mechanism Dec 1, 2019; the rationale for this decision was to
chosen by the participants at the previous visit) within a facilitate timely communication of study results and was
target turnaround time of less than 72 h. Adherence discussed with the data and safety monitoring board on
counselling was provided if the viral load was higher April 29, 2021. We also prespecified that people who
than 200 copies per mL, and with repeat viral testing per withdrew consent, had formally transferred care, or
provider discretion as soon as 2 weeks later. Clinicians outmigrated (defined as moving more than a 3-h
were prompted to consider e-collaboratives if they sought travelling distance from the study clinic) would be
input on challenges such as adherence and stigma. excluded from the primary analysis; the rationale for this
E-collaborative discussions were done using WhatsApp, decision was that these people no longer represented the
providing end-to-end encryption, and without using target population—ie, adolescents and young adults
patient-specific information to protect the confidentiality living with HIV and in the catchment area of the study
of participants. clinics.
The COVID-19 pandemic introduced a high strain on The primary analysis included all remaining
participants and providers at our study sites during the participants, and among those participants, we classified
study period, including school closures and travel their primary endpoint as either suppressed or not. Then
restrictions (appendix p 9). The national health ministries we calculated the clinic-specific proportion of young

e520 www.thelancet.com/hiv Vol 10 August 2023

 Articles

people with virological suppression. To assess the For costing, we entered data into a standardised Excel
robustness of these decisions, we prespecified sensitivity workbook and analysed the data to estimate the
analyses to include people who outmigrated or incremental annual cost per participant associated with
transferred care, to exclude participants with missing SEARCH-Youth activities. Time-and-motion data were
endpoint viral loads, and to adjust for differences used to estimate the proportion of active work time that
between participants with measured versus missing was dedicated to HIV care and study activities for each
endpoint viral loads. staff cadre in each facility. These data were included in
Secondary, prespecified outcomes were care the micro-costing to estimate the direct and indirect costs
engagement, switches to dolutegravir, patient of patient visits in intervention and control clinics.
satisfaction, intervention implementation, severe adverse Annual costs for each patient assume three clinic visits
events, and costing. These outcomes were assessed in per year, US$65 per patient per year for a dolutegravir-
participants who enrolled before Dec 1, 2019. Additional based antiretroviral treatment regimen, and an annual
secondary, prespecified outcomes (ie, barriers and viral load test ($31·48 for a rapid viral load test in
facilitators of the intervention, longitudinal virological intervention facilities and $110 for a traditional viral load
suppression, alcohol use, HIV-free survival among test in control facilities). This trial is registered with
infants born to participants, rates of vertical transmission, ClinicalTrials.gov, NCT03848728.
and mental health) will be published separately. Further
details on primary and secondary outcomes are available Role of the funding source
in the statistical analysis plan (appendix p 10). The funder of the study had no role in study design, data
collection, data analysis, data interpretation, or writing of
Statistical analysis the report.
Using sample size formulas for cluster randomised trials
with proportion endpoints,10 we estimated that 28 clinics Results
(14 clinics per group) would provide 80% power to detect In January, 2019, we included 28 clinics and excluded four
at least a 24% relative increase in virological suppression due to their small patient population sizes. The eligible
at 2 years of follow-up from 65% in the control, assuming clinics were randomly assigned: 14 were assigned to the
a coefficient of variation of k=0·175 and a harmonic intervention and 14 to the control, balanced on country
mean of 50 participants per clinic. (14 in Kenya and 14 in Uganda). From March 14, 2019, to
In the primary analysis, we compared the average Nov 26, 2020, we recruited 1988 participants, of whom
proportion of young people with virological suppression 1834 had enrolled before Dec 1, 2019: 916 (90·5%) of
at 2 years of follow-up with targeted minimum loss-based 1012 participants at intervention clinics and 918 (94·1%) of
estimation, an approach that accounts for the dependence
of outcomes within clusters and adaptively adjusts for
baseline covariates to maximise precision.11 Specifically, 28 clinics randomly assigned
we used leave-one-out cross-validation to select from the
following prespecified candidate adjustment variables:
the clinic-specific number of adolescents and young
14 clinics assigned to 14 clinics assigned to control
adults in HIV care at baseline, the clinic-specific intervention
proportion of young people with virological suppression
among those engaged at baseline, or no adjustment.
Using the Student’s t-distribution, we calculated two- 1031 people screened 987 people screened
sided 95% CIs and tested the null hypothesis that the
SEARCH-Youth intervention did not improve virological 19 people excluded 11 people excluded
suppression compared with the control intervention, 18 not in age range 11 not in age range
with a one-sided test at the 5% significance level. 1 declined

Prespecified subgroups included sex, age group

(15–19 years and 20–24 years), and baseline care status 1012 people enrolled 976 people enrolled
(recently engaged [started treatment within 6 months of
or at enrolment], engaged [started treatment more than
96 enrolled on, or after, 58 enrolled on, or after,
6 months before enrolment and had a clinic visit within Dec 1, 2019 Dec 1, 2019
6 months before enrolment], and re-engaging [started 1 withdrawal 106 outmigrated
treatment more than 6 months before enrolment and 98 outmigrated 48 transferred
32 transferred
without a clinic visit within 6 months of enrolment]).
Prespecified sensitivity and secondary analyses, including
the impact of transition to dolutegravir-based regimens, 785 people included in analysis 764 people included in analysis
are described in the statistical analysis plan (appendix
p 10).12 All analyses were done in R, version 4.0.3. Figure 1: Trial profile

www.thelancet.com/hiv Vol 10 August 2023 e521

 Articles

976 participants at control clinics (figure 1), who were

Intervention (n=785) Control (n=764) Total (n=1549)
followed up until March 1, 2022. During the 2-year follow-
Age, years 21 (19–23) 22 (19–23) 21 (19–23) up, one participant at an intervention clinic withdrew
Sex consent. Between groups, similar proportions of
Female 643 (81·9%) 605 (79·2%) 1248 (80·6%) participants outmigrated (98 [10·7%] of 915 participants in
Male 142 (18·1%) 159 (20·8%) 301 (19·4%) intervention clinics vs 106 [11·5%] of 918 participants in
Country of residence control clinics) or transferred care (32 [3·5%] vs 48 [5·2%]).
Kenya 332 (42·3%) 324 (42·4%) 656 (42·3%) All clusters were included in the analysis. The remaining
Uganda 453 (57·7%) 440 (57·6%) 893 (57·7%) 1549 participants (785 [50·7%] participants in intervention
Education (started or completed) clinics and 764 [49·3%] participants in control clinics)
No school 31 (3·9%) 24 (3·1%) 55 (3·6%) comprised the prespecified analytical cohort for the
Primary school 507 (64·6%) 525 (68·7%) 1032 (66·6%) primary endpoint.
Secondary school 194 (24·7%) 175 (22·9%) 369 (23·8%) Participants’ median age was 21 years (IQR 19–23) and
Tertiary school 53 (6·8%) 40 (5·2%) 93 (6·0%) 1248 (80·6%) of 1549 participants were female (table 1).
At boarding school 56 (7·1%) 56 (7·3%) 112 (7·2%) Most participants were already engaged in care at the
Employment status time of enrolment (1026 [66·3%] of 1547), with similar
Employed 283 (36·1%) 308 (40·3%) 591 (38·2%) distribution of care status within groups. More female
In school 165 (21·0%) 155 (20·3%) 320 (20·7%) participants (47 [3·8%] of 1246) were re-engaging in care
Unemployed 337 (42·9%) 301 (39·4%) 638 (41·2%) at baseline than male participants (eight [2·7%] of 301).
Marital status The most common antiretroviral regimen at enrolment
Single, never married 342 (43·6%) 291 (38·1%) 633 (40·9%) was tenofovir, lamivudine, and efavirenz (1118 [72·9%] of
Married, monogamous 306 (39·0%) 327 (42·8%) 633 (40·9%) 1533). The overall proportion of participants with
Married, polygamous 44 (5·6%) 51 (6·7%) 95 (6·1%) virological suppression at baseline was 74·8% (1147 of
Widowed 3 (0·4%) 6 (0·8%) 9 (0·6%) 1533): 73·5% (573 of 780) in intervention clinics and
Divorced 90 (11·5%) 89 (11·6%) 179 (11·6%) 76·2% (574 of 753) in control clinics. Baseline
Number of children characteristics of the entire cohort (appendix pp 5–6)
0 330 (42·0%) 323 (42·3%) 653 (42·2%)
were similar to that of the analytical cohort (table 1).
1 258 (32·9%) 227 (29·7%) 485 (31·3%)
Baseline characteristics at the clinic level are provided in
2 129 (16·4%) 131 (17·1%) 260 (16·8%)
the appendix (p 6).
Endpoint viral loads were obtained in 1425 (92·0%) of
3–5 51 (6·5%) 64 (8·4%) 115 (7·4%)
1549 participants: 731 (93·1%) of 785 in intervention
Drinks alcohol 137 (17·5%) 94 (12·3%) 231 (14·9%)
clinics and 694 (90·8%) of 764 in control clinics in the
Mobile* 210 (26·8%) 229 (30·0%) 439 (28·3%)
analytical cohort. 17 (1·1%) deaths occurred: eight (1·0%)
Antiretroviral regimen at enrolment†
in intervention clinics and nine (1·2%) in control clinics.
TDF–3TC–EFV 591 (76·0%) 527 (69·8%) 1118 (72·9%)
The remaining 107 (6·9%) participants did not have their
TDF–3TC–DTG 81 (10·4%) 101 (13·4%) 182 (11·9%)
endpoint viral load measured: 46 (5·9%) in intervention
AZT–3TC–NVP 33 (4·2%) 42 (5·6%) 75 (4·9%)
clinics and 61 (8·0%) in control clinics. No severe adverse
TDF–3TC–ATV/r 10 (1·3%) 18 (2·4%) 28 (1·8%)
events thought to be associated with the intervention
AZT–3TC–ATV/r 12 (1·5%) 14 (1·9%) 26 (1·7%)
were reported.
ABC–3TC–EFV 12 (1·5%) 11 (1·5%) 23 (1·5%)
In the primary analysis, the mean proportion of
Other 39 (5·0%) 42 (5·6%) 81 (5·3%)
participants with virological suppression at 2 years of
Baseline viral load <400 copies per mL‡ 573 (73·5%) 574 (76·2%) 1147 (74·8%) follow-up was 88% (95% CI 85–92) in intervention clinics
Baseline care status§ and 80% (77–84) in control clinics, corresponding to a
Recently engaged¶ 252 (32·1%) 214 (28·0%) 466 (30·1%) 10% beneficial effect from the SEARCH-Youth
Engaged|| 496 (63·3%) 530 (69·5%) 1026 (66·3%) intervention (risk ratio [RR] 1·10, 95% CI 1·03–1·16;
Re-engaging** 36 (4·6%) 19 (2·5%) 55 (3·6%) p=0·0019; appendix p 6). Similar results were observed
Data are n (%) or median (IQR). Data are restricted to the analytical population of participants who enrolled before
in the prespecified sensitivity analyses including
Dec 1, 2019, and who did not withdraw from the study, transfer, or outmigrate. 3TC=lamivudine. ABC=abacavir. participants who outmigrated or transferred care (1·12,
ATV/r=ritonavir-boosted atazanavir. AZT=zidovudine. DTG=dolutegravir. EFV=efavirenz. NVP=nevirapine. 1·04–1·20; p=0·0016), excluding participants with
TDF=tenofovir disoproxil fumarate. *Lived away from home for more than 1 month in the previous 6 months. †Data
missing for 16 (1·0%) participants: seven (0·9%) in the intervention group and nine (1·2%) in the control group.
missing endpoints (1·06, 1·01–1·12; p=0·0076), and
‡Data missing for 16 (1·0%) participants: five (0·6%) in the intervention group and 11 (1·4%) in the control group. adjusting for differences in characteristics between
§Data missing for two (0·1%) participants: one (0·1%) in the intervention group and one (0·1%) in the control group. people with measured versus missing endpoints (1·08,
¶Started treatment within 6 months of enrolment or at enrolment. ||Started treatment more than 6 months before 1·02–1·13; p=0·0036; appendix p 6).
enrolment and had an HIV care visit within 6 months of enrolment. **Started treatment more than 6 months before
enrolment, but did not have an HIV care visit within 6 months of enrolment. Across the prespecified subgroups, the intervention
was beneficial but varied in effect size. Among strata of
Table 1: Baseline characteristics of study participants in the analytical cohort HIV care at baseline, the greatest benefit was among
participants re-engaging in care (RR 1·60, 95% CI

e522 www.thelancet.com/hiv Vol 10 August 2023

 Articles

1·00–2·55; p=0·025; figure 2). The SEARCH-Youth countries). The annual cost of HIV care per patient was
intervention increased virological suppression in both US$7·71 more expensive in intervention facilities than in
female participants (1·06, 1·00–1·13; p=0·026) and male control facilities ($269·68 and $261·97, respectively;
participants (1·11, 0·98–1·25; p=0·044), as well as in appendix p 8). The intervention was associated with
older (aged 20–24 years) participants (1·05, 0·99–1·11; higher costs per patient annually in recurrent goods
p=0·040) and younger (aged 15–19 years) participants
(1·13, 1·01–1·26; p=0·0015; appendix p 7). Control
During the 2-year study period, 1003 (64·8%) of Intervention

1549 participants switched to an antiretroviral RR 1·07

RR 1·10 (95% CI
combination regimen that included dolutegravir, with a 100 (95% CI RR 1·60
1·00–1·13); RR 1·09 (95% CI
1·03–1·16);
higher proportion of participants in the intervention p=0·0019
p=0·020 (95% CI 1·00–2·55);
group (77%, 95% CI 73–82) than in the control group 0·96–1·24); p=0·025
p=0·085
(71%, 66–75; RR 1·10, 95% CI 1·00–1·20; p=0·041;
Participants with viral suppression (%)
figure 3). The intervention had a beneficial effect on 75
virological suppression in both participants who had
switched to a dolutegravir-based treatment and those who
did not, although it did not reach statistical significance
when stratified (table 2). The joint probability of switching 50
to a dolutegravir-based treatment and attaining virological
suppression by 2 years was also 11% higher in participants
in the intervention group (70%, 95% CI 65–75) than in
those in the control group (63%, 59–67; RR 1·11, 95% CI 25

1·00–1·22; p=0·020).
The SEARCH-Youth intervention also increased
80% 88% 85% 91% 73% 80% 53% 85%
retention in HIV care. Specifically, the proportion of
young people who had contact with the clinic in the 0
All Engaged Recently engaged Re-engaging
previous 6 months to endpoint ascertainment was 91%
(95% CI 89–94) among participants in intervention Figure 2: Proportion of participants with virological suppression at 2 years of follow-up, stratified by baseline
clinics and 72% (61–82) among participants in control care status
RR=risk ratio.
clinics, corresponding to a 27% increase (RR 1·27,
95% CI 1·10–1·47; p=0·0010).
100 + Control
Patient satisfaction surveys were completed by + Intervention
716 (91·2%) of 785 participants in intervention clinics
and 713 (93·3%) of 764 participants in control clinics. 80
++++++
+++++++++++++
+++++++ +
The intervention was associated with increased trust that ++ +++++++++++++++++
Cumulative probability (%)

+++++ +
+
+++
the provider would “keep my information private”, a 60
perception of ease in the ability to “get in touch with my
provider”, and a sense that providers knew “how to treat +
young people with HIV” (table 3). 40

In the intervention group, there was a high use of the

life-stage assessment. Specifically, 663 (84·5%) of 20
+

785 participants had at least four assessments. Offsite +

+

appointments, phone visits, out-of-hours appointments, +

+ ++
0 + +
and offsite drug delivery were selected by many 0 3 6 9 12 15 18 21 24 27
participants, and access choice options varied by clinic Time since enrolment (months)
(appendix p 7). Across the 14 intervention clinics, the
median proportion of viral load results delivered within Figure 3: Cumulative probability of switching to a dolutegravir-based
treatment
72 h was 95%, and the median delivery time was 1·35 days
(appendix p 8). Providers used the e-collaborative
Intervention Control Effect of intervention p value
platform for 262 discussions about 127 study participants
(95% CI) (95% CI) (95% CI)*
with challenging management issues.
Switched 92% (89–95) 88% (84–92) 1·04 (0·99–1·10) 0·057
To estimate the cost of HIV care and intervention
activities, we did 2-week site visits in 14 facilities during Did not switch 70% (61–79) 64% (54–74) 1·09 (0·89–1·34) 0·19

enrolment (seven intervention and seven control *Relative scale using two-stage targeted minimum loss-based estimation.
facilities, evenly distributed between Kenya and Uganda)
Table 2: Proportion of individuals with virological suppression among participants who switched and
and 23 facilities during follow-up (15 intervention and
did not switch to dolutegravir-based treatment
eight control facilities, evenly distributed between

www.thelancet.com/hiv Vol 10 August 2023 e523

 Articles

peer-mentor programme in clinics resulted in increased

Effect of intervention p value
(95% CI)* virological suppression in young individuals aged
15–24 years.18 There is also increasing evidence showing
“The time and date of my appointments are convenient for me.” 0·34 (0·07 to 0·60) 0·0069
benefit from combination service delivery approaches, as
“It is easy to get in touch with my provider.” 0·36 (0·11 to 0·61) 0·0029
recommended by UNAIDS.19 The Zvandiri programme
“Phone visits are (or would be) convenient for me.” –0·04 (–0·32 to 0·24) 0·61
combined community adolescent treatment supporters
“I like drug delivery outside of the clinic at a location convenient for me.” 0·46 (0·04 to 0·89) 0·017
and monthly support groups with text messaging, calls,
“I trust my provider will keep my information private.” 0·30 (0·1 to 0·49) 0·0020
home visits, and clinic-based counselling; in a cluster
“My provider knows how to treat young people with HIV.” 0·31 (0·1 to 0·53) 0·0028
randomised trial, fewer participants in the Zvandiri
“The staff at this clinic care about me.” 0·21 (–0·06 to 0·48) 0·064
intervention group (52 [25%] of 209) had non-virological
“I would recommend this clinic to other young people living with HIV.” 0·25 (0·02 to 0·47) 0·018
suppression (>1000 copies per mL) or died than
“Overall, I feel satisfied with the care I receive at this clinic.” 0·30 (0·05 to 0·54) 0·011 participants in the standard HIV care group (97 [36%] of
*Difference scale using two-stage targeted minimum loss-based estimation. 270; p=0·03).20
The SEARCH-Youth intervention is distinct from other
Table 3: Differences in patient satisfaction in the intervention and control groups
care delivery models by combining elements at the
levels of client–provider relationships (life-stage-based
($43·05 higher), personnel costs ($24·97 higher), facility discussion) and facility (offering out-of-hours, offsite,
costs ($5·17 higher), and capital goods (the equipment and phone access appointments) with a biomedical tool
needed for Xpert viral load testing; $0·31 higher). These (rapid viral load feedback) and a communication platform
additional costs were partly offset by lower services costs to facilitate collaboration among providers at different
($65·79 lower), primarily due to lower costs associated rural facilities. The intervention was designed to add
with rapid viral load testing compared with the standard resilience against pressures on adherence and care, and
practice of sending blood plasma samples to central to be dynamic in identifying and adapting to new issues
laboratories for testing. over time. The COVID-19 pandemic introduced high
external strain on participants in our study in many ways,
Discussion including restrictions on travel implemented by the
88% of adolescents and young adults living with HIV Governments of Kenya and Uganda (appendix p 9).
and participating in the SEARCH-Youth intervention One adult clinical programme in Uganda reported a
had virological suppression at 2 years, representing an 46% decrease in clinic visits overall in 2020 compared
improvement compared with standard care (80%). with the previous year, with 42% of participants (n=14 632)
Although there has been great progress in HIV treatment reporting difficulty in travelling to the clinic.21
coverage globally, improvements in clinical outcomes for Additionally, there has been particular concern about the
adolescents and young adults have lagged behind those impact of the COVID-19 pandemic on adolescents living
for older adults.2 The gains in care engagement and with HIV globally. The SEARCH-Youth intervention
virological suppression derived from the SEARCH-Youth showed benefit even within the constraints of the
intervention brought adolescents and young adults with COVID-19 pandemic and when many of the access
HIV in this cohort closer to the UNAIDS 2030 95-95-95 components of the intervention (eg, phone-based
targets. counselling and 6 months drug refills) were being
The development of interventions to improve outcomes recommended by the ministries of health across all sites.
among adolescents and young adults with HIV in sub- It is possible that due to our pre-existing phone-based
Saharan African has been slow, but some approaches are activities, intervention clinicians could more easily reach
now showing success. A 2019 review of interventions to patients and help them arrange alternate drug pick-up
improve antiretroviral therapy adherence among away from their usual providers while counselling and
adolescents and young people identified reports of three life-stage assessment were done by phone.
patient-level interventions and four health services The life-stage tool was intended to create a collaborative
interventions, but none were shown to increase relationship between providers and clients, helping them
virological suppression.13 Implementation of the Kenyan identify and anticipate barriers to adherence and care
Adolescent Package of Care did not increase virological that can shift through this period of life. At the start of
suppression among adolescents in one study;14 however, this trial, we examined the prevalence of recent life-stage
another retrospective study found that the presence of events in the intervention group using data from the first
youth-friendly services plus trained providers was life-stage assessment and found that two or more major
associated with increased rates of virological suppression life-stage events (eg, changes in schooling, employment,
at clinics.15 Peer-based strategies are now recommended or partnerships, or becoming a parent) were associated
by WHO and show promising results.16 The Baylor with no virological suppression.22 The tool was also
College of Medicine International Pediatric AIDS designed to help providers elicit and address issues of
Initiative Teen Clubs increased retention in a case-control stigma with individualised solutions such as alternative
study of adolescents in Malawi.17 In Project YES!, a youth access options (eg, coming to the clinic at early or late

e524 www.thelancet.com/hiv Vol 10 August 2023

 Articles

hours to accommodate school or work schedules). (eg, alternative access options varying by participant), did
Stigma at individual, interpersonal, community, and not disrupt participant access to other local programmes,
organisational levels intersects with and continues to and generally aligned with standard country follow-up
produce substantial barriers to treatment among young schedules. The timing of this trial, during the roll-out of
people living with HIV in Africa. Qualitative study of our dolutegravir-based regimens in these regions, also
participant population at baseline suggested that non- increases its relevance to the current context. During the
disclosure of HIV status, medication schedules, and 2-year study period, both Kenya and Uganda transitioned
clinic appointments all elicited stigma at school.23 The to tenofovir–lamivudine–dolutegravir for people living
results of our patient satisfaction survey suggest that the with HIV, as per WHO guidance.31 Evidence from clinical
intervention succeeded in increasing trust in providers to programmes of adults and clinical trials in adolescents
keep information private and that convenience in suggest that dolutegravir-based regimens result in
alternative visit times was appreciated. improved rates of virological suppression.32 We found
Providers in intervention clinics were very successful that the SEARCH-Youth intervention facilitated the
in implementing rapid viral load feedback, with a median transition of young people to tenofovir–lamivudine–
of 95% of viral load results delivered to clients within dolutegravir, with a higher proportion of intervention
72 h (appendix p 8). Point-of-care viral load assays with participants having switched by 2 years (figure 3).
rapid turnaround times are now being implemented Transition to dolutegravir was probably a key driver to the
more widely in low-income and middle-income high rates of virological suppression in the control group
countries, but outcomes have been mixed. In a (80%). Importantly, we also found that the intervention
randomised controlled trial, point-of-care viral load appeared to confer benefit not just to participants who
testing every 3 months for Kenyan children younger than switched to dolutegravir-based treatment, but also among
14 years was not associated with increased prevalence of those who had not switched (table 2).
virological suppression.24 However, point-of-care viral Increasing evidence including from studies done in
load testing was associated with increased virological Uganda shows that rapid viral load feedback is feasible to
suppression in a study of South African adults.25 We implement at scale. A cluster randomised trial in
postulated that the rapid feedback of viral loads—in both 20 clinics in Uganda showed that rapid viral testing using
supporting successful adherence and identifying the GeneXpert was feasible, with median turnaround
lapses—had the potential to be especially effective for time of 1 day.33 The Ugandan Ministry of Health is
adolescents and young people who are just developing currently prioritising improvements in turnaround time
the cognitive skills of abstract thinking and appreciating for viral load results and moving to communicate results
long-term consequences of actions.3 However, we could to clinics electronically; clinics in Uganda are also
not discern the specific contribution of the rapid viral implementing same-day point-of-care viral load testing
load feedback within the combination SEARCH-Youth with the GeneXpert assay in their follow-up programme
model of care. for infants born to women with HIV. An additional
A key component of the success of the SEARCH-Youth consideration in implementing this intervention on a
intervention was the higher rate of retention in care in large scale would be its cost. We estimated that the
intervention clinics (91%) compared with control clinics SEARCH-Youth intervention was associated with a
(72%). One of the main drivers of poor outcomes in modest increase of $7·71 (3%) in annual cost per patient.
young people living with HIV in sub-Saharan Africa is If point-of-care viral load testing became standard of care,
disengagement from care, especially among adolescents. then our intervention would be associated with $30·35
A systematic review estimated that more than 15% of (12%) in additional costs per patient per year. These costs
adolescents are lost to follow-up across sub-Saharan are lower than what has been observed in several studies
Africa,26 but rates as high as 52% in South Africa27 and of non-clinical interventions for adolescents and young
32% in rural Uganda28 have been reported. We postulate adults with HIV, ranging from $49·50 to $166·02 in
that a perception of satisfaction with care is important in additional costs per patient per year.34,35
maintaining clients in care. Overall patient satisfaction Our study had limitations. As is characteristic for this
has been associated with re-engagement with HIV care age group, our participants had a high degree of
in adults.29 We found that participants in the intervention mobility, resulting in transfers of care and outmigration.
group were much more satisfied with care than To address this shortcoming, we tracked all participants
participants in the control group (table 3). who had outmigrated or transferred care and included
Several features of the design and timing of this trial them in a secondary analysis that also showed benefit
contribute to the generalisability of our results to the of the intervention among those who had separated
current treatment landscape for young people and their from the study community. Another limitation of our
utility to policy makers. We used a pragmatic30 design study is the challenge of identifying which elements of
that included minimal enrolment criteria and clinic- our multicomponent intervention had the greatest
based recruitment to better approximate real-world benefit. Ongoing qualitative studies of providers and
implementation. We built flexibility into the intervention participants might provide insight into this aspect. Our

www.thelancet.com/hiv Vol 10 August 2023 e525

 Articles

costing estimates were limited to the ongoing costs of 3 Sawyer SM, Afifi RA, Bearinger LH, et al. Adolescence:
a foundation for future health. Lancet 2012; 379: 1630–40.
HIV care for adolescents and young adults. Programmes
4 UNAIDS. Get on the fast-track: the life-cycle approach to HIV.
implementing this intervention would incur additional Finding solutions for everyone at every stage of life. 2016. http://
costs associated with training of personnel. The www.unaids.org/sites/default/files/media_asset/Get-on-the-Fast-
implication of these costs on the additional cost per Track_en.pdf (accessed June 13, 2022).
5 Green LW, Kreuter MW, Deeds SG, Partrige KB. Health education
person would depend on the number of adolescents planning: a diagnostic approach. Palo Alto, CA: Mayfield, 1980.
and young adults served by this intervention. 6 Havlir DV, Balzer LB, Charlebois ED, et al. HIV testing and
As they transition into adulthood, adolescents and treatment with the use of a community health approach in rural
Africa. N Engl J Med 2019; 381: 219–29.
young adults living with HIV in sub-Saharan Africa face
7 Ugandan Ministry of Health. Consolidated guidelines for the
numerous challenges to remaining in care and having prevention and treatment of HIV in Uganda, 2nd edn.
virological suppression. Dynamic models of care that September, 2018. https://platform.who.int/docs/default-source/
mca-documents/policy-documents/guideline/UGA-RH-43-01-
accommodate their individual needs and increase GUIDELINE-2018-eng-Final-Uganda-HIV-Guidelines.pdf (accessed
resilience in their relationships with providers and June 13, 2023).
clinics are needed. The multilevel SEARCH-Youth 8 Kenyan Ministry of Health. Guidelines on use of antiretroviral
intervention improved outcomes among young people drugs for treating and preventing HIV in Kenya. 2018. https://
chskenya.org/wp-content/uploads/2018/08/Kenya-ARV-
when added to standard practices, providing benefits Guidelines-2018.pdf (accessed July 10, 2023).
beyond transition to dolutegravir, and was effective 9 Shade SB, Osmand T, Luo A, et al. Costs of streamlined HIV care
despite COVID-19 pandemic-related restrictions. delivery in rural Ugandan and Kenyan clinics in the SEARCH
study. AIDS 2018; 32: 2179–88.
Systematic life-stage-based assessment and similar 10 Hayes RJ, Bennett S. Simple sample size calculation for cluster-
multilevel flexible support could help bring adolescents randomized trials. Int J Epidemiol 1999; 28: 319–26.
and young adults living with HIV closer towards a goal of 11 Balzer LB, van der Laan M, Ayieko J, et al. Two-stage TMLE to
reduce bias and improve efficiency in cluster randomized trials.
universal virological suppression. Biostatistics 2023; 24: 502–17.
Contributors 12 Balzer LB, Ruel T, Havlir DV. Statistical analysis plan for primary
FM, EC, JA, WEM, FO, DB, JP, JK, MRK, CC, SBS, DH, and TR and selected secondary health endpoints of the SEARCH-Youth
conceptualised and designed the study. All authors participated in study study. arXiv 2022; published online Nov 4. https://doi.org/10.48550/
operations, data collection, and design modifications. LBB, JA, MN, arXiv.2211.02771 (preprint).
WEM, JK, BK, and SBS did the statistical analysis, with input from 13 Reif LK, Abrams EJ, Arpadi S, et al. Interventions to improve
JP, JSchr, JSchw, TR, FM, and DH. TR wrote the first draft of the antiretroviral therapy adherence among adolescents and youth in
manuscript, with input from FM, LBB, and DH. LBB, WEM, BK, FM, low- and middle-income countries: a systematic review 2015–2019.
AIDS Behav 2020; 24: 2797–810.
and TR directly accessed and verified the study data. All authors
reviewed the initial draft and approved the final manuscript. All authors 14 Mburu M, Guzé MA, Ong’wen P, et al. Evaluating the effectiveness
of the HIV adolescent package of care (APOC) training on viral load
had full access to all the data in the study and had final responsibility for
suppression in Kenya. Public Health 2019; 173: 146–49.
the decision to submit for publication.
15 Wilson K, Onyango A, Mugo C, et al. Kenyan HIV clinics with
Declaration of interests youth-friendly services and trained providers have a higher
DH has received research grants from the National Institutes of Health, prevalence of viral suppression among adolescents and young
and non-financial support from Gilead and AbbVie, all via her adults: results from an observational study.
institution. EC has received payments for consultation on unrelated J Assoc Nurses AIDS Care 2022; 33: 45–53.
studies of HIV and hypertension from the Infectious Diseases Research 16 WHO. Adolescent-friendly health services for adolescents living
Collaboration in Uganda. All other authors declare no competing with HIV: from theory to practice. Geneva: World Health
interests. Organization, 2019.
17 MacKenzie RK, van Lettow M, Gondwe C, et al. Greater retention in
Data sharing care among adolescents on antiretroviral treatment accessing “Teen
De-identified study data will be made available approximately 1 year after Club” an adolescent-centred differentiated care model compared
completion of the ongoing trial (NCT03848728), following approval of a with standard of care: a nested case-control study at a tertiary
concept sheet summarising the analyses to be done. Further inquiries referral hospital in Malawi. J Int AIDS Soc 2017; 20: e25028.
can be directed to the SEARCH Scientific Committee at 18 Denison JA, Burke VM, Miti S, et al. Project YES! Youth engaging
[email protected]. for success: a randomized controlled trial assessing the impact of a
clinic-based peer mentoring program on viral suppression,
Acknowledgments adherence and internalized stigma among HIV-positive youth
We are grateful to the local research teams, clinic staff, and study (15–24 years) in Ndola, Zambia. PLoS One 2020; 15: e0230703.
participants. This study was supported by a grant from 19 UNAIDS. Improving HIV service delivery for infants, children, and
Eunice Kennedy Shriver National Institute of Child Health and Human adolescents: a framework for country programming. 2020. https://
Development at the National Institutes of Health (UH3HD096915, with library.unaids.org/?publication=improving-hiv-service-delivery-for-
grants awarded to all authors, except BK, to their institutions). This infants-children-and-adolescents-a-framework-for-country-
project was part of the ​Prevention and Treatment through a programming (accessed June 13, 2023).
Comprehensive Care Continuum for HIV-affected Adolescents in 20 Mavhu W, Willis N, Mufuka J, et al. Effect of a differentiated service
Resource Constrained Settings (PATC³H) consortium delivery model on virological failure in adolescents with HIV in
(https://www.patc3h.org/). The content is solely the responsibility of the Zimbabwe (Zvandiri): a cluster-randomised controlled trial.
authors and does not necessarily represent the official views of the Lancet Glob Health 2020; 8: e264–75.
National Institutes of Health. 21 Wagner Z, Mukasa B, Nakakande J, Stecher C, Saya U, Linnemayr S.
Impact of the COVID-19 pandemic on use of HIV care, antiretroviral
References therapy adherence, and viral suppression: an observational cohort
1 UNICEF. Global and regional trends. 2021. https://data.unicef.org/ study from Uganda. J Acquir Immune Defic Syndr 2021; 88: 448–56.
topic/adolescents/hiv-aids/ (accessed June 13, 2022). 22 Mwangwa F, Charlebois ED, Ayieko J, et al. Two or more significant
2 Inzaule SC, Kroeze S, Kityo CM, et al. Long-term HIV treatment life-events in 6-months are associated with lower rates of HIV
outcomes and associated factors in sub-Saharan Africa: treatment and virologic suppression among youth with HIV in
multicountry longitudinal cohort analysis. AIDS 2022; 36: 1437–47. Uganda and Kenya. AIDS Care 2023; 35: 95–105.

e526 www.thelancet.com/hiv Vol 10 August 2023

 Articles

23 Johnson-Peretz J, Lebu S, Akatukwasa C, et al. “I was still very 30 Ford I, Norrie J. Pragmatic trials. N Engl J Med 2016; 375: 454–63.
young”: agency, stigma and HIV care strategies at school, baseline 31 WHO. Consolidated guidelines on HIV prevention, testing,
results of a qualitative study among youth in rural Kenya and treatment, service delivery and monitoring: recommendations for a
Uganda. J Int AIDS Soc 2022; 25 (suppl 1): e25919. public health approach. July 16, 2021. https://www.who.int/
24 Patel RC, Oyaro P, Thomas KK, et al. Point-of-care HIV viral load publications/i/item/9789240031593 (accessed July 10, 2023).
and targeted drug resistance mutation testing versus standard care 32 Dorward J, Sookrajh Y, Khubone T, et al. Implementation and
for Kenyan children on antiretroviral therapy (Opt4Kids): outcomes of dolutegravir-based first-line antiretroviral therapy for
an open-label, randomised controlled trial. people with HIV in South Africa: a retrospective cohort study.
Lancet Child Adolesc Health 2022; 6: 681–91. Lancet HIV 2023; 10: e284–94.
25 Drain PK, Dorward J, Violette LR, et al. Point-of-care HIV viral load 33 Jain V, Owaraganise A, Black D, et al. RAPID-VL intervention
testing combined with task shifting to improve treatment outcomes improves viral load ordering, results turnaround time and viral
(STREAM): findings from an open-label, non-inferiority, suppression: a cluster randomized trial in HIV clinics in Uganda.
randomised controlled trial. Lancet HIV 2020; 7: e229–37. International AIDS Society Conference; July 18–21, 2021
26 Leshargie CT, Demant D, Burrowes S, Frawley J. The proportion of (abstr 2418).
loss to follow-up from antiretroviral therapy (ART) and its 34 Tozan Y, Capasso A, Sun S, et al. The efficacy and cost-effectiveness
association with age among adolescents living with HIV in of a family-based economic empowerment intervention
sub-Saharan Africa: a systematic review and meta-analysis. (Suubi + Adherence) on suppression of HIV viral loads among
PLoS One 2022; 17: e0272906. adolescents living with HIV: results from a cluster randomized
27 Nyakato P, Schomaker M, Fatti G, et al. Virologic non-suppression controlled trial in southern Uganda. J Int AIDS Soc 2021;
and early loss to follow up among pregnant and non-pregnant 24: e25752.
adolescents aged 15–19 years initiating antiretroviral therapy in 35 Ekwunife OI, Ofomata CJ, Okafor CE, et al. Cost-effectiveness and
South Africa: a retrospective cohort study. J Int AIDS Soc 2022; feasibility of conditional economic incentives and motivational
25: e25870. interviewing to improve HIV health outcomes of adolescents living
28 Okoboi S, Ssali L, Yansaneh AI, et al. Factors associated with long- with HIV in Anambra State, Nigeria. BMC Health Serv Res 2021;
term antiretroviral therapy attrition among adolescents in rural 21: 685.
Uganda: a retrospective study. J Int AIDS Soc 2016; 19: 20841.
29 Mukamba N, Chilyabanyama ON, Beres LK, et al. Patients’
satisfaction with HIV care providers in public health facilities in
Lusaka: a study of patients who were lost-to-follow-up from HIV
care and treatment. AIDS Behav 2020; 24: 1151–60.

www.thelancet.com/hiv Vol 10 August 2023 e527