Fee for home delivery and monitoring of antiretroviral therapy for HIV infection compared to standard clinic-based services: a randomised trial in South Africa

SUMMARY

Background:

Home delivery and monitoring of antiretroviral therapy (ART) is convenient, overcomes logistic barriers, and could increase individual ART adherence and viral suppression. With client payment and sufficient health benefits, this strategy could be scalable. The aim of the Deliver Health Study was to test the acceptability and efficacy of a user fee for home ART monitoring and delivery.

Methods:

We conducted a randomized trial, the Deliver Health Study, of a fee for home delivery of ART compared to free clinic ART delivery in South Africa. People with HIV were randomized to: 1) fee for home delivery and monitoring of ART, including community ART initiation if needed; or 2) clinic-based ART (standard of care). The one-time fee for home delivery (ZAR 30, 60, and 90; equivalent to $2, 4, 6) was tiered based on participant income. The primary outcomes were recorded fee payment and acceptability assessed via questionnaire. The secondary outcome was viral suppression with the difference between study groups assessed through robust Poisson regression including participants with viral load measured at exit (modified intention-to-treat analysis (mITT)).

Findings:

From October 2019-January 2020,162 participants were enrolled; 82 were randomized to the fee for home delivery group and 80 to the clinic-based group with comparable characteristics at baseline. Overall, 87 participants (54%) were men, 101 (62%) were on ART, and 98 (60%) were unemployed. In the home delivery group, 40 (49%), 33 (40%), and 9 (11%) participants qualified for the ZAR 30, 60, and 90 fee, respectively. Median follow-up was 47 weeks (IQR 43–50 weeks) with 96% retention. Eighty (98%) participants paid the user fee with high acceptability and willingness to pay. In the mITT analysis of 157 (96%) subjects who completed follow-up, fee for home delivery and monitoring statistically significantly increased viral suppression from 74% to 88% overall (RR=1.21, 95% CI: 1.02–1.42); and from 64% to 84% among men (RR=1.31, 95% CI: 1.01–1.71).

Interpretation:

Among South African adults with HIV, a fee for home delivery and monitoring of ART significantly increased viral suppression compared to clinic-based ART. Clients’ paying a fee for home delivery and monitoring of ART was highly acceptable in the context of low income and high unemployment, and improved health outcomes as a result. Home ART delivery and monitoring, potentially with a user fee to offset costs, should be evaluated as a differentiated service delivery strategy to increase access to care.

Funding:

National Institutes of Mental Health (NIMH)

Trial registration:

ClinicalTrials.gov number NCT04027153

Introduction

In KwaZulu-Natal (KZN), South Africa (SA), after an abrupt decline in the 1990s, life expectancy rebounded starting in 2004 with the introduction of antiretroviral therapy (ART).^1,2 This more than 10-year increase in lifespan was driven primarily by substantial access to effective, safe, well-tolerated, once-daily, oral ART for treatment and prevention of HIV.¹ Effective and efficient ART delivery is key for increasing population coverage and maintaining suppression for life, the key indicator of treatment success. However, in South Africa, of the 7.8 million persons with HIV, 72% are on ART, and only 66% are virally suppressed.³ Men are less likely to be suppressed than women (58% compared to 72%, respectively),³ due to masculine gender norms and barriers to care including clinic bottle-necks in ART provision; men more often report the time taken to access clinic-based care is a barrier.^4,5 Further, there are limited options for collection of medication outside the clinic setting in rural areas.⁶ HIV-associated mortality continues to be high, particularly among priority populations who do not regularly visit the clinic on a sufficiently frequent schedule to maintain uninterrupted ART and viral suppression.^7,8 Over four years, a third of persons living with HIV in SA are no longer retained in care.⁹ Clinically stable persons (i.e. without advanced HIV) can safely and effectively receive ART and monitoring in the community.^10,11 In South Africa, medication delivery outside the clinic is available through the Centralised Chronic Medicines Dispensing and Distribution (CCMDD), supported by Project Last Mile (PLM), which packages and delivers medication to external pick-up points across the country and to clinic-based pick-up points; services outside the clinic are primarily available in urban areas.¹² Home-delivery could offer an alternative to CCMDD, particularly in rural settings and incorporating clinical screening could reduce clinic visits. Gaps in data exist on the efficacy of home ART delivery and acceptability of a fee for home delivery service in a rural setting. A fee for home delivery has the potential to offset programmatic staffing and transportation costs. Innovations in efficient service delivery—including expanding community ART initiation, monitoring, and refill delivery—are needed for people living with HIV, particularly men among whom the largest gap in viral suppression is seen, to achieve the UNAIDS 95-95-95 goal leading to 85% viral suppression among all persons living with HIV.^3,13

Community-based ART services – i.e., ART initiation and refills outside the clinic – increase viral suppression by removing logistical barriers to clinic access and engaging persons living with HIV in care.^10,14 However, gaps in service provision and access persist. Routing science, i.e. the scheduling of the route or itinerary of people or goods, uses data-driven algorithms, e.g., the traveling salesman algorithm, to match client delivery preferences and identify efficient delivery routes to ensure on-time delivery of the right goods to the right person. Leveraging the experience and expertise of routing science, home ART delivery could address gaps in access, meet client preferences, improve individual health, minimize carbon emissions through combined deliveries, and increase viral suppression over time. Further, since minimal financial resources are available (the South African National Department of Health pays a relatively small fee per refill distributed) delivery optimization, a standard component of routing algorithms, is essential. Few data from randomised trials directly compare the efficacy of data-driven home ART delivery and monitoring to clinic services.

We conducted a randomised trial to evaluate the efficacy of client payment of a fee for community ART initiation as indicated, monitoring, and ART resupply, compared to standard clinic ART initiation as indicated, monitoring, and ART resupply for adults in HIV in KwaZulu-Natal, South Africa. The objectives were to evaluate client payment of the fee for home delivery; acceptability of the delivery service; and the relative efficacy of a client fee for home delivery versus clinic-based ART on the proportion of persons with HIV who were virally suppressed at one year.

Methods

Study design

We conducted an unblinded, individually-randomised trial of home ART delivery (including ART initiation if the participant had not yet initiated ART), monitoring, and ART resupply compared to standard clinic ART services among South African adults living with HIV. We hypothesized that a data-driven ART delivery algorithm would match efficient supply with client preferences, clients would pay a fee for the service, and home delivery would overcome logistic barriers and increase viral suppression. Further, we hypothesized that paying a fee for delivery and requesting a specific time for ART delivery and monitoring would increase engagement in care and implementation intention (i.e., a strategy that automates action control, in this case receiving medication delivery which could increase adherence¹⁵). Finally, we hypothesized that home ART delivery and monitoring would increase viral suppression, especially among men who have a lower rate of viral suppression in part due to work and other opportunity costs,¹⁶ and address disparities in viral suppression by gender.

The study was conducted in rural and peri-urban areas of high HIV prevalence in South Africa; Pata, Azalea and Dambuza communities in uMgungundlovu District, KwaZulu-Natal, South Africa. Population HIV prevalence in KwaZulu-Natal was 36%¹⁷ representing high prevalence settings in southern Africa. Public clinics in South Africa offer access to ART at no cost. These communities are characterized by high unemployment, low per capita income (below USD $2 per day), and income inequality.

The Human Sciences Research Council Research Ethics Committee, in South Africa, and the University of Washington Institutional Review Board, Seattle, WA, approved this study. The study was registered at ClinicalTrials.gov, number NCT04027153.

Study participants

Following community mobilization which included discussing the proposed study with the Community Advisory Board, community stakeholders, and the local Department of Health, participants were recruited through HIV clinics and HIV testing at community locations. At community-based HIV testing and counselling service points in high traffic areas, such as transportation depots, and at clinics, information about the study was provided to potential participants.

Trained nurses and supervised lay counsellors conducted study activities. Staff received standardized national nurse-initiated and managed ART (NIMART) training in nurse-led HIV testing and counselling, clinical evaluation for ART initiation, ART initiation, monitoring, adverse effects, and national algorithms for HIV care. A nurse was responsible for clinical oversight, prescriptions, and blood draws. Participants completed an interview during which staff electronically collected preferences for delivery, demographics, and HIV exposure via the REDCap Mobile App.¹⁸ Lay counsellors conducted HIV testing with standardized pre- and post-test counselling. Persons living with HIV received additional point-of-care testing to stage their HIV and assess clinical eligibility for community-based ART initiation: CD4 cell count, the World Health Organization (WHO) clinical HIV stage, pregnancy testing, creatinine testing to assess renal function, and symptom screening for tuberculosis. A dried blood spot was collected to assess HIV viral load at baseline.

Persons living with HIV were eligible for randomisation if they were able to provide informed consent, 18 years or older, a resident in the participating communities, clinically stable (CD4 count >100 cells/μL, WHO HIV stage 1–3, not pregnant, normal renal function, and had no symptoms on a standardized symptom screen for active tuberculosis), and willing to pay for home delivery of ART. Clinic records were reviewed to facilitate transfer of participants on ART to the study. ART was initiated at enrolment for participants who reported not being on ART and for whom no clinic records were found; a clinic file was started for those participants. As the participants were clinic patients, their medications were available from a central pharmacy and were packaged into three months supply for the purposes of the study. Participants who were not eligible for randomisation for clinical reasons or pregnancy were referred to care and followed until they linked to clinic-based services. We chose a CD4 threshold of >100 cells/μL to facilitate clinic care for persons with advanced HIV/AIDS who were at risk for opportunistic infections. Prior to enrolment, all participants provided written informed consent, which included counselling about randomisation, procedures in each study group, and their rights as research participants.

Randomization

The study analyst generated the randomisation allocation using randomly selected blocks of size 2, 4, or 6. The randomisation allocation was performed using codes in sequentially numbered envelopes. Persons living with HIV were randomised in a 1:1 ratio to the 1) client payment of a fee for community ART initiation as indicated, monitoring, and ART resupply, or 2) standard clinic ART initiation as indicated, monitoring, and ART resupply group (Figure 1). Due to the infeasibility of blinding the study team and study participants to ART delivery method, the study was unblinded; however, the laboratory staff, who assessed the outcome of plasma HIV viral load, were blinded to the allocation of participants as were the study investigators.

Figure 1: Trial profile.

Study procedures

Participants reported their monthly income, which was used to determine the cost of ART delivery in the home ART delivery group. For income <ZAR 500, ZAR 500–3200, and >ZAR 3200 the cost of delivery was ZAR 30, 60, and 90 (equivalent to USD 2, 4, 6), respectively, with a one-time fee covering the cost of delivery for the duration of the study analogous to an annual fee. The fee was paid in cash to the study team who recorded payment and monitored payment receipt. Participants also completed a delivery preferences survey indicating suitable delivery times, confirming location at home or work, and updating contact details. Participants received same-day ART initiation if not already on ART including standardized counselling and the national HIV program’s first line ART regimen at that time of efavirenz, tenofovir, and emtricitabine. Seven days after ART initiation, participants received a phone call to ask about symptoms, ART side effects, and adverse events. Using the preferred delivery times and locations provided by participants, a custom scheduling algorithm solved the “traveling salesman problem” (i.e. mathematical optimization function) to optimize the timing and order for each week’s deliveries, minimizing the total distance travelled while matching client availability, and ensuring that clients had an uninterrupted supply of ART. The final driving route was determined using a commercially available route planning smartphone app. Deliveries took place two to three weeks prior to participants exhausting their ART supply; the algorithm accounted for remaining ART and also included an option for urgent deliveries to avoid participants running out of medication. The algorithm also accounted for the average drive time at that time of day and the typical duration of the home monitoring and delivery visits. Participants received a text message to confirm the date and time of their delivery and could reschedule the visit by text message, request a vacation supply, and nominate someone else to collect their medication by contacting the study staff. Missed deliveries, if participants were not at home to receive the delivery, were added to the following week’s delivery algorithm. The study team members received a weekly delivery memo including the order of participants receiving ART on each day and the driving directions.

Following enrolment, in the home-delivery group, participants received month one and then quarterly home visits for ART resupply, clinical monitoring, counselling, and ascertainment of adverse events and social harms. ART was dispensed with one month supply, two months, and then every three months thereafter. Trimethoprim-sulfamethoxazole prophylaxis was dispensed according to country guidelines and tuberculosis preventive therapy (isoniazid) was provided. Participants received appointments for their home visits. Participants who missed visits were contacted and their visit rescheduled. The home delivery service was regularly available on evenings and on the weekends. Staff used a phone-based application to conduct standardised monitoring that included counselling guidelines. For HIV and ART monitoring, participants completed a clinical questionnaire to screen for symptoms of side effects associated with ART, tuberculosis, and other common opportunistic infections. Point-of-care creatinine testing was done to monitor renal function. HIV plasma viral load was assessed at exit for treatment success and the results provided to participants to guide adherence counselling. Participants who required additional clinical services were referred for care and followed until they linked. Participants in the home delivery group were administratively linked to the clinic and their files kept up-to-date.

Participants in the clinic group were referred to established local ART clinics for ART initiation (if required), monitoring, and refills. Participants in the clinic group attended the clinic of their choice from the eleven available. They received quarterly phone calls to document ART initiation and adverse events.

Social harms and adverse events were assessed at every in-person visit and with every phone call. Participants were asked about adverse events, including serious adverse events; we included hospitalisations related to HIV and ART among serious adverse events. Chart abstraction was planned for all participants to capture additional clinical events and test results.

At the exit visit, planned for month six, all home ART delivery group procedures were conducted as outlined above, with the addition of collection of plasma for measurement of HIV viral load. In addition, all participants completed a questionnaire regarding their experience in accessing care, acceptability of home ART delivery, and barriers for not visiting the clinic in the clinic group. Participants receiving home ART delivery were then transferred to the clinic or differentiated service delivery, e.g. decentralized medication dispensing, as appropriate.

The trial was fully enrolled in January 2020, prior to the COVID-19 pandemic. During the course of the pandemic, the local Department of Health requested that the study continue to limit the number of participants waiting at the clinic to collect refills as a mitigation strategy for COVID-19 transmission. Deliveries halted for two weeks in March 2020 (Figure 2) and then home delivery continued as no-contact deliveries with clinical assessments via telemedicine from March 2020 until guidance from the Department of Health allowed in person visits for research to restart in September 2020, which increased follow-up time from the planned six months to 12 months for both study groups. In the clinic group, the Department of Health increased access to fast track ART collection as a COVID-19 transmission mitigation strategy. Participants in both groups received a standard three-month resupply of ART per visit during the COVID-19 pandemic.

Figure 2: Transport Log compared to ART dispensation.

*Transport logs were not collected before 2021-01-04

Outcomes and assessment

The primary trial endpoints were the proportion of participants paying the delivery fee and the acceptability of home delivery. The secondary endpoints were achieving HIV viral suppression (<20 copies/mL) assessed at month 12 among all participants and among men; plasma viral load testing was conducted by an accredited laboratory. Safety, clinical adverse events and social harm, was a secondary outcome and was assessed through adverse event reporting.

Laboratory methods

Rapid HIV testing was conducted according to national guidelines: Determine HIV 1–2 (Abbott) and First Response HIV 1-2-0 Card Test (Prima Medical) with SD BIOLINE HIV-1/2 Rapid (Standard Diagnostics Inc.) as a tie-breaker. Creatinine testing was done using point-of-care StatSensor Xpress (Nova Biomedical). Point-of-care CD4 testing (Pima, Alere) was conducted using a finger-stick specimen. Plasma HIV viral load was assessed by bDNA (bioMérieux, with a limit of detection of 20 copies/mL) at Global Laboratories in Durban, South Africa.

Statistical analysis

For the primary outcome, we reported the proportion paying the fee, and we did not conduct a power calculation. To estimate power for the secondary outcome of viral suppression, we assumed a total sample size of 120 participants, randomized 1:1. We assumed that viral suppression among participants in the standard of care group would be 75%. With an estimated 5% loss to follow-up and 57 participants retained per group, the power was >80% to see a 20% difference (75% vs. 95%) in viral suppression. The sample size calculation did not account for correlation of viral suppression within the clinic; the effect of within clinic correlation was explored in sensitivity and exploratory analyses.

All available exit assessments from participants contributed to assignment of whether the participant was virally suppressed. Acceptability was assessed through an exit questionnaire that reviewed willingness to continue to pay a fee for home ART delivery and monitoring,

We calculated the proportion of participants who paid the fee, who answered affirmatively to the acceptability questions (Table S1), and the rates of viral suppression (95% CIs were calculated using the Wilson score method). The endpoint of viral suppression was assessed among participants who had viral load assessed at exit (the modified intent-to-treat [mITT] cohort). Effects of the randomisation groups on viral suppression were estimated as relative risks (RRs) and risk differences (RDs) produced with robust Poisson regression.¹⁹ Models were adjusted a priori for gender and age younger than 30, which are known covariates of viral suppression. Tests for superiority of the intervention group compared to the clinic group were based on two-sided Wald p-values <0·05. This regression analysis was repeated separately for men. To account for correlation within the clinic and cluster variation between clinics, in a sensitivity analysis, we estimated the RRs and RDs using generalized estimating equations with exchangeable correlation structure. In an exploratory analysis, we also evaluated using an independence correlation structure.

Delivery vehicles were equipped with GPS devices to monitor the distance driven and logs were kept separately to record stops and medication dispensation. These data were evaluated to estimate the total distance driven, miles driven per successful ART dispensation and monitoring visit, time taken per visit, and to calculate the carbon emissions and equivalent carbon offset from driving a delivery vehicle with a combustion engine. We did all analyses with R version 4.1.

Role of the funding source

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

Results

Between October 7^th, 2019, and January 30^th, 2020, 400 participants were identified for study screening; 180 participants were identified as living with HIV, of which 173 (96%) completed screening for study eligibility (Figure 1). Of the 11 who were ineligible for randomisation, 6 (55%) had a CD4 count of <100 cells/μL, 3 (27%) screened positive using the symptomatic TB screening questionnaire, 1 (9%) screened positive in the clinical assessment for WHO stage 4 disease, and 1 (9%) was pregnant. A total of 162 participants were randomised: 80 to the clinic group, and 82 to the home ART group. Seven participants were lost to follow-up: 6 in the clinic group, and 1 in the home ART group; of the 7, 3 withdrew, 2 moved out of the study area, and 2 were lost for an unknown reason. At least 96% of participants completed each visit (months one, three, and six visits; and a month nine visit that was added to mitigate the COVID-19 movement restrictions until exit visits could be conducted in person) in the home ART delivery group. Data for 96% of participants (155/162) were included for the fee payment and viral load endpoints analyses (modified intention-to-treat analysis).

The baseline characteristics for the 162 participants are shown in Table 1, of whom 87 (54%) were men and all of whom were Black race and non-Hispanic. Participants had a median age of 36 years (interquartile range (IQR): 31–43). Seventy-four percent (120/162) of participants completed secondary education and 60% (98/162) reported that they were not employed. Fifty four percent (87/162) of participants qualified for the ZAR 30 fee tier, 38% (62/162) qualified for the ZAR 60 fee, and 8% (13/1622) for the ZAR 90 fee. Most participants were asymptomatic and clinically stable: 96% (156/162) were WHO clinical stage I, 81% (132/162) had a CD4 count ≥350 cells/μL, and 85% of participants initiating ART (52/61) had normal renal function (creatinine <106 μmol/L).

Table 1:

Baseline characteristics by randomisation group

		Total		Clinic Group		Fee for Delivery Group
		(n = 162)		n = (80)		n = (82)
Gender	Women	75	(46%)	37	(46%)	38	(46%)
	Men	87	(54%)	43	(54%)	44	(54%)
Age, years	18 – 29	35	(22%)	23	(29%)	12	(15%)
	30 – 49	108	(67%)	49	(61%)	59	(72%)
	≥ 50	19	(12%)	8	(10%)	11	(13%)
Education	Primary	42	(26%)	26	(32%)	16	(20%)
	Secondary	102	(63%)	44	(55%)	58	(71%)
	Tertiary and above	18	(11%)	10	(12%)	8	(10%)
Employed		64	(40%)	31	(39%)	33	(40%)
Knowledge of HIV status	Persons known to be living with HIV	107	(66%)	53	(66%)	54	(66%)
	Persons newly identified as living with HIV	55	(34%)	27	(34%)	28	(34%)
Past ART use (among persons known to be living with HIV)	Currently on ART	101/107	(94%)	50/53	(94%)	51/54	(94%)
	Taken ART in the past	3/107	(3%)	2/53	(4%)	1/54	(2%)
	Never taken ART	3/107	(3%)	1/53	(2%)	2/54	(4%)
Fee tier	30 ZAR	87	(54%)	47	(59%)	40	(49%)
	60 ZAR	62	(38%)	29	(36%)	33	(40%)
	90 ZAR	13	(8%)	4	(5%)	9	(11%)

^*Fee tier for the Clinic Group is the tier in which they would have been placed, based on their monthly income, had they been randomized to the Fee for Delivery Group

Median follow-up was 47 weeks (IQR 43–50 weeks). In the fee payment group, 98% (80/82) of participants paid the full user fee; 100% (9/9) in the ZAR 90 group, 94% (31/33) in the ZAR 60 group, and 100% (40/40) in the ZAR 30 group. Acceptability was high with 100% (81/81) reporting willingness to continue to pay a fee, reporting that the fee was reasonable (81/81), and that they would recommend participation to others (81/81) (Table 2). One hundred percent (79/79) of participants who paid the fee also reported that the fee helped them remember to take their medication.

Table 2:

Rates of fee payment by fee amount

Fee amount	n	%	(95% CI)
Total	80/82	98%	(92–99)
R30	40/40	100%	(91–100)
R60	31/33	94%	(80–98)
R90	9/9	100%	(70–100)

Overall home ART delivery and monitoring increased viral suppression at 47 weeks compared to the clinic group (88% vs. 74%, RR=1·21, 95% CI: 1·02–1·42) (Table 3). Viral suppression was high for men (84%) and women (92%) in the home ART delivery group, compared to 64% for men and 86% for women in the clinic group. The home ART delivery strategy significantly increased viral suppression among men: community-based ART (84%, RR=1·31, 95% CI: 1·01–1·71), compared to standard of care (64%). Rates of viral suppression were comparable in the two groups among women; we were not powered to see a difference in viral suppression among women. The absolute increase in viral suppression, the risk difference, was 21% (95% CI 2 – 39) for men in the home ART group compared to the clinic (Table 3). When accounting for clustering among participants at clinics, the results were consistent with the primary findings (overall RR=1·16, 95% CI: 1·06–1·27 and among men RR=1·22, 95% CI: 1·03–1·45) (Table S2–S4). The intracluster correlation coefficient (ICC) was 0.025 overall and −0.005 among men. No serious adverse events were reported and no social harms related to study participation were reported.

Table 3:

Rates, relative risks, a of viral suppression, overall and by gender

	Rate of viral suppression				Adjusted* RR of viral suppression			Adjusted* RD of viral suppression
	Clinic group (n = 80)		Fee for delivery group (n = 82)		Fee for delivery group versus clinic group			Fee for delivery group versus clinic group
	n/N	(%)	n/N	(%)	RR	(95% CI)	p value	RD	(95% CI)	p value
Overall	55/74	(74%)	71/81	(88%)	1.21	(1.02–1.42)	0.026	16%	(3–29)	0.016
Men	25/39	(64%)	37/44	(84%)	1.31	(1.01–1.71)	0.045	21%	(2–39)	0.030
Women	30/35	(86%)	34/37	(92%)	1.08	(0.88–1.33)	0.449	7%	(−10–25)	0.420

RR = Relative risk; RD = Risk difference

^*Adjusted for gender and age younger than 30 years

From January 2020 onwards, in the home ART delivery and monitoring group, medication dispensation logs compared well with transportation logs visually indicating good correlation between travel and successful ART delivery and monitoring (Figure 2). Four hundred and twenty six successful medication dispensations were conducted. Extrapolating data from 289 dispensations with routing logs, a distance of 3956 km (2458 miles) was driven which was 18 km (11 miles) per successful stop, with it taking an average of 35 minutes per dispensation including driving, monitoring, ART dispensing, multiple dispensations at a single stop, and unsuccessful stops. To offset the annual carbon footprint for ART delivery and home monitoring, one tree would need to be planted for every 10 participants in the program.²⁰

Discussion

This randomised trial conducted in a high HIV prevalence setting in South Africa during the global COVID-19 pandemic and consequent movement restrictions provides evidence that home delivery of ART, including same-day ART initiation as needed, monitoring and ART resupply, increases viral suppression among people with HIV, particularly among men, compared to standard clinic-based services. The adaptation to no-contact delivery and clinical assessments via telemedicine also demonstrate the resilience of home delivery to COVID-19 movement restrictions. Further, the use of routing algorithms that prioritize client preferences and optimize delivery logistics can be leveraged to automate care provision and also meet the requirements of clinical protocols and constraints of medication supply. Client-centred services that overcome barriers to care may increase the proportion of persons living with HIV who start ART, achieve viral suppression, and sustain engagement in care.

The COVID-19 pandemic has accelerated the adoption of differentiated service delivery (DSD) services for HIV care, such as multi-month dispensing, fast-track ART,¹¹ and additional pick-up points for medications with more than 800 points available in KwaZulu-Natal. DSD has helped maintain rates of viral suppression but new HIV diagnoses were delayed.^21,22 However, current viral suppression levels falls short of the UNAIDS goal of overall 86%; interventions are needed to identify and link persons with HIV to ART and sustained viral suppression.²³ Further, a gap in evidence exists regarding how to achieve high levels of viral suppression including among men for their own health benefits and to lower HIV incidence among women.^24,25 Using standard vehicle routing algorithms that match client preferences for ART delivery with existing ART supply, we demonstrated high levels of viral suppression that met the UNAIDS goals and eliminate gender disparities. Finally, even though 60% of participants reported being unemployed, payment of the delivery fee was high (98%) indicating the prioritization of accessible services within the context of constrained client resources.

We hypothesized that home HIV delivery and monitoring would increase the proportion of persons with HIV who achieve viral suppression, based on evidence that community-based HIV interventions increased suppression,²⁶ by overcoming barriers to care including standard work hours, stigma, unfavourable perceptions of clinics and staff, cost of transport, and lost wages, and, more recently, the risk of COVID-19 acquisition.^8,27,28 Further, persons in informal employment may need to choose between visiting the clinic or earning daily wages,⁵ limiting their use of clinic-based services.^5,29 The use of delivery algorithms that account for client preferences increases flexibility and access for clients. We also hypothesized that payment of the fee by the client would increase treatment intention and adherence, which participants reported in their survey; however, this comparison was not randomized. To assess this, home ART delivery service would need to be tested with and without a fee to determine if the fee payment by the client increased their adherence. Offering convenient home ART delivery and monitoring including times after hours and on weekends, being flexible to meet travel/mobility needs, offering quarterly refills, and streamlining monitoring and resupply had better viral suppression outcomes, especially for men. Real-world equivalent services for home-delivery include private pharmacies that deliver medications, delivery services which use pre-packaged items, and delivery algorithms which are available through standard computing tools. These could be adapted for implementation of home ART delivery and monitoring. While the study collected plasma for viral load testing for the study, dried blood spots (DBS) perform equally well for measuring viral load and could be incorporated into a home ART delivery service with collection of the DBS for viral load testing.³⁰

There were no serious and severe adverse events across the two study groups, indicating that home ART delivery is likely to be as safe as clinic-based ART. However, the proportion of persons achieving viral suppression was higher through home ART delivery which could increase health gains over time. Lastly, for scale-up, implementation may need a mix of behavioural and social interventions to see the same (or larger) effect size across heterogeneous settings.

We acknowledge several limitations of the study. The standard of clinic care changed during the COVID-19 pandemic to increase access to fast-track ART and multi-month refills;¹¹ both are evidence-based DSD strategies to maintain viral suppression. More than three quarters of participants achieved viral suppression at the clinic and it is likely the changes to the standard of care at the clinic improved viral suppression in the clinic-based group, although the COVID-19 pandemic may have also limited clinic access. Movement restrictions during COVID-19 may have increased the impact of a home ART delivery and monitoring service since participant were more likely to be at home. However, while clinic-based DSD services increased viral suppression overall, disparities in viral suppression by gender persisted in clinic-based services which were eliminated in the home ART delivery group, which had overall significantly higher viral suppression. The client fee for delivery of ART was relatively low (ZAR 30, 60 or 90); although it is higher than the ZAR 10 the Department of Health pays for CCMDD ART delivery to external pick-up points such as pharmacies. The cost of fuel per successful dispensation in South Africa is approximately ZAR 25 (for driving 14.2 km); thus, the fee could offset some of the costs of delivery. A formal costing and cost-effectiveness analysis have not been conducted to date. Our study was limited to settings with medium and high HIV prevalence and may not be generalizable to settings with lower prevalence, as sufficient number of clients within a geographic radius is required for home ART delivery and monitoring to be cost-efficient. The study was limited to adults since the ability to pay a fee was a requirement for enrolment. However, adolescents and children are a priority group and should be included in future evaluations including home ART delivery paid for by the health care service. We did not include persons lost-to-follow-up in the analysis; however, if we assumed that they were not virally suppressed that would have strengthened the study outcome. Lastly, a relatively small number of participants were ART naïve at enrolment and further studies should evaluate efficacy and safety among persons initiating ART for the first time who may require additional counseling.

The strengths of the study include the randomised design, the use of a routing algorithm to meet client delivery preferences, successful enrolment of men (54% of the study population), and high retention across the randomised groups. The study primary aims were fee payment and acceptability so that the viral suppression outcome could be considered in the context of client fee payment, acceptability, and willingness to continue to pay a fee for service. Using a mobile app, standardized care was provided following clinical algorithms, thus limiting medical errors and facilitating task shifting.

The next steps for client-centred ART delivery and monitoring are to test a menu of strategies to match the services provided to client preferences and to allow for those to change over time and be responsive to external circumstances such as the COVID-19 pandemic. Also, evaluating the health impact and cost at scale in high HIV prevalence settings with a baseline gender gap in suppression would provide evidence for generalizability. While home ART achieved viral suppression among almost 90% of women and men living with HIV, additional services are needed to reach the remaining 10%. Specifically, expanded home delivery services to address barriers to care and long-acting injectable antiretroviral therapy may overcome remaining logistical and other barriers. Future research directions should focus on scalable, client-centred strategies to deliver ART.

Research in context.

Evidence before this study

Antiretroviral therapy (ART) for HIV has prevented almost 10 million premature HIV-associated deaths globally. ART is safe, well tolerated, generally taken once daily, and widely available at low or no cost in many settings worldwide. However, coverage of ART, as measured by the gold-standard of viral suppression among persons with HIV, has fallen short of the 86% suppression target set by Joint United Nations Programme on HIV/AIDS (UNAIDS), with suppression higher among women compared to men. We searched PubMed in January 2022 to identify studies evaluating community-based strategies to increase viral suppression with these keywords: HIV, viral load, and community-based. Prior to this study, Dave and colleagues conducted a systematic review and meta-analysis evaluating the effectiveness of community-based HIV initiatives. For the outcome of viral suppression, community health workers and peers increased the relative risk of viral suppression by 40% (pooled OR: 1·40, 95% CI: 1·06–1·86). In a randomised trial of community-based ART initiation, monitoring, and resupply (the DO ART Study) viral suppression significantly increased overall from 63% to 74%. However, a quarter of persons with HIV continued to have detectable viral loads. Client-centred, accessible services such as home ART delivery and monitoring have the potential to adapt to changing restrictions, overcome barriers, and increase viral suppression.

Added value of this study

This study presents new evidence from a randomised clinical trial in South Africa on the effectiveness of home-based ART delivery and monitoring to achieve higher viral suppression, particularly among men, and meet the UNAIDS goal for 86% viral suppression. Importantly, despite high unemployment, 98% of participants paid a fee that could offset some of the delivery costs and serve as an incentive for daily adherence with high reported satisfaction with the service and willingness to pay for home ART delivery. The automated delivery algorithm increased flexibility of the service by accounting for client preferences for delivery times, remaining medication supply, and need for clinical monitoring.

Implications

We demonstrated that home ART delivery and monitoring is an effective strategy and could be scaled up to address the gap in viral suppression overall and for men. While this client-centred approach will require adaptation of current services including expanding to new delivery platforms, cost may not be a limiting factor.

Data sharing

A complete de-identified patient dataset sufficient to reproduce the study findings will be made available approximately 1 year after completion of the trial (NCT02929992), following approval of a concept sheet summarising the analyses to be done. Further inquiries can be directed to the Deliver Health Scientific Committee at lnakatsuka@partners.org.