Effect of community-led delivery of HIV self-testing on HIV testing and antiretroviral therapy initiation in Malawi: A cluster-randomised trial.

BACKGROUND: Undiagnosed HIV infection remains substantial in key population subgroups including adolescents, older adults, and men, driving ongoing transmission in sub-Saharan Africa. We evaluated the impact, safety, and costs of community-led delivery of HIV self-testing (HIVST), aiming to increase HIV testing in underserved subgroups and stimulate demand for antiretroviral therapy (ART). METHODS AND
FINDINGS: This cluster-randomised trial, conducted between October 2018 and July 2019, used restricted randomisation (1:1) to allocate 30 group village head clusters in Mangochi district, Malawi to the community-led HIVST intervention in addition to the standard of care (SOC) or the SOC alone. The intervention involved mobilising community health groups to lead the design and implementation of 7-day HIVST campaigns, with cluster residents (≥15 years) eligible for HIVST. The primary outcome compared lifetime HIV testing among adolescents (15 to 19 years) between arms. Secondary outcomes compared: recent HIV testing (in the last 3 months) among older adults (≥40 years) and men; cumulative 6-month incidence of ART initiation per 100,000 population; knowledge of the preventive benefits of HIV treatment; and HIV testing stigma. Outcomes were measured through a post-intervention survey and at neighboring health facilities. Analysis used intention-to-treat for cluster-level outcomes. Community health groups delivered 24,316 oral fluid-based HIVST kits. The survey included 90.2% (3,960/4,388) of listed participants in the 15 community-led HIVST clusters and 89.2% (3,920/4,394) of listed participants in the 15 SOC clusters. Overall, the proportion of men was 39.0% (3,072/7,880). Most participants obtained primary-level education or below, were married, and reported a sexual partner. Lifetime HIV testing among adolescents was higher in the community-led HIVST arm (84.6%, 770/910) than the SOC arm (67.1%, 582/867; adjusted risk difference [RD] 15.2%, 95% CI 7.5% to 22.9%; p < 0.001), especially among 15 to 17 year olds and boys. Recent testing among older adults was also higher in the community-led HIVST arm (74.5%, 869/1,166) than the SOC arm (31.5%, 350/1,111; adjusted RD 42.1%, 95% CI 34.9% to 49.4%; p < 0.001). Similarly, the proportions of recently tested men were 74.6% (1,177/1,577) and 33.9% (507/1,495) in the community-led HIVST and SOC arms, respectively (adjusted RD 40.2%, 95% CI 32.9% to 47.4%; p < 0.001). Knowledge of HIV treatment benefits and HIV testing stigma showed no differences between arms. Cumulative incidence of ART initiation was respectively 305.3 and 226.1 per 100,000 population in the community-led HIVST and SOC arms (RD 72.3, 95% CI -36.2 to 180.8; p = 0.18). In post hoc analysis, ART initiations in the 3-month post-intervention period were higher in the community-led HIVST arm than the SOC arm (RD 97.7, 95% CI 33.4 to 162.1; p = 0.004). HIVST uptake was 74.7% (2,956/3,960), with few adverse events (0.6%, 18/2,955) and at US$5.70 per HIVST kit distributed. The main limitations include the use of self-reported HIV testing outcomes and lack of baseline measurement for the primary outcome.
CONCLUSIONS: In this study, we found that community-led HIVST was effective, safe, and affordable, with population impact and coverage rapidly realised at low cost. This approach could enable community HIV testing in high HIV prevalence settings and demonstrates potential for economies of scale and scope. TRIAL REGISTRATION: Clinicaltrials.gov NCT03541382.

Introduction

In 2018, approximately 1.7 million people were newly infected with HIV, with most cases in sub-Saharan Africa [1]. Regionally, almost one-fifth of people living with HIV were unaware of their HIV status [1]. Gaps remain more substantial among adolescents 15 to 19 years, older adults 40 years and above, and men [2]. While HIV incidence has been declining, undiagnosed HIV infection in these key population subgroups are drivers of ongoing transmission, impeding achievement of HIV elimination goals [1]. Routine HIV testing is a critical component of providing early diagnosis and treatment to reduce HIV-related morbidity and mortality and maximise HIV prevention benefits [3].

HIV testing services (HTS) are being provided within the context of declining undiagnosed HIV [4]. Most HTS are facility-based, though barriers including HIV-related stigma and discrimination, lack of convenience, and economic costs for clients have hindered uptake among underserved subgroups [5,6]. Community-based HTS can diagnose individuals at earlier stages of disease [7] and improve treatment and viral suppression when combined with convenient antiretroviral therapy (ART) services [8]. Despite their contributions, costs are higher for community-based HTS and global funding for community health programmes has been declining [7,9]. More efficient and scalable community strategies are needed to reach and maintain universal HIV testing in HIV-prevalent populations.

Among the most promising approaches are community-led strategies for disease prevention and management, which aim to involve underserved communities in decision-making, align health programmes with community-specific needs and experiences, and expand and sustain coverage [10]. Prior studies have shown increased coverage and efficiency and improved health outcomes when communities participate in the design and implementation of health services [11-14]. Within HIV, community groups have led demand creation activities for HIV services [12,15]. Recent innovations in self-care technologies are now expanding the breadth of services that could be delivered by communities [16].

HIV self-testing (HIVST) is a recommended approach that can facilitate novel HIV testing strategies [17]. Previous studies have demonstrated the effectiveness of door-to-door distribution of HIVST kits on increased HIV testing in Malawi and Zambia [18,19]. Given the impact of vertical community-based HIVST, we evaluated community-led delivery of HIVST. Specifically, we investigated the impact, safety, and costs of mobilising community health groups to lead the design and implementation of 7-day HIVST campaigns, aiming to increase HIV testing in underserved subgroups and stimulate demand for ART in a rural, high HIV prevalence area of Malawi.

Methods

Design

We conducted a cluster-randomised trial and allocated 30 group village head clusters to the community-led HIVST intervention in addition to the standard of care (SOC) or the SOC alone (S1 CONSORT Checklist). A cluster-randomised design was used since the intervention was delivered at group village head level. The study aimed to determine whether the intervention increased the proportion of the population who tested for HIV at cluster level, focusing on adolescents, older adults, and men. The trial also assessed impacts on population-level ART initiation, knowledge of the preventive benefits of HIV treatment, and HIV testing stigma; adverse events; and costs. A detailed protocol was published separately [https://hivstar.lshtm.ac.uk/protocols/] [20].

Setting and participants

Mangochi is a rural district bordering Lake Malawi and Mozambique with adult HIV prevalence of 10.1% (Fig 1) [21]. Group village heads hold customary authority over a group of villages. Government community health workers (CHWs) oversee provision of basic health services with community health action groups at the group village head level. Community volunteers, including village health committees, provide services at village level.

Fig 1

Map of clusters in Mangochi district.

HIVST, HIV self-testing; SOC, standard of care. Map of Mangochi district with health facilities and group village head-defined clusters. Malawi National Spatial Data Centre, www.masdap.mw.

Group village head clusters serviced by 5 government primary health centres were assessed by the study team for eligibility. Clusters were defined according to the boundaries of the group village head catchment area. Inclusion criteria for clusters prioritised a minimum population of 2,000 residents, distance of at least 5 kilometres to the health facility, and geographical separation between clusters. The study team obtained verbal consent from group village heads for cluster enrolment.

Randomisation

The 30 group village head clusters were randomised (1:1) to the community-led HIVST or SOC arm. Restricted randomisation was used to ensure balance between arms for key factors that could influence the effect of the intervention [22]. Restriction criteria included health facility, population, distance from facility, and number of villages (S1 Text). From 12,540 unique combinations falling within the restriction parameters, PPI and KF drew a computer-generated random sample of 1,000 combinations, which were sequentially numbered.

On July 16, 2018, group village head clusters were randomised at a public ceremony with community and government representatives. Volunteers selected numbered balls corresponding to one combination and one arm allocation from an opaque bag. Masking of community implementers and residents was not feasible since the intervention was delivered at cluster level, but data were managed and analysed without reference to arm allocation where possible.

Procedures

Community-led HIV self-testing

The community-led HIVST intervention involved engaging established community health groups from 15 group village head clusters to lead the design and implementation of HIVST campaigns in their areas. Implementation was staggered, with 2 to 3 clusters receiving the intervention every 14 days. Implementation was administered by the study team, including Population Services International Malawi, the Malawi-Liverpool-Wellcome Trust Clinical Research Programme, and the Ministry of Health. Formative research and piloting informed the design [20].

Following entrance meetings, the intervention proceeded in 3 stages, adapting participatory learning and action methods [23]. First, community health action groups and CHWs attended 2-day participatory workshops. Participants identified drivers of HIV infection, mapped HIV services and barriers to access, defined priority subgroups, and designed a 7-day HIVST campaign to be delivered in their areas. Specifically, participants planned strategies for distribution of HIVST kits, support for linkage to routine HIV care, demand creation for HIVST, social harms reporting, and monitoring and evaluation.

Second, community volunteers attended 2-day trainings on supporting use and interpretation of HIVST kits and providing information on linkage to routine HIV services, specifically confirmatory testing and ART initiation for reactive results, voluntary medical male circumcision (VMMC) for nonreactive results among men, and couples testing for serodiscordant results among partners. Volunteers were also trained on communication of HIV prevention messages, including effectiveness of ART, management of social harms, handling and storage of kits, and data collection.

Lastly, community volunteers delivered the campaign in their areas, supervised by community health action groups and CHWs. Implementation was based on strategies defined during participatory workshops for each cluster. HIVST sensitisation and distribution was conducted door-to-door and from fixed locations (e.g., schools, mosques, boreholes), social hotspots (e.g., fishing docks, sports fields, video shows), and community meetings. Support for linkage to routine services provided by community health groups included phone referrals and donated transportation funds. In addition to support provided by communities, the study team supplied the OraQuick HIV Self-Test (Orasure Technologies, Thailand), instructional materials, data collection tools, and nationally standardised gratuity of MWK 7,000 (US$10) per volunteer. HIVST kits could be taken by cluster residents aged 15 years and older. Residents could take an additional kit for secondary distribution and self-test with volunteer support or in private, with or without disclosing results.

Standard of care

The SOC, which was also available in community-led HIVST clusters, included HIV testing available through the Ministry of Health. HTS are provided by lay counsellors at health facilities through periodic community-based outreach. HIV testing follows standard serial testing algorithms using finger-prick rapid diagnostic tests, with ART universally available immediately following an HIV-positive diagnosis.

Outcomes and measurement

Outcomes were selected to understand the effect of the intervention on uptake of HIV services, especially among low-coverage subgroups. The primary outcome compared the proportion of adolescents (15 to 19 years) who self-reported lifetime testing for HIV between arms. Lifetime testing was a more relevant measure for adolescents since we anticipated that a high proportion of adolescents would have never tested [21], with the need for testing among this age group highly variable and dependent on the onset of sexual debut and HIV risk. We therefore hypothesised that the intervention would increase coverage of lifetime testing in a subgroup with limited previous experience of HIV testing, with a similar effect achieved on recent testing.

Secondary outcomes compared: self-reported recent HIV testing (in the last 3 months) among older adults (≥40 years) and men; cumulative 6-month incidence of ART initiation per 100,000 population; knowledge of the preventive benefits of HIV treatment; and HIV testing stigma. Exploratory outcomes compared: mutual knowledge of HIV status between sexual partners; recent HIV testing for adolescents; lifetime HIV testing for older adults and men; and HIV testing in the last 12 months for adolescents, older adults, and men. Cumulative incidence of VMMC was not assessed as specified in the protocol as services were discontinued prior to study initiation.

Outcomes were measured at cluster level through a post-intervention survey, except for ART initiations, which were captured at the 5 health facilities. The survey was administered 8 to 12 weeks after the intervention start in the community-led HIVST clusters or matched dates in the SOC clusters. Cluster residents were sampled for the survey to form the evaluation population. Within each cluster, villages with at least 500 residents and that included or was located near the group head village were randomly selected per cluster. In villages with approximately 500 residents, all households were eligible for the survey. In larger villages, 150 households were recruited in a clockwise spiral starting with the village head household, with multiple visits made to schedule interviews. Written informed consent or assent was obtained for residents aged 15 years and older in recruited households. Participants were interviewed on household and sociodemographic characteristics, HIV service use, and sexual behavior. A random sample (approximately 20%) received an HIV knowledge and attitudes module (Text A in S2 Text).

Clinic assistants at the 5 health facilities interviewed ART patients aged 15 years and older to establish cluster eligibility for 6 months following the intervention. Population estimates for cluster residents aged 15 years and older were obtained from village and facility registers and used as the denominator for cumulative incidence of ART initiations.

Process indicators measuring HIVST coverage were assessed through the survey and HIVST registers, which recorded sociodemographic information for residents collecting HIVST kits. Adverse events related to HIVST were captured through the survey and classified by severity [24].

Economic data on the total and unit costs of the intervention were collected from the provider perspective, with financial costs from expenditure records supplemented with full costs from direct observations and interviews (Text B in S2 Text). Costs are reported in 2018 US Dollars.

Planned outcome evaluation of a repeat community-led HIVST intervention [20] was not completed due to delayed study initiation, leaving insufficient time between post-intervention assessments. The study reported here had standalone outcomes and was not affected.

Sample size

The study was powered to detect a 20% absolute difference between study arms in the primary outcome of lifetime HIV testing for adolescents [18]. We assumed 35% to 50% prevalence of testing among adolescents in the SOC arm based on national estimates [21]. Fifteen group village head clusters per arm with 50 adolescents of 250 residents per cluster provided 90% power at a 5% significance level. We assumed a coefficient of variation (k) of 0.25 based on guidelines for cluster-randomised trials [22]. The study was also powered to measure a difference in recent HIV testing in older adults and men and cumulative ART initiations (S1 Text).

Statistical analysis

Analysis used intention-to-treat, that is participants within clusters were analysed based on cluster assignment to study arms rather than individual-level exposure to the intervention. Outcomes were analysed at cluster level using established methods for cluster-randomised trials with a small number of clusters [22]. Specifically, risk differences, mean differences, and risk ratios for the intervention effect were calculated from cluster-level risks, means, and log risks, respectively (S1 Text). Cluster-level summaries were then compared between arms with a t test.

Using a 2-stage approach [22], effect estimates were adjusted for sex and age group a priori and any imbalance between arms in adolescent covariates. To estimate the risk difference and risk ratio, the first stage used logistic regression to adjust for confounding bias at the individual level. Predicted risks were then summed at cluster level and used to calculate the difference and ratio of observed and predicted values. A log transformation was applied to summaries as appropriate. The second stage used a t test to compare covariate-adjusted summary values between arms. To calculate the mean difference, similar procedures were applied using linear regression in the first stage.

A priori subgroup analysis compared the primary outcome by sex and age group (15 to 17 years, 18 to 19 years). Post hoc analysis compared cumulative incidence of ART initiation by first and last 3-month period. Statistical analysis used Stata version 14.0.

Ethical considerations

The study is registered with ClinicalTrials.gov, NCT03541382. Ethical approvals were granted by the University of Malawi College of Medicine (P.01/18/2332), London School of Hygiene & Tropical Medicine (14761), and WHO (STAR-comm led CRT-Malawi). The study is part of the Unitaid/Population Services International HIV Self-Testing Africa Initiative (STAR) [http://hivstar.lshtm.ac.uk/].

Results

The study population included 44,543 residents in 15 community-led HIVST clusters and 39,806 residents in 15 SOC clusters. The intervention was delivered from October 5, 2018 to January 17, 2019 by 157 community health action group members (cluster mean 10.5) and 190 community volunteers (cluster mean 12.7). Overall, 24,316 HIVST kits (cluster mean 1,621) were distributed, with 47.2% (n = 11,472) of kits distributed to men.

Outcomes were measured from December 5, 2018 to March 30, 2019 for the post-intervention survey and to July 31, 2019 for facility data collection. Fig 2 shows the trial flow diagram.

Fig 2

Trial flow diagram.

Flow diagram of the cluster-randomised trial.

The survey included 90.2% (3,960/4,388) and 89.2% (3,920/4,394) of listed participants, respectively, in the community-led HIVST and SOC arms. Adolescent participation rates were similar at 90.2% (910/1,002) in the community-led HIVST arm and 86.4% (867/1,004) in the SOC arm.

Participant characteristics are summarised in Table 1. Overall, the proportion of men was 39.0% (3,072/7,880), which was below expected [21] with 84.6% (1,577/1,863) and 82.4% (1,495/1,814) responding in the community-led HIVST and SOC arms, respectively.

Table 1

Comparison of population characteristics by study arm.

	Community-led HIVST	SOC
	n (%)	n (%)
Household characteristics	(N = 1,994)	(N = 2,015)
Adults (median/range)*	2 (0–8)	2 (0–10)
Children (median/range)*	1 (0–1)	1 (0–1)
Household wealth index†
Lowest	368 (20.3%)	341 (18.6%)
Second	353 (19.4%)	395 (21.6%)
Third	361 (19.9%)	362 (19.8%)
Fourth	358 (19.7%)	373 (20.4%)
Highest	375 (20.7%)	358 (19.6%)
Individual characteristics	(N = 3,960)	(N = 3,920)
Male	1,577 (39.8%)	1,495 (38.1%)
Age (median/range)	29 (15–96)	29 (15–98)
Age group
15–19 years	910 (23.0%)	867 (22.1%)
20–24 years	631 (15.9%)	675 (17.2%)
25–39 years	1,253 (31.6%)	1,267 (32.3%)
≥40 years	1,166 (29.4%)	1,111 (28.3%)
Marital status‡
Married or living together	2,428 (61.3%)	2,467 (62.9%)
Separated, divorced, or widowed	612 (15.5%)	542 (13.8%)
Never married	918 (23.2%)	910 (23.2%)
Educational attainment§
None	1,730 (43.7%)	1,764 (45.0%)
Primary	1,902 (48.0%)	1,838 (46.9%)
Secondary or higher	328 (8.3%)	317 (8.1%)
Literate||	2,196 (55.5%)	2,066 (52.7%)
Muslim	2,840 (71.7%)	3,008 (76.7%)
Ethnicity
Yao	2,778 (70.2%)	2,942 (75.1%)
Ngoni	546 (13.8%)	443 (11.3%)
Other	636 (16.1%)	535 (13.6%)
Resident in the last 2 months	3,877 (97.9%)	3,830 (97.7%)
Self-rated health status¶
Very good	1,546 (39.1%)	1,314 (33.5%)
Good	1,738 (43.9%)	1,810 (46.2%)
Fair	338 (8.5%)	389 (9.9%)
Poor	337 (8.5%)	407 (10.4%)
Reported current sexual partner**	2,875 (72.6%)	2,931 (74.8%)
Circumcised (for men)††	1,335 (84.9%)	1,285 (86.0%)

HIVST, HIV self-testing; SOC, standard of care.

*32 missing values in the community-led HIVST arm and 8 missing values in the SOC arm.

†179 missing values in the community-led HIVST arm and 186 missing values in the SOC arm.

‡2 missing values in the community-led HIVST arm and 1 missing value in the SOC arm.

§1 missing value in the SOC arm.

||1 missing value in the community-led HIVST arm.

¶1 missing value in the community-led HIVST arm.

**1 missing value in the SOC arm.

††5 missing values in the community-led HIVST arm.

Most participants obtained primary-level education or below. The majority were married and reported a sexual partner. Characteristics were well balanced by arm, though differences in literacy, religion, ethnicity, and self-reported health status were observed for adolescents (Table 2).

Table 2

Comparison of characteristics for adolescents (15–19 years) by study arm.

	Community-led HIVST	SOC
	n (%)	n (%)
Individual characteristics	(N = 910)	(N = 867)
Male	387 (42.5%)	381 (43.9%)
Age (median/range)	18 (15–19)	18 (15–19)
Age group
15–17 years	400 (44.0%)	384 (44.3%)
18–19 years	510 (56.0%)	483 (55.7%)
Marital status*
Married or living together	138 (15.2%)	147 (17.0%)
Separated, divorced, or widowed	34 (3.7%)	20 (2.3%)
Never married	738 (81.1%)	699 (80.7%)
Educational attainment
None	239 (26.3%)	262 (30.2%)
Primary	604 (66.4%)	552 (63.7%)
Secondary or higher	67 (7.4%)	53 (6.1%)
Literate	667 (73.3%)	577 (66.6%)
Muslim	672 (73.8%)	686 (79.1%)
Ethnicity
Yao	665 (73.1%)	684 (78.9%)
Ngoni	126 (13.8%)	88 (10.1%)
Other	119 (13.1%)	95 (11.0%)
Resident in the last 2 months	879 (96.6%)	844 (97.3%)
Self-rated health status†
Very good	416 (45.8%)	328 (37.8%)
Good	406 (44.7%)	449 (51.8%)
Fair	46 (5.1%)	42 (4.8%)
Poor	41 (4.5%)	48 (5.5%)
Reported current sexual partner	389 (42.7%)	390 (45.0%)
Circumcised (for men)	340 (87.9%)	346 (90.8%)

HIVST, HIV self-testing; SOC, standard of care.

*1 missing value in the SOC arm.

†1 missing value in the community-led HIVST arm.

Primary and secondary outcomes

Lifetime HIV testing among adolescents was higher in the community-led HIVST arm (84.6%, 770/910) than the SOC arm (67.1%, 582/867), with adjusted risk difference (RD) of 15.2% (95% CI 7.5% to 22.9%; p < 0.001; Table 3 and Table A in S2 Text and Fig A in S2 Text). There was strong evidence that the effect of the intervention differed by age group (p-value for interaction = 0.02), with a more pronounced difference among 15 to 17 year olds (adjusted RD 21.5%, 95% CI 10.4% to 32.6%; p < 0.001) than 18 to 19 year olds (adjusted RD 10.8%, 95% CI 4.3% to 17.3%; p = 0.002). Lifetime testing was also higher for boys (adjusted RD 20.5%, 95% CI 10.7% to 30.3%; p < 0.001) than girls (adjusted RD 11.1%, 95% CI 2.8% to 19.4%; p = 0.01; p-value for interaction = 0.06).

Table 3

Primary and secondary outcomes by study arm.

	Community-led HIVST		SOC		Risk or mean difference (95% CI)	Adjusted risk or mean difference (95% CI)*	Risk ratio (95% CI)	Adjusted risk ratio (95% CI)*
	n/N (%)	GM	n/N (%)	GM	p-value	p-value	p-value	p-value	k
Primary outcome
Lifetime HIV testing among adolescents 15–19 years	770/910 (84.6%)	84.6%	582/867 (67.1%)	67.2%	16.4% (7.8%–25.0%) <0.001	15.2% (7.5%–22.9%) <0.001	1.26 (1.11–1.43) <0.001	1.24 (1.11–1.39) <0.001	0.13
Stratified by age group†
15–17 years	320/400 (80.0%)	79.5%	219/384 (57.0%)	54.3%	22.5% (9.8%–35.3%) 0.001	21.5% (10.4%–32.6%) <0.001	1.47 (1.15–1.87) 0.003	1.44 (1.16–1.79) 0.002
18–19 years	450/510 (88.2%)	88.0%	363/483 (75.2%)	76.0%	11.5% (4.3%–18.7%) 0.003	10.8% (4.3%–17.3%) 0.002	1.16 (1.06–1.27) 0.003	1.15 (1.06–1.25) 0.002
Stratified by sex‡
Male	309/387 (79.8%)	79.6%	218/381 (57.2%)	56.6%	22.3% (11.9%–32.7%) <0.001	20.5% (10.7%–30.3%) <0.001	1.41 (1.19–1.66) <0.001	1.37 (1.17–1.6) <0.001
Female	461/523 (88.1%)	88.0%	364/486 (74.9%)	74.9%	11.8% (2.6%–21.0%) 0.01	11.1% (2.8%–19.4%) 0.01	1.17 (1.04–1.33) 0.014	1.17 (1.04–1.31) 0.01
Secondary outcomes
HIV testing in the last 3 months among adults ≥40 years§	869/1,166 (74.5%)	73.1%	350/1,111 (31.5%)	30.9%	42.3% (34.7%–50.0%) <0.001	42.1% (34.9%–49.4%) <0.001	2.37 (2.0–2.79) <0.001	2.36 (2.01–2.77) <0.001	0.16
HIV testing in the last 3 months among men	1,177/1,577 (74.6%)	73.8%	507/1,495 (33.9%)	33.3%	40.8% (32.9%–48.6%) <0.001	40.2% (32.9%–47.4%) <0.001	2.22 (1.91–2.57) <0.001	2.19 (1.91–2.51) <0.001	0.17
ART initiation per 100,000 population across 6 months||	136/44,543 (305.3)	270.5	90/39,806 (226.1)	207.3	72.3 (−36.2–180.8) 0.18		1.31 (0.84–2.03) 0.23		0.34
Stratified by post-intervention period ¶
First 3 months	83/44,543 (186.3)	184.2	37/39,806 (93.0)	97.5	97.7 (33.4–162.1) 0.004		1.89 (1.21–2.95) 0.007
Last 3 months	53/44,543 (119.0)	108.0	53/39,806 (133.2)	122.8	−10.7 (−80.5–59.2) 0.76		0.88 (0.51–1.51) 0.63
Knowledge of the preventive benefits of HIV treatment**		15.0		14.7	0.3 (−0.6–1.3) 0.51	0.4 (−0.4–1.3) 0.29
HIV testing stigma††		7.4		7.6	−0.2 (−0.6–0.2) 0.36	−0.2 (−0.5–0.2) 0.37

ART, antiretroviral therapy; GM, geometric mean (of cluster-level proportions); HIVST, HIV self-testing; k, coefficient of variation in group village head-defined clusters; SOC, standard of care.

*Analysis adjusted for sex, age group, literacy, religion, ethnicity, and health status. Analysis among adolescents defines levels of age group as 16–17 years and 18–19 years. Analysis among adults ≥40 years defines levels of age group as 40–49 years and ≥50 years. Analysis among men adjusts for the same covariates except for sex.

†p-Value for interaction, p = 0.02.

‡p-Value for interaction, p = 0.06.

§Testing in the last 3 months was ascertained based on the most recent test date. If the month of the test date was not reported, we counted the test as being in the last 3 months if the test date was in 2018 for interview dates in 2018 or if the test date was in 2018 or 2019 for interview dates in 2019. 113 and 156 participants in the community-led HIVST and SOC arms, respectively, did not report month data. We conducted a sensitivity analyses where test dates with missing months were not counted as being in the last 3 months. Among adults ≥40 years, community-led HIVST: 72.0% (839/1,166), SOC: 27.0% (300/1,111); adjusted RD 43.7%, 95% CI 36.0%–51.5%; p < 0.001. Among men, community-led HIVST: 72.0% (1,135/1,577), SOC: 30.7% (459/1,495); adjusted RD 40.4%, 95% CI 32.7%–48.0%; p < 0.001.

||Denominator for ART initiations is the estimated cluster population of adults ≥15 years, which was estimated using village and health facility registers and the proportion of adults reported in household enumeration.

¶Post hoc analysis. p-Value for interaction, p = 0.02.

**N = 1,925, with 30 missing values. Score is the sum of 5 questions using 5-point likert scale, with range of 5–25 (low to high knowledge).

††N = 1,929, with 26 missing values. Score is the sum of 6 questions using 3-point likert scale, with range of 3–18 (low to high stigma).

Recent HIV testing (in the last 3 months) among older adults was higher in the community-led HIVST arm (74.5%, 869/1,166) than the SOC arm (31.5%, 350/1,111), with adjusted RD of 42.1% (95% CI 34.9% to 49.4%; p < 0.001). The proportion of recently tested men was 74.6% (1,177/1,577) in the community-led HIVST arm and 33.9% (507/1,495) in the SOC arm (adjusted RD 40.2%, 95% CI 32.9% to 47.4%; p < 0.001). Knowledge of the preventive benefits of HIV treatment and HIV testing stigma measures showed no differences between arms (Table 3).

Cumulative 6-month incidence of ART initiation was, respectively, 305.3 and 226.1 per 100,000 population in the community-led HIVST and SOC arms (RD 72.3, 95% CI −36.2 to 180.8; p = 0.18). In post hoc analysis, cumulative incidence in the 3-month post-intervention period was, respectively, 186.3 and 93.0 per 100,000 population in the community-led HIVST and SOC arms, with a larger effect in the first 3 months (RD 97.7, 95% CI 33.4 to 162.1; p = 0.004) than the last 3 months (RD −10.7, 95% CI −80.5 to 59.2; p = 0.76; p-value for interaction = 0.02).

In exploratory analyses, the intervention increased HIV testing in 3-month, 12-month, and lifetime periods, overall and among defined subgroups, and mutual knowledge of HIV status between sexual partners (adjusted RD 14.1%, 95% CI 8.6% to 19.5%; p < 0.001; Table A in S2 Text).

Process outcomes

Self-reported HIVST uptake was 74.7% (2,956/3,960) in the community-led HIVST arm, ranging from 68.5% in older men to 84.7% in young women (20 to 24 years), and 3.7% (145/3,920) in the SOC arm (Table 4 and Fig B in S2 Text). The proportion of participants aware of HIVST was 95.3% (3,771/3,960) and 32.4% (1,268/3,920) in the community-led HIVST and SOC arms, respectively.

Table 4

Fidelity to community-led HIV self-testing intervention.

	Community-led HIVST	SOC
	n (%)	n (%)
	(N = 3,960)	(N = 3,920)
Heard of self-testing*	3,771 (95.3%)	1,268 (32.4%)
Ever self-tested†	2,956 (74.7%)	145 (3.7%)
Self-tested in the last 3 months‡	2,919 (73.7%)	128 (3.3%)
For most recent self-test:	(N = 2956)
Self-test distributor§	2,775 (93.9%)
Community health volunteer	2,775 (93.9%)
Family member	130 (4.4%)
Other	49 (1.7%)
Self-test collection location||
Home	2,392 (80.9%)
Home of community health volunteer	220 (7.4%)
Other	343 (11.6%)
First test ever	306 (10.4%)
Self-test result**
Positive††	70 (2.4%)
Negative	2,873 (97.4%)
Invalid	8 (0.3%)
Harmed before or after self-testing‡‡	18 (0.6%)

HIVST, HIV self-testing; SOC, standard of care.

*1 missing value in the community-led HIVST arm and 1 missing value in the SOC arm.

†1 missing value in the community-led HIVST arm.

‡1 missing value in the community-led HIVST arm.

§2 missing values in the community-led HIVST arm.

||1 missing value in the community-led HIVST arm.

¶1 missing value in the community-led HIVST arm.

**5 missing values in the community-led HIVST arm.

††40% (n = 28) were newly HIV–positive, 11.4% (n = 8) were previously diagnosed and not on treatment, 48.6% (n = 34) were previously diagnosed and on treatment. Of 36 HIV–positive and not on treatment, 58.3% (n = 21) initiated on antiretroviral therapy.

‡‡1 missing value in the community-led HIVST arm.

Of 2,956 self-testers in the community-led HIVST arm, most obtained HIVST kits through primary distribution from community health groups (93.9%, n = 2,775). Only 4.4% (n = 130) received HIVST kits through secondary distribution from family members.

The majority of kits were obtained at the home of the participant (80.9%, n = 2,392) followed by the home of community health volunteers (7.4%, n = 220). Further, 10.4% (n = 306) reported no previous HIV testing and 2.4% (n = 70) reported an HIV-positive result, of whom 40.0% (n = 28) were newly identified and 11.4% (n = 8) were previously diagnosed and not on treatment. Self-reported ART initiation was 58.3% (21/36).

Adverse events related to HIVST were reported by 0.6% of participants (18/2,955) and classified by severity [24]. Reports included forced self-testing or results disclosure (moderate grade) and one case of physical harm (moderate to severe grade).

Costs

Total provider cost of community-led HIVST was US$138,624, with a mean cost of US$ US$5.70 per HIVST kit distributed (Table B in S2 Text). Unit costs were US$241 per HIV–positive identified, US$602 per new HIV–positive identified, and US$468 per HIV–positive identified not on treatment.

Discussion

Community-led delivery of 7-day HIVST campaigns linked to HIV treatment and prevention increased HIV testing in underserved subgroups. Lifetime testing increased by 15.2% for adolescents, with more pronounced differences among younger adolescents and boys. Recent testing increased by 42.1% for older adults and by 40.2% for men. Mutual knowledge of HIV status between sexual partners also improved. Cumulative incidence of ART initiation per 100,000 population apparently increased 3 months post-intervention, with 186.3 residents treated in the community-led HIVST arm compared with 93.0 residents treated in the SOC arm. Difference in ART initiations between arms was not found for the predefined 6-month period. The intervention also achieved 74.7% HIVST uptake with limited adverse events and at US$5.70 per HIVST kit distributed. Our study provides evidence of an effective, safe, and affordable community strategy that rapidly achieved high impact and coverage at low cost and could be scaled in priority settings to meet and maintain HIV elimination goals.

To our knowledge, this is the first randomised trial to assess the impact of community-led delivery of HTS, which was recently enabled by the introduction of HIVST. This is also one of the few studies to report high coverage of HIV testing among subgroups with substantial undiagnosed HIV infection. Community participation has long been advocated as fundamental to primary healthcare and an approach that could increase coverage and efficiency of health programmes, improve outcomes, enhance the capacity of communities to address ill-health, and contribute to the sustainability of community health programmes [10]. We used participatory methods to engage established community health groups in designing and implementing HIVST campaigns adapted to their respective contexts. Our study builds on “top-down” community-based HIV testing and self-testing [18, 19] and “bottom-up” community-led demand creation for HIV services [12,15] by using a community-led HIVST model. Future iterations of this intervention could engage groups over time to provide repeat or multidisease services, including interventions to address malaria, tuberculosis, and other priority disease areas [25]. With the COVID-19 epidemic, community-led disease control programmes have potential to contribute to surveillance and early detection, reporting, and management. HIVST may also enable ongoing provision of HIV testing as routine HIV services are disrupted and reduce demand on healthcare workers to provide in-person HIV testing [26].

We found that community-led HIVST can lead to high coverage and effective targeting, with our study reporting substantially higher HIVST uptake from a community-led approach than a previous study of door-to-door HIVST [18]. Uptake was consistent across adolescents, older adults, and men. In contrast, vertical distribution of HIVST kits by community-based distribution agents achieved 42.5% uptake across a 12-month period in Malawi [18]. Uptake may be driven by successful context-informed planning, trust between community health groups and community members, and the value and novelty of HIVST among HIV-prevalent communities. The intervention also had minimal adverse events, alleviating safety concerns around decentralising management of HIVST implementation [16]. Further, our results may be applicable to high HIV prevalence settings in sub-Saharan Africa with similar cadres of community health workers and volunteers.

We showed increased lifetime and recent HIV testing in adolescents, especially younger adolescents and boys, older adults, and men, with prevalence of undiagnosed HIV disproportionately concentrated in these subgroups. Mutual knowledge of HIV status between sexual partners also increased. The intervention effect, while slightly lower than assumed for sample size calculations, was achieved against a SOC that included a high saturation of HIV services, with Mangochi a priority district for the Ministry of Health. Diagnosis of recent HIV infection is critical for HIV prevention, with our study reinforcing the importance of community strategies in reaching underserved subgroups [7]. Further, the impact attained within a short period of time makes community-led HIVST a promising candidate for national HIV programmes to consider for periodic implementation to reach underserved subgroups.

Community-led HIVST had an immediate impact on ART initiation 3 months post-intervention, though the effect diminished at 6 months. Population-level impact was measured even as the post-intervention survey reported that 1.2% of self-testers were newly HIV–positive or previously diagnosed but not on treatment, underscoring the potential for HIVST to influence ART demand. The intervention involved engaging government CHWs and health facilities to facilitate linkage to routine HIV services, likely contributing to successful referrals. However, self-reported ART initiation was 58.3% at follow-up. Optimising timely linkage to HIV treatment and prevention services is essential to maximise the health benefits from HIV testing and self-testing [27]. Neither VMMC nor preexposure prophylaxis were available at primary care level during the study, so we were unable to evaluate uptake of these services as intended. Further, despite providing training and materials to community volunteers on HIV prevention messages, the intervention did not improve knowledge of the preventive benefits of HIV treatment. The absence of effect may reflect insufficient discussion on the topic or difficulties conveying HIV risk reduction concepts.

Our analysis reported average cost of US$5.70 per HIVST kit distributed, which was lower than the cost of door-to-door HIVST models in nearby rural districts (2017 US$8.15) and urban Blantyre (2014 US$8.78) [28,29]. Average costs of community-based HTS in sub-Saharan Africa were similar [7]. Community health programmes are important for epidemic preparedness and management but can be costly to implement [9]. A community-led approach to HIVST is likely to realise significant economies of scale, with potential cost savings when community health groups are mobilised nationally and recurrently by Ministries of Health in nonresearch settings. Economies of scope can also lead to greater efficiency by implementing HIVST within a package of interventions addressing a broader set of conditions, including through the use of self-care products. Further, we found that the cost per HIV–positive identified through HIVST was US$241 to US$602, with modelling studies suggesting a cost-effectiveness threshold of US$315 per new HIV diagnosis [4]. Further cost reductions and uptake among undiagnosed, untreated HIV–positive persons or high HIV risk persons linking to HIV prevention would ensure greater probability of community-led HIVST as a cost-effective strategy.

Our study had multiple limitations. Due to the pragmatic nature of the intervention, there was some contamination of HIVST in the SOC arm, although reported events were nominal. The study design did not allow us to isolate the effects of specific intervention components, including the use of participatory methods and introduction of HIVST. Primary and secondary outcomes on HIV testing were self-reported and subject to misreporting due to recall or social desirability bias, including overreporting in the community-led HIVST arm following exposure to the intervention. We had a small number of clusters per arm and aimed to mimimise bias through randomisation of clusters, using restriction of factors likely to be associated with the outcome, and adjustment for imbalances between arms in individual characteristics. However, we did not measure primary and secondary outcomes in a baseline sample and adjust for baseline HIV testing. Our sampling frame may have included households that had better access to the HIVST intervention due to their location, with potential overestimation of the intervention effect. However, the effect size was relatively large, and our conclusions would have likely remained unchanged. The survey included fewer men than expected with almost one-fifth of eligible men not found. Implementation occurred within a controlled research setting as part of a mature HIVST programme that had been operating since 2015, potentially affecting the generalisability of our costs. Adverse events were reported in the survey, with follow-up to obtain details not feasible. We also did not evaluate accuracy of HIVST, which was previously shown to be high in rural settings when given optimised instructional materials and brief demonstrations [30].

Community-led delivery of 7-day HIVST campaigns linked to HIV treatment and prevention was effective in increasing HIV testing in adolescents, older adults, and men and mutual knowledge of HIV status between sexual partners. Population-level ART initiation apparently increased within a 3-month period but showed no difference at 6 months. Community-led delivery of HIVST was safe and associated with higher uptake and lower costs compared with previous evaluations of vertical community-based HIV testing and self-testing. Given evidence of high population impact and coverage rapidly realised at low cost, community-led HIVST shows much promise as an effective, safe, and affordable strategy, while entrusting communities with leading solutions for disease control. This approach could enable community HIV testing in high HIV prevalence settings and demonstrates potential for economies of scale and scope.

Checklist of information to include when reporting a cluster-randomised trial.

(DOCX)

Click here for additional data file.

Statistical analysis plan.

Statistical analysis plan for primary and secondary outcomes.

(DOCX)

Click here for additional data file.

Supporting Information.

Supporting text, tables, and figures.

(DOCX)

Click here for additional data file.

5 Aug 2020

Dear Dr Indravudh,

Thank you for submitting your manuscript entitled "Effect of community-led delivery of HIV self-testing on HIV testing and antiretroviral therapy initiation in Malawi: a cluster-randomised trial" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external assessment.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by .

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Richard Turner, PhD

Senior editor, PLOS Medicine

rturner@plos.org

9 Sep 2020

Dear Dr. Indravudh,

Thank you very much for submitting your manuscript "Effect of community-led delivery of HIV self-testing on HIV testing and antiretroviral therapy initiation in Malawi: a cluster-randomised trial" (PMEDICINE-D-20-03726R1) for consideration at PLOS Medicine.

Your paper was evaluated by the editors and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to invite you to submit a revised version that addresses the reviewers' and editors' comments fully. You will appreciate that we cannot make a decision about publication until we have seen the revised manuscript and your response, and we expect to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We hope to receive your revised manuscript by Sep 30 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

Please let me know if you have any questions, Otherwise, we look forward to receiving your revised manuscript in due course.

Sincerely,

Richard Turner, PhD

Senior Editor, PLOS Medicine

rturner@plos.org

-----------------------------------------------------------

Requests from the editors:

Please state briefly in the submission form where study data can be found - e.g., in supplementary files.

Please quote aggregate demographic details for study participants in the abstract.

Please add a new final sentence to the "Methods and findings" subsection of your abstract, quoting 2-3 of the study's main limitations.

At line 69, please begin the sentence with "In this study, we found that ..." or similar.

In the abstract and elsewhere, please quote p values alongside 95% CI, where available.

After the abstract, we will need to ask you to add a new and accessible "author summary" section in non-identical prose. You may find it helpful to consult one or two recent research papers in PLOS Medicine to get a sense of the preferred style.

Please remove the information on funding from the end of the main text. In the event of publication, this information will appear in the article metadata via entries in the submission form.

Throughout the paper, please adapt reference call-outs to the following style: "... [5,6]." (i.e., no spaces within the square brackets).

In reference 25, should that be "Taylor and Francis Group"?

Please adapt the attached CONSORT checklist so that individual items are referred to by section (e.g., "Methods") and paragraph numbers rather than by line or page numbers, as the latter generally change in the event of publication. Please refer to this in the Methods section (e.g., "See S1_CONSORT_Checklist").

Comments from the reviewers:

*** Reviewer #1:

This article describes a cluster randomized trial in Malawi of community-led delivery of HIV self-testing with a focus on impact safety and cost. The study takes on an important topic in that lack of awareness of HIV, while decreasing overall, remains high in some sub-populations and testing is the first step towards increasing testing, increasing treatment initiation, decreasing community viral suppression and eventually decreasing incidence. The study was well conducted and the cluster randomized design was both appropriate and helps to increase the validity of the results. I commend the authors on their work and I am generally disposed toward supportingg publication. Still, there are some important limitation to the work which need to be dealt with prior to this study being ready for publication.

Below are specific comments about the manuscript.

Running the intervention in conjunction with local community groups is a very nice strength of the approach and will hopefully make it more sustainable.

I appreciate that the program provided support for travel for linkage to care. But is this a sustainable model? Can you talk more about this in the discussion.

The number of clusters was small, so the potential for bias remains and this is worth discussing. Still, the participants were well balanced so the amount of bias is hopefully small. What is missing is the most important thing, however, which is lifetime risk of testing before the intervention period. This should be in the discussion as well.

SOC was not well described. How much testing was already being done? It would seem that the effectiveness of the intervention would be in part driven by how much people already test so this is important. Also, for the intervention arm, did facility testing continue such that the intervention is really home and facility testing? Were any other testing campaigns being done at the time.

Could there have been any crossover/contamination? I suspect not, but I can't tell from the paper how close the populations were to each other and if the intervention were delivered in a way that someone could access them from outside.

To that point, I found details on delivery of the intervention to be lacking. You did a nice job describing the process by which the intervention was developed, but practically how were test kits delivered? They went door to door, but how often? Did they have a list of all homes to go to?

It isn't clear in the methods who the study population is. Everyone in the community? Just those who wanted testing? This is a bit clearer in the results, but the idea of an evaluation population isn't made very clear in the methods. Why wasn't everyone in the cluster included in the population? I assume for cost and time reasons, but then why wasn't a random sample taken?

I have some concern about the way the evaluation population was chosen as much as I understand it. Starting with the village head household could potentially lead to those houses with more access, high probability of uptake being chosen and those households farther away which may be less likely to uptake any HIV testing being less likely to participate.

The fact that outcomes are self-reported is a limitation. You mention this in the discussion but this needs more discussion. What is the likely direction and magnitude of this bias? I would think that those in the home testing group might have more incentive to say they tested due to the increased community mobilization, so it would bias away from the null.

In the statistical analysis, I was unclear on what this means "Risk/mean differences and

risk ratios were calculated from cluster-level summaries". Most importantly, what approach was used to adjust for the clustered design? I assume yes because you accounted for it in your sample size and using your data without cluster adjustment, I get narrower intervals, but I'm just not clear on how it was done.

In the sample size section, and this is not really very important, but I'm curious why your control arm had much higher lifetime testing than you anticipated.

In the sample size you specified a 20% difference. Was this the meaningful difference? I ask because you didn't meet this even though you did find a significant result.

*** Reviewer #2:

Thank you for the opportunity review this manuscript on the effect of community-led delivery of HIV self-testing on case finding and ART initiation in Malawi. While some countries are approaching epidemic control, others are still lagging, and we must work in those countries to identify undiagnosed PLHIV. Particularly in the time of COVID-19, HIV self-testing represents a unique opportunity to provide a wide coverage of testing services to those who are unable to access testing sites due to stigma, transportation cost, lock-down restrictions, lack of awareness, etc. This manuscript provides important results indicating the efficacy of community HIV self-testing in Malawi, and will be important for moving case identification strategies forward.

Minor Revisions:

Methods: The inclusion/exclusion criteria is not entirely clear. Please better define how participants were selected and what the inclusion/exclusion criteria were for each study arm.

Was counseling or social support provided to those who participated in HIV self-testing?

Linkage to ART services for those who tested positive in SOC was mentioned, but was linkage to ART services also provided to those participating in HIV self-testing?

Statistical analysis: An alpha value was not denoted in the statistical analysis section, and therefore it is unclear what level of significance is being tested.

Discussion: In reference to the current atmosphere with COVID-19, the use of HIV self-testing as a way to limit clinic visits, reduce HRH testing demands, and reduce in-person testing services could be added to strengthen the argument for community-led HIV self-testing implementation and expansion.

*** Reviewer #3:

This is a useful cluster-randomised trial on the effect of community-led delivery of HIV self-testing on HIV testing and antiretroviral therapy initiation in Malawi. However, there are quite a few issues needing attention.

1) Strictly speaking, The analyses are not intention to treat analyses. The definition of intention to treat analysis is 'once randomised always analysed'. However, around 10% of adolescents in each arm was missing for the primary analysis, and only 25% of those > 15 years old in each arm was available for the secondary analyses. These all happened after randomisation so could introduce potential biases. The current analyses are basically 'complete case analysis'.

2) The primary outcome is the lifetime HIV testing among adolescents (15-19 years) self-testing vs standard care. However, it's confusing on how this short-term intervention (7 days) over a period less than a year was related to lifetime HIV testing? Because it's more or less a cross-section survey on HIV testing. There was neither lifetime follow-up nor prior information before the trial on the testing. The second outcome on short-term measure might be more related to the effect of the intervention, but lifetime testing? Can authors justify the appropriateness of the primary outcome.

3) Sample size. Sample size was calculated on 20% efficacy in lifetime testing. Where did this 20% assumption come from? also what's justification of using coefficient of variation of 0.25 (assuming it is the intracluster correlation?)? This 0.25 seems too big and basically the sample size could not be reproduced. Detailed and clearer sample size calculation is needed with justifications of all the assumptions.

4) Statistical analysis. It said "Risk/mean differences and risk ratios were calculated from cluster-level summaries and compared between arms with t-test". This is not correct as t-test is for continuous variables. For categorical variables, need some other tests. It said "effect estimates were adjusted for sex and age group a priori and any imbalance between arms in adolescent covariates". However, can authors make it clear what exactly the stats models used for the adjusted analyses? What are the outcomes? What are the covariates?

5) Study design. Although the lifetime testing for adolescents is the primary outcome, it seems the investigators were more interested in households and adult residents in each cluster when designing the trial and randomising the clusters. It's a bit confusing what's rationale of the study design. Household first or adolescents first? It looks like the sample size calculation and identification of primary outcome is a post hoc decision as there are so many focus on secondary analyses which are not even powered. As we can see, the table 1 should be on the comparison of adolescents in two arms but it becomes about households? The current table 1 can be removed to supplementary information.

6) Randomisation. It only said "restricted randomisation" but exactly by what method? stratification, minimisation or blocking? Details are needed. Also there are professional software and CTU service around for the randomisation but two study PIs did this in a way that is not really understandable by professional statisticians. It should be simpler and more straightforward than what's written in the paper.

***

Any attachments provided with reviews can be seen via the following link:

[LINK]

20 Oct 2020

Submitted filename: CL trial - Response to reviewers V1.2.docx

Click here for additional data file.

13 Nov 2020

Dear Dr. Indravudh,

Thank you very much for submitting your revised manuscript "Effect of community-led delivery of HIV self-testing on HIV testing and antiretroviral therapy initiation in Malawi: a cluster-randomised trial" (PMEDICINE-D-20-03726R2) for consideration at PLOS Medicine.

Your paper was again evaluated by our independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

We will not yet be able to accept the manuscript for publication in the journal in its current form, but we would like to invite you to submit a further revised version that addresses the reviewers' and editors' comments fully. You will note that we cannot make a decision about publication until we have seen the revised manuscript and your response, and we may seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We hope to receive your revised manuscript by Dec 04 2020 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. For instructions see http://journals.plos.org/plosmedicine/s/submission-guidelines#loc-methods.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

Please let me know if you have any questions. Otherwise, we look forward to receiving your revised manuscript soon.

Sincerely,

Richard Turner, PhD

Senior Editor, PLOS Medicine

rturner@plos.org

-----------------------------------------------------------

Requests from the editors:

Please finalize the arrangements for data release.

We suggest amending the text at line 34 to "... HIV infection ... in key groups including ...".

Please revisit the cumulative incidence of ART initiation, risk difference "73.2%" quoted at line 68. This appears to be quoted as "72.7%" in table 2.

Also at line 68, please adapt the wording to make it clear what "Differences were observed ..." refers to, perhaps by linking this to the previous sentence. Please adapt the wording if this is a post-hoc analysis. Again, please make sure that the numbers quoted can be reconciled with those at line 104 and in table 2.

At line 104, we ask you to adapt the wording to "apparently increased" or similar, as we believe you are quoting a secondary outcome finding where the risk of type 1 error is elevated. Please make a similar change at line 462.

At line 547, we suggest using the word "contamination" rather than "spillover".

Comments from the reviewers:

*** Reviewer #1:

Overall I am happy with the responses except for:

The number of clusters was small, so the potential for bias remains and this is worth discussing. Still, the participants were well balanced so the amount of bias is hopefully small. What is missing is the most important thing, however, which is lifetime risk of

testing before the intervention period. This should be in the discussion as well.

We have provided additional details on the sample size calculations (pg. 14), which are based on standard methods outlined by Hayes and Moulton for cluster-randomised trials [1]. Although the number of clusters was small, the trial was designed with sufficient power to

measure relevant differences between arms in the primary and secondary outcomes. Our study did not measure lifetime risk of testing through a baseline sample; we have expanded on this as a limitation in the discussion (pg. 26).

My point is not about power, my point is about bias. Small numbers of clusters means that the chance of lack of exchangeable populations increases and this is an important limitation of the design that needs to be discussed.

*** Reviewer #2:

Thank you for the opportunity re-review this paper looking at the impact of community-led HIVST on case identification and initiation on treatment. This is an important and relevant study that shows the effectiveness of HIVST in improving case identification, treatment initiation, viral suppression, and eventually reducing incidence. The author has addressed all of my and my fellow reviewer's comments in this paper with sufficient justification, and I therefore recommend this paper be accepted for publication.

*** Reviewer #3:

Many thanks authors for their great effort to improve the manuscript. The authors have addressed most of my concerns and comments adequately. However, there is still one remaining issue needing attention. I am still not satisfied/convinced with the answer for using the lifetime HIV testing among adolescents (15-19 years) as primary outcome of the study. I can understand it from practical point of view but couldn't understand what's research hypothesis and potential causal relationship between the short-term intervention and lifetime HIV testing, direct or indirect? any evidence (similar studies) from the existing literature? The secondary outcomes are more straightforward and making sense so it's fine.

***

Any attachments provided with reviews can be seen via the following link:

[LINK]

4 Dec 2020

Submitted filename: STAR_CL trial_Response to reviewers_v1.3.docx

Click here for additional data file.

17 Dec 2020

Dear Dr. Indravudh,

Thank you very much for re-submitting your manuscript "Effect of community-led delivery of HIV self-testing on HIV testing and antiretroviral therapy initiation in Malawi: a cluster-randomised trial" (PMEDICINE-D-20-03726R3) for consideration at PLOS Medicine.

I have discussed the paper with editorial colleagues and our academic editor, and it was also seen again by one reviewer. I am pleased to tell you that, provided the remaining editorial and production issues are fully dealt with, we expect to be able to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We hope to receive your revised manuscript after the holidays. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

Please let me know if you have any questions. Otherwise, we look forward to receiving the revised manuscript soon.

Sincerely,

Richard Turner, PhD

Senior Editor, PLOS Medicine

rturner@plos.org

------------------------------------------------------------

Requests from Editors:

Please finalize the arrangements for study data deposition.

At line 64, please amend the text to "the proportions ... were ...".

Please revisit the sentence at lines 69-70, where it seems that a few additional words would be helpful to indicate that the difference is in favour of the HIVST group.

At line 71, please make that "few adverse events".

We note that reference 16 contains some superfluous information about competing interests - please read through the reference list to ensure that this and any other material not needed is removed.

Please move table S1 to the main paper and adjust the table numbering as appropriate.

Please break up the supplementary material document into individual files, so that specific items and files are referred to in the text (e.g., "See S2_Figure").

The attached published trial protocol can be removed and simply referred to as a reference.

Comments from Reviewers:

*** Reviewer #3:

The authors have addressed my question very well. I am satisfied with the response and revision. No further issues needing attention.

***

Any attachments provided with reviews can be seen via the following link:

[LINK]

15 Jan 2021

Submitted filename: STAR_CL trial_Response to reviewers_v1.4.docx

Click here for additional data file.

4 Apr 2021

Dear Dr Indravudh,

On behalf of my colleagues and our Academic Editor, Dr Fox, I am pleased to inform you that we have agreed to publish your manuscript "Effect of community-led delivery of HIV self-testing on HIV testing and antiretroviral therapy initiation in Malawi: a cluster-randomised trial" (PMEDICINE-D-20-03726R4) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

Prior to final acceptance, please add a few words around line 160 to note that the trial is reported according to CONSORT.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Richard Turner, PhD

Senior Editor, PLOS Medicine

rturner@plos.org