Differentiated antiretroviral therapy (ART) delivery models, in which patients are provided with care relevant to their current status (e.g., newly initiating, stable on treatment, or unstable on treatment) has become an essential part of patient-centered health systems. In 2015, the South African government implemented Chronic Disease Adherence Guidelines (AGLs), which involved five interventions: Fast Track Initiation Counseling for newly initiating patients, Enhanced Adherence Counseling for patients with an unsuppressed viral load, Early Tracing of patients who miss visits, and Adherence Clubs (ACs) and Decentralized Medication Delivery (DMD) for stable patients. We evaluated two of these interventions in 24 South African facilities: ACs, in which patients meet in groups outside usual clinic procedures and receive medication; and DMD, in which patients pick up their medication outside usual pharmacy queues.
Methods and findings
We compared those participating in ACs or receiving DMD at intervention sites to those eligible for ACs or DMD at control sites. Outcomes were retention and sustained viral suppression (<400 copies/mL) 12 months after AC or DMD enrollment (or comparable time for controls). 12 facilities were randomly allocated to intervention and 12 to control arms in four provinces (Gauteng, North West, Limpopo, and KwaZulu Natal). We calculated adjusted risk differences (aRDs) with cluster adjustment using generalized estimating equations (GEEs) using difference in differences (DiD) with patients eligible for ACs/DMD prior to implementation (Jan 1, 2015) for comparison. For DMD, randomization was not preserved, and the analysis was treated as observational. For ACs, 275 intervention and 294 control patients were enrolled; 72% of patients were female, 61% were aged 30–49 years, and median CD4 count at ART initiation was 268 cells/μL. AC patients had higher 1-year retention (89.5% versus 81.6%, aRD: 8.3%; 95% CI: 1.1% to 15.6%) and comparable sustained 1-year viral suppression (<400 copies/mL any time ≤ 18 months) (80.0% versus 79.6%, aRD: 3.8%; 95% CI: −6.9% to 14.4%). Retention associations were apparently stronger for men than women (men RD: 13.1%, 95% CI: 0.3% to 23.5%; women RD: 6.0%, 95% CI: −0.9% to 12.9%). For DMD, 232 intervention and 346 control patients were enrolled; 71% of patients were female, 65% were aged 30–49 years, and median CD4 count at ART initiation was 270 cells/μL. DMD patients had apparently lower retention (81.5% versus 87.2%, aRD: −5.9%; 95% CI: −12.5% to 0.8%) and comparable viral suppression versus standard of care (77.2% versus 74.3%, aRD: −1.0%; 95% CI: −12.2% to 10.1%), though in both cases, our findings were imprecise. We also noted apparently increased viral suppression among men (RD: 11.1%; 95% CI: −3.4% to 25.5%). The main study limitations were missing data and lack of randomization in the DMD analysis.
In this study, we found comparable DMD outcomes versus standard of care at facilities, a benefit for retention of patients in care with ACs, and apparent benefits in terms of retention (for AC patients) and sustained viral suppression (for DMD patients) among men. This suggests the importance of alternative service delivery models for men and of community-based strategies to decongest primary healthcare facilities. Because these strategies also reduce patient inconvenience and decongest clinics, comparable outcomes are a potential success. The cost of all five AGL interventions and possible effects on reducing clinic congestion should be investigated.
Clinical Trial registration
Why was this study done?
- Antiretroviral therapy (ART) has brought enormous benefits, enabling HIV patients to live longer and healthier lives. However, many patients struggle to adhere to this lifelong treatment, and their retention in care is suboptimal.
- To address this, South Africa’s National Department of Health has introduced National Adherence Guidelines for Chronic Diseases that recommend key interventions suitable for new patients, clinically stable patients, patients with advanced HIV disease or on a failing ART regimen, and lost-to-follow-up patients.
- The early implementation of these guidelines from 2015 was evaluated to understand the effectiveness of five interventions and learn about improvements for the guidelines’ rollout.
- This paper provides the results on adherence clubs and decentralized medication delivery, two interventions for efficient medication pick-up by clinically stable patients.
What did the researchers do and find?
- We used a cluster-randomized evaluation design for adherence clubs and an observational study for decentralized medication delivery across 12 intervention and 12 control health facilities in 4 provinces.
- The 275 adherence club patients had improved retention in care at one year compared to the 294 controls, especially men, and comparable viral suppression within 18 months.
- The 232 patients receiving decentralized medication delivery had apparently lower retention in care than the 346 controls (possibly due to medication pick-up data not reaching clinics), and viral suppression was comparable overall, but men on decentralized medication delivery had apparently increased viral suppression.
What do these findings mean?
- This programmatic evaluation with a comparison group provides the National Department of Health with many lessons for the national rollout of the guidelines and the effect one can expect from such alternative medication pick-up services.
- The control facilities were already offering some adherence and retention services, and the observed benefits of the guideline interventions are therefore additional.
- The comparable outcomes for decentralized medication delivery against standard of care, the retention benefit by adherence clubs, and the apparent increased effects among men suggest that these interventions can play an important role in South Africa’s ART program.
- Because the interventions also decentralize ART and reduce patients’ inconvenience, comparable outcomes are a potential success, and further work should investigate the cost–benefit relation and possible effects on decongesting clinics.
Citation: Fox MP, Pascoe S, Huber AN, Murphy J, Phokojoe M, Gorgens M, et al. (2019) Adherence clubs and decentralized medication delivery to support patient retention and sustained viral suppression in care: Results from a cluster-randomized evaluation of differentiated ART delivery models in South Africa. PLoS Med 16(7): e1002874. https://doi.org/10.1371/journal.pmed.1002874
Academic Editor: Marie-Louise Newell, University of Southampton, UNITED KINGDOM
Received: January 16, 2019; Accepted: June 28, 2019; Published: July 23, 2019
Copyright: © 2019 Fox et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: In accordance with the World Bank’s Open Data and Data Privacy Policies and as per agreement with the Government of South Africa, the anonymized micro data set excluding individual patient viral load and CD4 count test results is available for public use in the World Bank’s Development Data Hub and in the World Bank’s Open Knowledge Repository at https://microdata.worldbank.org/index.php/catalog/3466.
Funding: This work was supported by World Bank trust funds from several governments to finance the impact evaluation data collection, cleaning and analysis components, and from the Government of South Africa domestic health financing for the implementation of the adherence interventions that were evaluated in this impact evaluation. The World Bank trust fund donors supporting this evaluation had no role in study design; data collection, analysis, or interpretation; or writing of the manuscript. The corresponding author has had full access to all the data in the study and had final responsibility for the decision to submit for publication. Part of AH’s time was supported by the National Institutes of Health Fogarty International Center (#D43TW009340).
Competing interests: SR is a member of the Editorial Board of PLOS Medicine. The authors have declared that no other competing interests exist.
Abbreviations: AC, Adherence Club; AGL, Adherence Guideline; aRD, adjusted risk difference; ART, antiretroviral therapy; CCMDD, Centralized Chronic Medicine Dispensing and Distribution; DiD, difference in differences; DMD, Decentralized Medication Delivery; GEE, generalized estimating equation; HREC, Human Research Ethics Committee; IRB, Institutional Review Board; NDOH, National Department of Health; NGO, nongovernmental organization; NHLS, National Health Laboratory Service; PHC, primary healthcare clinic; PuP, pick-up point; TB, tuberculosis
The benefits of the rollout of antiretroviral therapy (ART) programs in resource-limited settings have been massive, including increased survival [1,2], reduced morbidity [3–6], and potential reductions in transmission [7,8]. Now that HIV treatment programs have had time to mature, attention has turned to the challenges of keeping patients adherent to lifelong therapy. Ample evidence from studies that have reviewed evidence from all over sub-Saharan Africa has shown that retention in HIV care is suboptimal. Current evidence suggests that 5-year retention in sub-Saharan Africa is close to 60% [9–14].
One way to attempt to improve retention is by reducing the burden on the patient of seeking care. Differentiated ART delivery models, in which patients with different needs receive tailored care [15–18], has been proposed as a solution. Under differentiated ART delivery, patients who are clinically stable on treatment and have demonstrated good adherence can be offered a repeat prescription collection strategy that allows them to pick up their medication in a less time-consuming manner than general clinic pharmacy queues as part of a package of services, including counseling and peer support. Two such approaches are Adherence Clubs (ACs), in which patients meet in a small groups outside the usual clinic queues, pick up their prepacked medication, and discuss adherence, and Decentralized Medication Delivery (DMD), in which patients pick up their medication at a pick-up point (PuP) away from the clinic such as a private pharmacy or church. For both strategies, medication may be prepacked by the clinic, a government district pharmacist, or a private provider. Such approaches may provide an incentive for already clinically stable patients to continue to adhere because the time burden for medication collection is reduced. They may also reduce the burden on the clinic because these patients require fewer clinic visits. While the literature does show some benefit of repeat prescription strategies along with a package of services in differentiated ART delivery, studies have largely focused on the effectiveness of ACs, with most data coming from observational studies [19–22]. Further, most studies were evaluations of programs that did not mimic routine conditions but rather evaluated pilot programs with enhanced or nongovernmental organization (NGO) support. Thus, while evidence is beginning to emerge on the effectiveness of differentiated ART delivery models, to date, there is not enough evidence to conclude the approaches are effective.
South Africa, which has the largest HIV treatment program in the world , has been grappling with retention and poor adherence [24,25] and, in response, developed a comprehensive strategy to tackle these issues. The approach was developed in 2014 by the National Department of Health (NDOH) and is described in the “National Adherence Guidelines for Chronic Diseases”  (AGLs), which call for a “minimum package” of interventions to mitigate the problem by identifying patients who require more intense care (patients with unsuppressed viral loads and those lost from care) and those who require less intense care (patients who have demonstrated an ability to achieve and maintain viral suppression). The AGL minimum package of interventions contains eight interventions focused on education and counseling, repeat prescription strategies, patient tracing, and integrated care, detailed in a set of guidelines  and standard operating procedures .
In order to develop an evidence base behind differentiated ART delivery models and to inform improvement of the rollout of the interventions, South Africa’s NDOH piloted rollout of the interventions in 12 primary healthcare clinics (PHCs) from 2015. We evaluated five of the interventions and present here an evaluation of two strategies, ACs and DMD, for clinically stable patients designed to reduce patient burden, which would in turn help clinicians and staff to provide increased support to clients with advanced HIV disease and for those on a failing ART regimen. The evaluation design has been reported previously [28,29].
The study used an unblinded cluster-randomized evaluation design for ACs and an observational study for DMD. For DMD, while the study was initially designed to have sites randomized to DMD roll out, changes in clinic mandates meant that some clinics began using DMD in control sites and not all intervention sites were able to begin implementing DMD. The full protocol can be seen in S1 Text. The study was conducted in 24 health facilities (12 intervention, 12 control sites) in Gauteng, KwaZulu Natal, Limpopo, and North West provinces. Details on the sites and viral suppression by site is provided in S1 Table. Matched clinic pairs were randomized 1:1 (by computer) by the NDOH to intervention and control with a single randomization for all interventions. Intervention sites implemented the AGL interventions, and control sites continued to provide standard of care. Standard of care for clinically stable patients would include a visit usually every other month. Some counseling may have been given during visits, and some facilities may have offered support groups or group counseling sessions in the waiting areas. In addition, four control sites implemented AC-like interventions as part of standard of care. All 24 sites were PHCs that were 1) high volume (total on ART > 1,000 as reported by sites, which was not always accurate); 2) not a National Health Insurance pilot site; 3) generating computerized TIER.Net data; and 4) not participating in other adherence-related studies. Sites were matched on district, total on ART, proportion virally suppressed, setting (rural/urban/formal/informal), and location.
ACs comprise clinically stable ART patients who meet at facilities or community locations in groups of up to 30 every 2 to 3 months to receive group counseling, have a brief symptom screen, and receive prepacked medications. Clubs are managed by lay staff and nurses at the facility with support from community health workers. The goal is to keep patients engaged and adherent by providing social support and facilitating medication delivery and treatment monitoring while reducing patient burden at the clinic because these patients only require 6-monthly clinic visits and assessments for blood tests and rescripting. While each of the components of the ACs (prepacked medications, social support, adherence counseling, etc.) are currently considered important parts of the intervention, to date, none of the components have been tested alone to determine which components are essential.
DMD comprises prepacking and distribution of medications to PuPs, which are at locations other than the clinic pharmacy. Patients only need to come to the clinic on a 6-monthly basis for a clinical exam and rescripting. The goal is to reduce the time and resource burden on the patient while also decongesting clinics. DMD was established using three models: 1) facility-led prepackaging of medicines delivered to external sites; 2) Centralized Chronic Medicine Dispensing and Distribution (CCMDD), in which private partners prepackaged medicines and delivered them to the facility, which then distributed packs to dedicated PuPs (such as an AC) or delivered direct to a pharmacy or other dedicated PuP; and 3) chronic dispensing unit, in which the district pharmacist prepackaged medication and delivered to facilities, PuPs, or ACs. For clarity, we note that patients enrolled in the DMD analyses could not be in an AC.
DOH staff at facilities started to be trained on these interventions in July 2015, and training continued until mid-2016. At the start of data collection, the intervention sites had been implementing these interventions for at least 1 month and at some sites up to 6 months. At intervention sites, clinically stable patients were supposed to be given a choice between any of the three repeat prescription collection strategies available: ACs, DMD, or spaced fast-lane appointments (which we did not evaluate), though in some cases, facilities did decide which intervention a patient would be offered. Patients could also choose to switch between strategies, but this was monitored in our cohorts, and very few switches were recorded. It was not possible, however, for a patient to be enrolled in more than one approach at the same time.
We included patients over 18 years old who were resident in the facility’s catchment area, had no documented plan to transfer facilities, and were neither pregnant nor eligible for prevention of mother-to-child transmission services. Patients had to be eligible for a repeat prescription collection strategy by 1) being on the same ART regimen for at least 12 months, 2) having had their most recent viral load in the past 3 months, and 3) having had two consecutive undetectable viral loads (<400 copies/mL).
As the study was designed to assess retention-based outcomes, we did not interact with patients because this could impact outcomes. Instead, we received a waiver of consent and reviewed medical records to identify eligible patients. Follow-up started when patients became eligible to receive the intervention: the date of prescription visits for ACs/DMD for intervention patients and, for control patients, the clinic visit that occurred closest to the date that the medical record was reviewed. Follow-up start dates were between March 2016 and March 2017. At intervention sites, we included patients identified as actually having received the intervention (AC or DMD) in a clinic register or patient files. When more patients than required could be identified, we took a random sample of those eligible (for ACs, from all possible ACs). We then confirmed eligibility with clinic files. At control sites, we used clinic records to identify persons who would have been eligible for a repeat prescription strategy if they had been implemented at the facility and randomly allocated (using a computer) eligible patients to ACs or DMD to serve as controls (e.g., even random numbers were allocated to ACs, and odd random numbers were allocated to DMD). Again, we took a random sample of all those eligible. However, we note that because these participants were not offered the intervention, we could not identify which control subjects would have received the intervention if offered, and this could lead to some lack of comparability in our populations. Study flow figures for ACs and DMD are presented in Figs 1 and 2; the CONSORT checklist is presented in S1 CONSORT Checklist.
We followed patients through routinely collected facility- and patient-level records. Follow-up data came from three sources: 1) clinic records (clinic registers, AC registers, and patient clinic files), 2) a national electronic patients database called TIER.Net, and 3) the National Health Laboratory Service (NHLS) database of all viral loads done in South Africa’s public sector [25,30]. Passive follow-up through medical record review continued for 18 months (to allow a 6-month window for 1-year outcomes to occur).
Our first long-term outcome was sustained viral suppression (<400 copies/mL) at 12 months after eligibility for ACs or DMD, indicated by any suppressed viral load from TIER.Net, NHLS, or paper files. We did not specify a window around 12 months for suppression, so we defined this post hoc based on any suppressed viral load within 2 to 18 months after eligibility for ACs or DMD to allow time for viral loads to be recorded. If there were discordant results within the follow-up period (one suppressed and one unsuppressed viral load result), the outcome was classified as suppressed. We note that we consider this as continued viral suppression because everyone in the study had to have a suppressed viral load at entry, even though in some cases (N = 18 AC and N = 20 DMD), patients did have both an unsuppressed and suppressed viral load in that time period. We also report viral suppression among those patients who had a viral load for comparison to the results including all patients.
Our second outcome was retention in care at 12 months after eligibility for ACs or DMD defined as 100% − % attrition, with attrition as the sum of reported deaths, loss to follow-up, and transfers. Loss to follow-up was defined based on clinic definitions—failure to attend the clinic within 90 days of a scheduled appointment. For retention, we used TIER.Net and file review to measure 12-month outcomes. The database was locked on May 23, 2018.
Sample size and data analysis
We designed the study to have power = 80%, two-sided alpha = 0.05, and a coefficient of variation of 0.1 to account for clustering. We assumed 80% of patients would reach the primary outcome in the control arm. To detect a difference of 15%, we estimated we would need 24 subjects per clinic for each intervention (576 total).
We followed a similar analytic plan for each intervention and outcome. We compared arms on the proportion with the outcome and calculated crude risk differences and 95% CIs. Because we included those who actually received the intervention, our results should be considered as an “as treated” analysis. Next, we accounted for clustering using a linear regression generalized estimating equation (GEE) model of risk with an unstructured correlation matrix with site clustering. The model was then adjusted for differences in baseline covariates. Despite the cluster-randomized design, we anticipated differences between groups in the period prior to the intervention period. To account for this, we implemented a difference-in-differences (DiD) approach using data from a period prior to the intervention rollout (Jan 1, 2015 through Dec 31, 2015). Our model is where outcomeij is a binary outcome for person i at site j, period is a dummy variable indicating the period (1 = intervention, 0 = preintervention), intervention is a dummy variable indicating randomization group (0 = control, 1 = intervention), and β4 is the effect of the intervention (i.e., difference between intervention and control groups in the intervention period minus differences in the preperiod). The model is further adjusted for a vector of covariates Xij (which included sex, age, and CD4 count at ART initiation) and for clustering by clinic using GEEs. We also conducted subgroup analyses by age and sex, but these were not prespecified subgroup analyses and, as such, should be considered hypothesis generating. We note further that the decision to add in the DiD analysis was not in our initial protocol but added prior to data analysis, once we determined full randomization could not be maintained and we could not identify all control patients who would have received the interventions if offered.
The study was approved by the Human Research Ethics Committee (HREC) of the University of the Witwatersrand and the Boston University Institutional Review Board (IRB). Both approved use of routine clinic data for the evaluation and a waiver of consent. The trial is registered at clinicaltrials.gov (NCT02536768).
Table 1 presents the baseline data, and the CONSORT figure is shown in Fig 1. Our sample was largely female (72%) and aged 30 to 49 years (61%). Median CD4 count at ART initiation was 268 cells/ml3. Arms were well balanced with respect to baseline demographics, but we saw some small imbalance in CD4 count at ART initiation (median 278 versus 256), and those in the intervention arm had been on treatment for substantially longer than the control arm.
About 84% of subjects had a repeat viral load within 18 months of club eligibility, but this did not differ by arm. Sustained viral suppression was high overall at about 80%, including those without a repeat viral load (suppression among those with a viral load was about 95%). This is expected given that those eligible for ACs are highly adherent already. In our crude analysis, ACs were associated with little change in sustained viral suppression (risk difference [RD]: 0.4%; 95% CI: −6.2% to 7.0%) (Table 2). We also saw higher sustained suppression in control sites (RD:−2.7%; 95% CI: −4.0% to −1.4%) prior to intervention rollout (S2 Table). When differences in sustained suppression prior to intervention rollout were combined with the enrolled cohort using DiD, we saw comparable suppression within 18 months (RD: 3.1%; 95% CI: −3.8% to 10.0%), and when adjusting for individual characteristics, CIs widened (RD: 3.8%; 95% CI: −6.9% to 14.4%) (full model in S3 Table). Overall, we saw evidence for comparable viral load outcomes between arms. We found little difference between intervention arms in sustained viral suppression results overall (Fig 1, left) and limited to those with a repeat viral load (Fig 3, right).
AC, Adherence Club; DiD, difference in differences; VL, viral load.
Retention at 12 months was high in both arms (85%), as expected given the targeted population. ACs were associated with a crude 7.8 (95% CI: 2.1% to 13.6%) percentage-point increase in retention (Table 3). Prior to intervention rollout, we saw no difference in retention between arms (RD: 0.4%; 95 CI: −0.4% to 1.2%) (S4 Table). Using DiD, we saw an increase in 12-month retention among those on ACs (RD: 7.4%; 95% CI: 2.9% to 11.9%) compared to controls and a similar result when controlling for clustering and individual characteristics (RD: 8.3%; 95% CI: 1.1% to 15.6%) (full model in S5 Table). Further, retention effects were apparently twice as large for men (RD: 13.1%; 95% CI: 0.3% to 23.5%) than women (RD: 6.0%; 95% CI: −0.9% to 12.9%), though estimates were imprecise. There seemed to be little benefit to those under 30, while we observed somewhat large benefits in some older age groups (S6 Table). This was true even though absolute values of retention increased with increasing age.
The DMD cohort baseline characteristics were similar to the AC cohort, with roughly 65% between 30 and 49 years and about 70% female. Median CD4 count at ART initiation was 270 cells/ml3 at enrollment. Intervention and control arms were similar with respect to sex and age, but again, we found some small imbalances in CD4 (279 versus 256 cells/ml3) and log viral load at eligibility (log10 1.62 versus 2.09 copies/ml) (Table 4, and the CONSORT figure is shown in Fig 3). As with ACs, those in the DMD intervention arm had been on treatment for substantially longer than the control arm. For DMD, because we were not able to preserve randomization, we included clinics actually implementing DMD in the intervention arm and those that were not in the control arm.
We found that sustained suppression rates were only about 75% overall, but this was largely because of patients without a repeat viral load. Among those with a repeat viral load, suppression was about 95%. Comparing DMD implementation sites to those not implementing DMD, sustained suppression was similar (RD: 2.9%; 95% CI: −4.2% to 10.0%) (Table 5). Prior to intervention rollout, sustained suppression was slightly higher in implementation sites compared to control sites (RD: 3.3%; 95% CI: 2.0% to 4.6%) (S7 Table). Controlling for baseline differences, we found no difference in sustained suppression (RD: −0.5%; 95% CI: −7.5% to 6.6%). Results were largely unchanged when adjusting for clustering and individual characteristics (RD: −1.0%; 95% CI: −12.2% to 10.1%) (full model in S8 Table), though results were less precise. Further, while there was no benefit overall, there may be some among men (RD: 11.1%; 95% CI: −3.4% to 25.5%) (S9 Table). When limited to those with a viral load, viral suppression was sustained in 8% more of those patients in the DMD compared to the standard of care (from 90.2% to 98.3%) in the intervention period, though this disappeared using DiD with cluster adjustment (RD: −2.2%; 95% CI: −11.7% to 7.3%, data not presented in tables). Still, this suggests sustained suppression among those tested is higher in implementation sites, while repeat testing may be somewhat worse. Fig 4 presents the full results on the left and limited to those with a viral load on the right.
DiD, difference in differences; DMD, Decentralized Medication Delivery; VL, viral load.
Retention was high overall (about 85%) with, apparently, somewhat lower retention in DMD than control sites (RD: −5.8%; 95% CI: −11.7% to 0.2%). Prior to intervention, we found nearly identical retention (RD 0.3%; 95% CI: −0.5% to 1.1%) (S10 Table). Adjusting for baseline differences using DiD, we found decreased retention at intervention sites (RD −6.0%; 95% CI: −10.6% to−1.1%) (Table 6) that changed little when adjusting for baseline differences between arms (RD −5.9%; 95% CI: −12.5% to 0.8%) (full final model in S11 Table). As with suppression the difference in retention, if real, appears to be largely among men (S12 Table).
In one of the first randomized trials of decentralized service delivery for HIV care, we found that AC patients had higher 1-year retention (89.5% versus 81.6%) and comparable sustained 1-year viral suppression (80.0% versus 79.6%) compared to those receiving standard of care. The retention association we observed was stronger for men than women (men RD: 13.1%, 95% CI: 0.3% to 23.5%; women RD: 6.0%, 95% CI: −0.9% to 12.9%). For DMD, we found that those in DMD had apparently lower retention (81.5% versus 87.2%) and comparable viral suppression compared to patients receiving standard of care (77.2% versus 74.3%), though our findings were imprecise. We also found apparently increased suppression among men (RD: 11.1%; 95% CI: −3.4% to 25.5%). Thus, our findings demonstrate that such approaches can be implemented with potential for benefit and, at a minimum, no harm among patients who already have high rates of viral suppression and retention.
South Africa’s National AGLs were developed to improve treatment outcomes and reduce the burden on clinically stable patients while allowing for provision of more intense care to individuals with advanced HIV disease. For repeat prescription strategies like DMD and for adherence approaches like ACs that also include repeat scripting, even if outcomes remain comparable to the control group, this could be seen as a benefit because these interventions are designed to make care easier for those clinically stable patients and help clinicians and staff to provide increased support to clients with advanced HIV disease and for those on a failing ART regimen, as well as freeing up capacity . In fact, we might expect that the benefits of these approaches would at best be small because these interventions are designed for patients who already have high suppression and retention. We previously showed ACs were preferred by patients . Here, we found ACs supported sustained viral suppression with comparable sustained viral suppression (RD: 3.8%; 95% CI: −6.9% to 14.4%) and an increase in 12-month retention (RD: 8.3%; 95% CI: 1.1% to 15.6%). Given that comparable outcomes would be acceptable for ACs, this finding, combined with our early outcomes, suggests ACs are effective. Our findings are in line with previous work that has also shown benefits to ACs [33–40]. Some have noted that ACs are inexpensive to implement and save patients time , and others have shown they are popular with patients [42,43]. Studies in Cape Town have shown retention to be over 90% after 12 months in clubs [21,44–47], much like in our sample . The same has been shown in Kenya . While these studies lacked comparison groups, in Khayelitsha, South Africa, ACs were associated with about a 12% increase in 12-month retention , slightly larger, though in line with, what we found. Unlike previous work, ours is one of the first to use at least some randomization for ACs (we were not able to do so for DMD), and this, combined with our DiD approach, allowed us to draw stronger conclusions. Still, caution is needed in taking the results to scale, as experience from the Western Cape has shown. There, qualitative research has shown that moving beyond the pilot phase came with complex challenges, and sites were less likely to see the benefits when the intervention was used on a large scale .
We did not look at short-term medication pick-up outcomes for DMD because data on pick-ups were kept by private providers and were not always fed back into patient records. Here, we looked at long-term outcomes, one of which (sustained viral suppression) did not rely on data from outside service providers. DMD was more difficult to evaluate because not all sites that were supposed to implement DMD did so, and some that were not supposed to did. Still, we found no differences in sustained viral suppression between study arms (RD: −1.0%; 95% CI: −12.2% to 10.1%). This could indicate a benefit of the intervention because it frees up clinic space. We did observe an increase in attrition associated with the intervention (RD: −5.9%; 95% CI: −12.5% to 0.8%), but we suspect this is unlikely to reflect real differences given that retention was likely influenced by visit information not being returned to clinics files. Thus, the fact that we saw only small differences suggests retention is likely unchanged. Overall, DMD appears to be a useful intervention. Unlike ACs, which have been implemented in numerous settings, there is little evidence on DMD with which to compare our data.
The evaluation was designed to evaluate the AGLs as implemented compared to standard of care. The use of a cluster-randomized evaluation helped increase validity, and use of routine data collection helped prevent the study influencing retention and suppression-based outcomes. At the same time, these choices also come with important limitations. First, it is critical when interpreting these results to understand that the control sites were not pure control groups because forms of the interventions were being implemented at many control sites. This would likely move estimates towards the null. Second, unlike individually randomized trials, our cluster-randomized design reduced power and increased the chances of confounding. Our DiD approach should help mitigate this, but some residual confounding may be left. Third, our use of routine data led to some missing data that we could not prevent. Fourth, not all those eligible for the intervention necessarily received it, and if those at intervention sites were a select group of patients, this could create bias. In addition, because subjects were not offered the interventions at control sites, we could not determine who would have received them if they had been offered. This could lead to some lack of comparability in our study populations at baseline. Fifth, while we did initially randomize clusters, this is not a fully randomized trial, both because for the DMD intervention, randomization was not maintained, but also because in the intervention sites, we only enrolled subjects who received the intervention, whereas in the control sites, we could only enroll a sample of those eligible for the interventions, even if they never would have received them if offered. The impact of this is not clear. Sixth, we note that those in the intervention arm had been on treatment for substantially longer than the control arm. We suspect this may be because providers prioritized targeting the interventions towards those who were on treatment far longer than the minimum required and had the longest track record of demonstrating stability on treatment. This could have the effect of making the interventions seem better than they are. Seventh, we conducted the study during the rollout, before sites had experience with the interventions, and as such, results under full implementation may differ. Eighth, DMD is largely run by parties outside the clinic so that data collection on the interventions was, during the study period, often controlled by third parties and did not always make it back into the data collection systems at the sites. This could have affected our long-term retention outcome. Ninth, we note that we did not a priori decide to stratify our analyses by sex, and therefore, sex-stratified results should be considered hypothesis generating and require confirmation. Finally, because these interventions were implemented at test sites, we do not know for sure that the results will generalize to all clinics in South Africa.
In conclusion, we found that ACs and DMD, in agreement with other studies, are feasible, acceptable, and for ACs, had positive outcomes. We saw an overall retention benefit to ACs and comparable outcomes with DMD against standard of care, which should still prove to be a benefit to clinics as they are designed to decongest the clinic overall. ACs as a medication refill model worked especially for male patients, given the benefits seen in retention and sustained viral suppression. As the AGL intervention package is taken to scale, it will be important to track patient retention and sustained viral suppression and to quantify the cumulative cost benefit of the package of interventions across the HIV care cascade.
S1 CONSORT Checklist. CONSORT 2010 checklist of information to include when reporting a cluster-randomized trial.
S1 Table. Population data (facility headcount and total active patients) at each facility and total numbers eligible by intervention (I) and control (C) for each intervention.
S2 Table. Viral suppression at 12 months (defined as within 2–18 months) for all those who would have been eligible for ACs in the period prior to the rollout of the interventions (Jan 1, 2015 through Dec 31, 2015) (preperiod).
AC, Adherence Club.
S3 Table. Regression coefficients for final model for DiD analysis of AC viral suppression at 12 months (defined as within 2–18 months) adjusted for site-level clustering.
AC, Adherence club; DiD, difference in differences.
S4 Table. Retention (alive and in care) at 12 months for all those who would have been eligible for ACs in the period prior to the rollout of the interventions (Jan 1, 2015 through Dec 31, 2015) (preperiod).
AC, Adherence Club.
S5 Table. Regression coefficients for final model for DiD analysis of AC retention within 12 months adjusted for site-level clustering.
AC, Adherence Club; DiD, difference in differences.
S6 Table. Effects for retention (alive and in care) at 12 months for those eligible for ACs during the intervention period (enrolled subjects only) by sex and age.
AC, Adherence Club.
S7 Table. Viral suppression at 12 months (defined as within 2–18 months) for all those who would have been eligible for DMD in the period prior to the rollout of the interventions (Jan 1, 2015 through Dec 31, 2015) (preperiod).
DMD, Decentralized Medication Delivery.
S8 Table. Regression coefficients for final model for DiD analysis of DMD viral suppression at 12 months (defined as within 2–18 months) adjusted for site-level clustering.
DiD, difference in differences; DMD, Decentralized Medication Delivery.
S9 Table. Associations for viral suppression at 12 months (defined as within 2–18 months) for those eligible for DMD during the intervention period (enrolled subjects only) by sex.
DMD, Decentralized Medication Delivery.
S10 Table. Retention (alive and in care) at 12 months for all those who would have been eligible for DMD in the period prior to the rollout of the interventions (Jan 1, 2015 through Dec 31, 2015) (preperiod).
DMD, Decentralized Medication Delivery.
S11 Table. Regression coefficients for final model for DiD analysis of DMD retention at 12 months adjusted for site-level clustering.
DiD, difference in differences; DMD, Decentralized Medication Delivery.
S12 Table. Associations for retention (alive and in care) at 12 months for those eligible for DMD during the intervention period (enrolled subjects only) by sex.
DMD, Decentralized Medication Delivery.
- 1. Cornell M, Grimsrud A, Fairall L, Fox MP, van Cutsem G, Giddy J, et al. Temporal changes in programme outcomes among adult patients initiating antiretroviral therapy across South Africa, 2002–2007. AIDS. 2010; 24(14):2263–2270. pmid:20683318
- 2. Cornell M, Johnson LF, Wood R, Tanser F, Fox MP, Prozesky H, et al. Twelve-year mortality in adults initiating antiretroviral therapy in South Africa. J Int AIDS Soc. 2017;20: 21902. pmid:28953328
- 3. Bor J, Herbst AJ, Newell M-L, Bärnighausen T. Increases in adult life expectancy in rural South Africa: valuing the scale-up of HIV treatment. Science. 2013;339: 961–5. pmid:23430655
- 4. Johnson LF, Mossong J, Dorrington RE, Schomaker M, Hoffmann CJ, Keiser O, et al. Life expectancies of South African adults starting antiretroviral treatment: collaborative analysis of cohort studies. PLoS Med. 2013;10: e1001418. pmid:23585736
- 5. Boulle A, Bock P, Osler M. Antiretroviral therapy and early mortality in South Africa. Bull World Health Organ. 2008;86: 678–687. pmid:18797643
- 6. Boulle A, Schomaker M, May MT, Hogg RS, Shepherd BE, Monge S, et al. Mortality in Patients with HIV-1 Infection Starting Antiretroviral Therapy in South Africa, Europe, or North America: A Collaborative Analysis of Prospective Studies. PLoS Med. 2014;11: e1001718. pmid:25203931
- 7. Granich RM, Gilks CF, Dye C, De Cock KM, Williams BG. Universal voluntary HIV testing with immediate antiretroviral therapy as a strategy for elimination of HIV transmission: a mathematical model. Lancet. 2009;373: 48–57. pmid:19038438
- 8. Cohen MS, Chen YQ, McCauley M, Gamble T, Hosseinipour MC, Kumarasamy N, et al. Prevention of HIV-1 infection with early antiretroviral therapy. N Engl J Med. 2011;365: 493–505. pmid:21767103
- 9. Fox MP, Rosen S. Retention of Adult Patients on Antiretroviral Therapy in Low- and Middle-Income Countries: Systematic Review and Meta-analysis 2008–2013. J Acquir Immune Defic Syndr. 2015;69: 98–108. pmid:25942461
- 10. Fox MP, Rosen S. Patient retention in antiretroviral therapy programs up to three years on treatment in sub-Saharan Africa, 2007–2009: Systematic review. Trop Med Int Heal. 2010;15: 1–15. pmid:20586956
- 11. Rosen S, Fox MP. Retention in HIV care between testing and treatment in sub-Saharan Africa: A systematic review. PLoS Med. 2011;8: e1001056. pmid:21811403
- 12. Mugglin C, Estill J, Wandeler G, Bender N, Egger M, Gsponer T, et al. Loss to programme between HIV diagnosis and initiation of antiretroviral therapy in sub-Saharan Africa: systematic review and meta-analysis. Trop Med Int Heal. 2012;17: 1509–20. pmid:22994151
- 13. Kranzer K, Govindasamy D, Ford N, Johnston V, Lawn SD. Quantifying and addressing losses along the continuum of care for people living with HIV infection in sub-Saharan Africa: a systematic review. J Int AIDS Soc. 2012;15: 17383. pmid:23199799
- 14. Fox MP, Shearer K, Maskew M, Meyer-Rath G, Clouse K, Sanne I. Attrition through multiple stages of pre-treatment and ART HIV care in South Africa. PLoS ONE. 2014;9: e110252. pmid:25330087
- 15. The International AIDS Society. Differentiated Care [Internet]. [cited 25 Jul 2017]. Available from: http://www.differentiatedcare.org/about.
- 16. Phillips A, Shroufi A, Vojnov L, Cohn J, Roberts T, Ellman T, et al. Sustainable HIV treatment in Africa through viral-load-informed differentiated care. Nature. 2015;528: S68–S76. pmid:26633768
- 17. Grimsrud A, Bygrave H, Doherty M, Ehrenkranz P, Ellman T, Ferris R, et al. Reimagining HIV service delivery: the role of differentiated care from prevention to suppression. J Int AIDS Soc. 2016;19: 21484. pmid:27914186
- 18. Ssonko C, Gonzalez L, Mesic A, da Fonseca MS, Achar J, Safar N, et al. Delivering HIV care in challenging operating environments: the MSF experience towards differentiated models of care for settings with multiple basic health care needs. J Int AIDS Soc. 2017;20: 21654. pmid:28770590
- 19. Khabala KB, Edwards JK, Baruani B, Sirengo M, Musembi P, Kosgei RJ, et al. Medication Adherence Clubs: a potential solution to managing large numbers of stable patients with multiple chronic diseases in informal settlements. Trop Med Int Heal. 2015;20: 1265–1270. pmid:25962952
- 20. Tsondai PR, Wilkinson LS, Grimsrud A, Mdlalo PT, Ullauri A, Boulle A. High rates of retention and viral suppression in the scale-up of antiretroviral therapy adherence clubs in Cape Town, South Africa. J Int AIDS Soc. 2017;20: 21649. pmid:28770595
- 21. Grimsrud A, Lesosky M, Kalombo C, Bekker L-G, Myer L. Implementation and Operational Research: Community-Based Adherence Clubs for the Management of Stable Antiretroviral Therapy Patients in Cape Town, South Africa: A Cohort Study. JAIDS J Acquir Immune Defic Syndr. 2016;71: e16–23. pmid:26473798
- 22. Luque-Fernandez MA, Van Cutsem G, Goemaere E, Hilderbrand K, Schomaker M, Mantangana N, et al. Effectiveness of patient adherence groups as a model of care for stable patients on antiretroviral therapy in Khayelitsha, Cape Town, South Africa. PLoS ONE. 2013;8: e56088. pmid:23418518
- 23. Joint United Nations Programme on HIV/AIDS (UNAIDS). The gap report–July 2014. Geneva, Switzerland: UNAIDS; 2014. UNAIDS / JC2656. ISBN 978-92-9253-062-4.
- 24. South Africa National Department of Health. Health Indicators Update: Antiretroviral Indicators. Directorate: Monitoring and Evaluation. South Africa, Pretoria: National Department of Health South Africa; 2013.
- 25. Fox MP, Bor J, Brennan AT, MacLeod WB, Maskew M, Stevens WS, et al. Estimating retention in HIV care accounting for patient transfers: A national laboratory cohort study in South Africa. PLoS Med. 2018;15: e1002589. pmid:29889844
- 26. National Department of Health. National adherence guidelines for chronic diseases (HIV, TB and NCDs), Version: 7 April 2015. South Africa, Pretoria: National Department of Health South Africa; 2015.
- 27. National Department of Health Republic of South Africa. Standard Operating Procedures for Minimum Package of Interventions to Support Linkage to Care, Adherence and Retention in Care, Adherence Guidelines for HIV, TB and NCDs. Pretoria, South Africa: National Department of Health South Africa; 2016.
- 28. Fox MP, Pascoe SJ, Huber AN, Murphy J, Phokojoe M, Gorgens M, et al. Assessing the impact of the National Department of Health’s National Adherence Guidelines for Chronic Diseases in South Africa using routinely collected data: a cluster-randomised evaluation. BMJ Open. 2018;8: e019680. pmid:29358446
- 29. Fox MP, Pascoe SJS, Huber AN, Murphy J, Phokojoe M, Gorgens M, et al. Effectiveness of interventions for unstable patients on antiretroviral therapy in South Africa: results of a cluster-randomised evaluation. Trop Med Int Heal. 2018;23: 1314–1325. pmid:30281882
- 30. Carmona S, Bor J, Nattey C, Maughan-Brown B, Maskew M, Fox MP, et al. Persistent High Burden of Advanced HIV Disease among Patients Seeking Care in South Africa’s National HIV Program: Data from a Nationwide Laboratory Cohort. Clin Infect Dis. 2018;66: S111–S117. pmid:29514238
- 31. Haberer JE, Sabin L, Amico KR, Orrell C, Galárraga O, Tsai AC, et al. Improving antiretroviral therapy adherence in resource-limited settings at scale: a discussion of interventions and recommendations. J Int AIDS Soc. Wiley-Blackwell; 2017;20: 21371. pmid:28630651
- 32. The World Bank. Evaluation of the National Adherence Guidelines for Chronic Diseases in South Africa: Healthcare Provider Perspectives on Different Care Models, 2017. Washington DC: World Bank; 2017.
- 33. Nsanzimana S, Remera E, Ribakare M, Burns T, Dludlu S, Mills EJ, et al. Phased implementation of spaced clinic visits for stable HIV-positive patients in Rwanda to support Treat All. J Int AIDS Soc. 2017;20: 21635. pmid:28770591
- 34. Mody A, Roy M, Sikombe K, Savory T, Holmes C, Bolton-Moore C, et al. Improved Retention with Six Month Clinic Return Intervals for Stable HIV-Infected Patients in Zambia. Clin Infect Dis. 2017;66: 237–243. pmid:29020295
- 35. Bemelmans M, Baert S, Goemaere E, Wilkinson L, Vandendyck M, van Cutsem G, et al. Community-supported models of care for people on HIV treatment in sub-Saharan Africa. Trop Med Int Health. 2014;19: 968–77. pmid:24889337
- 36. Prust ML, Banda CK, Nyirenda R, Chimbwandira F, Kalua T, Jahn A, et al. Multi-month prescriptions, fast-track refills, and community ART groups: results from a process evaluation in Malawi on using differentiated models of care to achieve national HIV treatment goals. J Int AIDS Soc. 2017;20: 21650. pmid:28770594
- 37. Kwarisiima D, Kamya MR, Owaraganise A, Mwangwa F, Byonanebye DM, Ayieko J, et al. High rates of viral suppression in adults and children with high CD4+ counts using a streamlined ART delivery model in the SEARCH trial in rural Uganda and Kenya. J Int AIDS Soc. 2017;20: 21673. pmid:28770596
- 38. Obua C, Kayiwa J, Waako P, Tomson G, Balidawa H, Chalker J, et al. Improving adherence to antiretroviral treatment in Uganda with a low-resource facility-based intervention. Glob Health Action. 2014;7: 24198. pmid:24909408
- 39. Alamo ST, Wagner GJ, Ouma J, Sunday P, Marie L, Colebunders R, et al. Strategies for optimizing clinic efficiency in a community-based antiretroviral treatment programme in Uganda. AIDS Behav. 2013;17: 274–83. pmid:22610422
- 40. Babigumira JB, Castelnuovo B, Stergachis A, Kiragga A, Shaefer P, Lamorde M, et al. Cost effectiveness of a pharmacy-only refill program in a large urban HIV/AIDS clinic in Uganda. PLoS ONE. 2011;6: e18193. pmid:21464895
- 41. Bango F, Ashmore J, Wilkinson L, van Cutsem G, Cleary S. Adherence clubs for long-term provision of antiretroviral therapy: cost-effectiveness and access analysis from Khayelitsha, South Africa. Trop Med Int Heal. 2016;21: 1115–1123. pmid:27300077
- 42. Venables E, Edwards JK, Baert S, Etienne W, Khabala K, Bygrave H. “They just come, pick and go.”; The Acceptability of Integrated Medication Adherence Clubs for HIV and Non Communicable Disease (NCD) Patients in Kibera, Kenya. PLoS ONE. 2016;11: e0164634. pmid:27764128
- 43. Dudhia R, Kagee A. Experiences of participating in an antiretroviral treatment adherence club. Psychol Health Med. 2015;20: 488–494. pmid:25168720
- 44. Grimsrud A, Sharp J, Kalombo C, Bekker L-G, Myer L. Implementation of community-based adherence clubs for stable antiretroviral therapy patients in Cape Town, South Africa. J Int AIDS Soc. 2015;18: 19984. pmid:26022654
- 45. Wilkinson L, Harley B, Sharp J, Solomon S, Jacobs S, Cragg C, et al. Expansion of the Adherence Club model for stable antiretroviral therapy patients in the Cape Metro, South Africa 2011–2015. Trop Med Int Heal. 2016;21: 743–9. pmid:27097834
- 46. Campion EW. Treating Millions for HIV—The Adherence Clubs of Khayelitsha. N Engl J Med. 2015;372: 301–303. pmid:25607424
- 47. Bateman C. MSF again paves the way with ART. S Afr Med J. 2013;103: 71–3. pmid:23374312
- 48. MacGregor H, McKenzie A, Jacobs T, Ullauri A. Scaling up ART adherence clubs in the public sector health system in the Western Cape, South Africa: a study of the institutionalisation of a pilot innovation. Global Health. 2018;14: 40. pmid:29695268