D ow nloaded from https://journals.lw w .com /jaids by BhD M f5ePH Kav1zEoum 1tQ fN 4a+kJLhEZgbsIH o4XM i0hC yw C X1AW nYQ p/IlQ rH D 3i3D 0O dR yi7TvSFl4C f3VC 4/O AVpD D a8K2+Ya6H 515kE= on 01/15/2021 Downloadedfromhttps://journals.lww.com/jaidsbyBhDMf5ePHKav1zEoum1tQfN4a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE=on01/15/2021 IMPLEMENTATION SCIENCE Twelve-Month Outcomes of Community-Based Differentiated Models of Multimonth Dispensing of ART Among Stable HIV-Infected Adults in Lesotho: A Cluster-Randomized Noninferiority Trial Betty B. Tukei, MSc, BSc,a Geoffrey Fatti, MBChB, MPH,b,c Appolinaire Tiam, MBChB, DipHIVMan, MMed,d,e Nicoletta Ngorima-Mabhena, MBChB, MSc,b Vincent J. Tukei, MBChB, MMed, MPH,d Itumeleng Tshabalala, Dipl.,f Veronica M. Sejana, BSc,a Trish Muzenda, MPH,b Lincoln M. Mokoroane, BSc,a Lebelang Sehlabo, BA,a Thapelo Maotoe, MBChB, MD,g Justine K. Mirembe, MBChB,h Ian Membe, MPH,h Francis Akpan, MBchB, MPH,g Khotso Maile, MSc,a Iyiola Faturiyele, MBChB, MPH,h Thembi Xulu, MB BCh, MPH,g Thomas Minior, MD, MPH,h Ian Sanne, MB BCh,g,i and Charles Chasela, BSc, MSc, DLSHTM, PhD,g,j for EQUIP Health Background: Lesotho adopted the test-and-treat approach for HIV treatment in June 2016, which increased antiretroviral treatment (ART) clinic volume. We evaluated community-based vs. facility- based differentiated models of multimonth dispensing of ART among stable HIV-infected adults in Lesotho. Methods: Thirty facilities were randomized to 3 arms, facility 3- monthly ART (3MF) (control), community ART groups (3MC), and 6-monthly community distribution points (6MCD). We estimated risk differences (RDs) between arms using population-averaged generalized estimating equations, controlling for baseline imbalances and specifying for clustering. The primary outcome was retention in ART care by intention-to-treat and virologic suppression as a secondary outcome (ClinicalTrials.gov: NCT03438370). Results: A total of 5,336 participants were enrolled, with 1898, 1558, and 1880 in 3MF, 3MC, and 6MCD, respectively. Retention in ART care was not different across arms and achieved the prespecified noninferiority limit (23.25%) between 3MC vs. 3MF (control); 6MCD vs. 3MF; and 6MCD vs. 3MC, adjusted RD = 20.1% [95% confidence interval (CI): 21.6% to 1.5%], adjusted RD = 21.3% (95% CI: 23.0% to 0.5%), and adjusted RD = 21.2% (95% CI: 22.9% to 0.5%), respectively. After 12 months, 98.6% (n = 1503), 98.1% (n = 1126), and 98.3% (n = 1285) were virally load (VL) suppressed in 3MF, 3MC, and 6MCD, respectively. There were no differences in VL between 3MC vs. control and 6MCD vs. control, risk ratio (RR) = 1.00 (95% CI: 0.98 to 1.01) and RR = 1.00 (95% CI: 0.98 to 1.01), respectively. Conclusions: There were no differences in retention and VL suppression for stable HIV-infected participants receiving multi- month dispensing of ART within community-based differentiated models when compared with the facility-based standard-of- care model. Key Words: HIV, differentiated models of care, community ART groups, Lesotho, antiretroviral treatment (J Acquir Immune Defic Syndr 2020;85:280–291) INTRODUCTION Background In June 2016, Lesotho adopted the WHO Test-and-Treat approach requiring immediate initiation of antiretroviral treat- ment (ART) for all PLHIV.1 ART is lifelong treatment and requires patient adherence to medication and retention in care. Barriers to ART adherence and retention are long distance to health facilities, long waiting times, travel costs, facility congestion, poor social support, and patient autonomy.2–8 Many PLHIV in sub-Saharan Africa undergo frequent visits Received for publication April 17, 2020; accepted June 22, 2020. From the aRight to Care/EQUIP Health, Maseru, Lesotho; bKheth’Impilo AIDS Free Living, Cape Town, South Africa; cDivision of Epidemiology and Biostatistics, Department of Global Health, Faculty of Medicine and Health Sciences, Stellenbosch University, Cape Town, South Africa; dElizabeth Glaser Pediatric AIDS Foundation, Washington, DC; eCentre for Interna- tional Health, University of Bergen, Bergen, Norway; fLesotho Ministry of Health Maseru, Lesotho; gRight to Care/EQUIP Health, South Africa; hDivision of Prevention Care and Treatment, U.S. Agency for International Development, Washington, DC; iDepartment of Internal Medicine, School of Clinical Medicine, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa; and jDepartment of Epidemi- ology and Biostatistics, School of Public Health, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, South Africa. Supported by The President’s Emergency Plan for AIDS Relief through the US Agency for International Development EQUIP Mechanism grant number AID-OAA-A-15-00070. The authors have no funding or conflicts of interest to disclose. Correspondence to: Charles Chasela, BSc, MSc, DLSHTM, PhD, Research Unit, Right to Care, 1006 Lenchen Avenue North, Centurion, Pretoria, South Africa (e-mail: charles.chasela@wits.ac.za). Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. 280 | www.jaids.com J Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. mailto:charles.chasela@wits.ac.za for ART pick ups every 1–2 months, congesting already burdened health systems, resulting in poor adherence and/or detachment from care for the patient. Scaling up of ART services requires adequate supply of ART of at least 3 months,9 thus minimizing frequent medication pick ups. Lesotho currently recommends 3–6-month dispensing of ARV drugs for patients with viral load ,1000 co- pies/mL.10 However, because of logistical and other structural challenges (storage space, adequate supplies, and standing country policies), multimonth dispensing (MMD) in Lesotho was only introduced at 3-monthly intervals in facilities only for stable ART patients.10 Community-based models of care are postulated to further strengthen adherence, retention, and viral suppression.11 Community ART Groups (CAGs) have been successfully implemented elsewhere in Southern Africa, and introducing MMD into them can have the same intended outcomes as facility-based interventions, while additionally decongesting and decentralizing facilities.12 We evaluated implementation of community-based vs. facility-based differ- entiated models of MMD of ART among stable HIV-infected adults in Lesotho. METHODS Study Design A detailed description of the study protocol is described elsewhere.13 Briefly, this was a cluster-randomized noninfer- iority trial with 3 differentiated models of multimonth scripting and dispensing (MMSD) undertaken in 30 facilities (clusters), in Lesotho. The selection criteria (Fig. 1) were as follows: (1) had at least 430 patients currently on ART to fulfill the facility sample size requirement, (2) were either already implementing CAGs or implementation of CAGs was deemed feasible, (3) had adequate supply chain procedures for MMSD, and (4) routine data collection procedures were in place. Each study arm had 10 facility clusters. The clusters were stratified into urban and rural, according to the geographic location, and randomized to: (1) control—3-monthly dispensing of ART at health facilities (3MF); (2) intervention 3-monthly dispensing of ART within CAGs (3MC); and (3) 6-monthly dispensing of ART at community distribution points (6MCD). The study was approved by the Lesotho National Health Research Ethics Committee and the Advarra Institutional Review Board, USA. Study Population The study population consisted of HIV-infected adults (age$18 years) who had received first-line ART for at least 6 months, had baseline viral load ,1000 copies/mL within the past 12 months (stable patient), provided written informed consent, and willing to participate in the arm their health facility was randomized to. Exclusion criteria included patients on second-line or third-line ART, those with comorbidities requiring more frequent facility visits, those who had ART modifications since the last viral load test, those who were pregnant and/or breastfeeding, those classi- fied as WHO clinical stage 3 or 4 in the past 3 months, or those participating in another study requiring dispensing of drugs. Before study participation, participants received ART on a monthly basis as per national guidelines or received a 2- monthly supply if working in South Africa. Study Intervention and Procedures Study Interventions The interventions in this study were the use of CAGs and community outreach distribution points for ART MMSD.14,15 Three-Month Dispensing Model at Health Facilities Providers at facilities randomized to 3-month dispens- ing provided all enrolled patients with a 3-month supply of ART. All other aspects of care were part standard of care for the enrolling clinic, and routine visits occurred every 90 days instead of the standard-of-care interval. CAG Model (3-Month Supply of Drug) CAGs are groups of stable patients on ART who meet in the community and manage their own health by taking turns obtaining medications from the clinic for the entire group, self-monitoring medication adherence, and providing support to each other.14,16,17 Study participants enrolled in this arm joined new/previously existing CAGs or were already members. The CAG consisted of 2–12 participants, who lived in a similar geographic location and attended the same health care facility. The members’ appointments were synchronized to ensure that scheduled clinic visits were on the same day. Stable participants were required to have clinical consultation and viral load (VL) testing at the facility at 12 and 24 months after enrollment. Each member of the CAG collected their own 3-month supply of ART from the clinic on this day. Participants who were ill at any stage of the study reported to the clinic as soon as it was possible. A CAG representative distributed the drugs to the other members at the CAG meeting on the same day or the following day. The dates of these visits were noted by the facility to ensure that all drugs were predispensed and ready during this visit date. Community ART Distribution Model by the Health Care Worker (6-Month Supply of Drug) Participants were identified from the facilities and sensitized on collecting ART in the community. The distri- bution points are community outreach points used for health service delivery. Their next ART refill was conducted in 6 months in the community at an outreach post. These encounters in the community were on an individual provider–patient basis, not as part of a CAG. The ART refill was conducted by a health care worker, appropriately trained and certified for dispensing ART to stable patients. Study participants’ adherence was assessed at this community distribution point. Participants continued to receive 6-month refills in the community if they remained stable (ie, have VL less than 1000 copies per mL). Stable participants were required to have a clinical consultation and VL done at the facility at 12 months and 24 months after enrollment. Community-Based Differentiated ModelsJ Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. www.jaids.com | 281 Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. PROCEDURES Screening and Enrollment Study nurses, stationed at each study facility, reviewed each patient’s clinical file to assess study eligibility. The nurses liaised with the facility health workers to refer all patients identified as potentially eligible for study enrollment. The study nurse invited referred patients to participate in the study and provided additional information on the study to all patients who expressed interest in the study. All study participants provided written informed consent before study screening and enrollment as part of the study informed consent process. Patients who met the inclusion criteria and provided informed consent were enrolled in the study. Each participant received a unique study number for identification during follow-up and assessment of participants’ records. All participants were enrolled from the facilities randomized to the 3 study arms. 3MF participants received the standard-of- care 3-monthly supply of ART at the facility. 3MC partici- pants were enrolled from members of existing/newly formed CAGs and were given 3-monthly supply of ART. For the facilities randomized to the 6MCD arm, enrolled participants were given their first 6-monthly ART supply at enrollment. Patients only returned to the facility for annual clinical assessments that include viral load blood draws. Follow-up Visits During follow-up, the study nurses reviewed the patient files and completed both a paper and electronic monitoring tool for each participant. The monitoring tool captured FIGURE 1. Trial flow diagram. CV, coefficient of variation of cluster size. Tukei et al J Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 282 | www.jaids.com Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. participants’ vitals on the study outcomes. Tracing of all participants who missed ART appointments was performed as part of routine activities by an existing community imple- menting partner, Lesotho Network of AIDS Services Orga- nizations (LENASO). Electronic data were captured using tablets and entered into a RedCap database onto a central server. Patients who were clinically unstable requiring more frequent or intense clinical follow-up were transitioned off the study arm and referred back to the facility. Outcomes The primary outcome was retention in ART care defined as the proportion of participants remaining in care 12 months after study enrollment. The secondary outcomes were (1) viral suppression, defined as the proportion of participants with viral load ,1000 copies/mL 12 months after study enrollment, and (2) the proportion of participants retained in the study model of care (study arm) after 12 months, defined as participants alive and continuously receiving ART at 3- or 6-monthly intervals in the same study arm as at enrollment in the study. The primary outcome of retention in ART care considered both death and loss to follow-up (LTFU) as attrition. The secondary outcome of retention in the study arm considered all participants who died, were LTFU, who transferred out to other facilities, and those transitioning off the arm due to needing more frequent dispensing of ART for clinical reasons as losses to the arm. Participants who missed a pick-up date for their ART medication for more than 90 days after the last missed appointment and who were not known to have transferred out to another facility or service or died were considered LTFU. Follow-up continued until 15 months to calculate outcomes at 12 months, allowing participants not returning for the 12- month visit 3 additional months before being classified as LTFU.15 Statistical Analysis Sample Size Estimation Participant sample size estimates were calculated for the primary outcome of retention in ART care 12 months after enrollment, for a noninferiority test for the difference in 2 proportions in a cluster-randomized design using PASS v.14 software. Participant enrollment numbers were assumed to be equal in each cluster. The probability of participant retention 12 months after study enrollment in the control group was assumed to be 95%. The probability of retention in the intervention arms after 12 months was assumed to be 95%. An intracluster correlation coefficient of 0.01 for retention amongst stable ART patients associated with the same healthcare facility was assumed.18 The noninferiority margin was prespecified as 23.25% [risk difference (RD)]. Assum- ing a = 0.05, power of 85%, and using the 1-sided Z-test (unpooled) statistic, the cluster sample size target was 192 enrolled participants per facility, with 1920 participants per arm and a total sample size of 5760. As retention in care was the primary outcome, no adjustment for LTFU was made. Data Analysis Descriptive measures of the study population at base- line in each study arm were evaluated using medians and interquartile ranges for continuous variables and proportions for categorical variables. Individual-level outcome analyses were conducted by intention-to-treat (ITT) including all enrolled participants in the arm to which they were originally allocated. For retention in ART care and retention in the study arm, RDs were estimated using binomial population-averaged generalized estimating equations using an identity link and an exchangeable correlation structure, specifying for clustering by facility and using robust standard errors.19,20 A small cluster size variance correction was used,20 and randomiza- tion strata (urban/rural location) were included in all models as a fixed-effect parameter. Mortality and LTFU (as separate outcomes) were also analyzed by ITT using RDs estimated from binomial population-averaged generalized estimating equation models. Multivariable regression analyses were performed including individual-level and cluster-level cova- riates that displayed imbalance at baseline and were associ- ated with the outcome in unadjusted analyses. Two prespecified VL suppression analyses were con- ducted as follows: (1) ITT analysis including all enrolled participants as allocated and irrespective of whether they had available follow-up VL results or completed the study, and (2) an analysis restricted to participants having an available VL result 12 months after study enrollment. VL results performed between 9 and 15 months after study enrollment were considered as 12-month VL results. Where participants had more than 1 VL result during this period, the result closest to 12 months was analyzed. For the ITT analysis of VS, a 3-level outcome variable was generated: VL suppressed, VL unsuppressed, and VL test not performed. Using a generalized structural equation framework, a multinomial logit model specified for clustering by facility was constructed to estimate the intervention effect of a suppressed vs. unsuppressed VL, specifying unsup- pressed VL as the base category. For the analysis restricted to participants with available VL results, log-binomial regres- sion with generalized estimating equations was used to estimate risk ratios of VS between study arms, specifying for clustering by facility. All regression models were adjusted for the randomization stratum. Statistical analyses were conducted using Stata 15. A Data Safety and Management Board oversaw the study (ClinicalTrials.gov Identifier NCT03438370). RESULTS From August 3, 2017, to April 30, 2018, a total of 3783, 2138, and 4636 ART patients were screened for eligibility, with 1898, 1558, and 1880 participants were enrolled in arms 3MF, 3MC, and 6MCD, respectively (Fig. 1). The most common reason for exclusion was nonavailabil- ity of viral load results (47% of those excluded), followed by having a viral load .1000 copies/mL (21% of those excluded). Enrollment in 6MCD was slightly slower than the other arms (median month of enrollment = November Community-Based Differentiated ModelsJ Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. www.jaids.com | 283 Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. http://ClinicalTrials.gov TABLE 1. Characteristics of Participants at Enrollment in the Multimonth Scripting and Dispensing of ART Cluster-Randomized Trial in Lesotho Baseline Characteristic Arm 3MF (N = 1898) Arm 3MC (N = 1558) Arm 6MCD (N = 1880) Age, yr, median (IQR) (n = 5336) 42.7 (34.7–54.0) 48.4 (39.4–57.8) 41.2 (33.3–51.4) Age categories, n (%) (n = 5336) 18–24 yrs 65 (3.4) 22 (1.4) 88 (4.7) 25–49 yrs 1223 (64.4) 821 (52.7) 1280 (68.1) $50 yrs 610 (32.1) 715 (45.9) 512 (27.2) Gender, n (%) (n = 5336) Female 1158 (61.0) 1117 (71.7) 1264 (67.2) Male 740 (39.0) 441 (28.3) 616 (32.8) CD4 cell count at enrollment, cells/mL, median (IQR) (n = 3388) 515 (365.0–698.0) 560 (396.0–743.0) 535.5 (365.0–726.5) CD4 cell count categories, n (%) ,200 cells/mL 78 (4.1) 39 (2.5) 58 (3.1) ,200–499 cells/mL 546 (28.8) 415 (26.6) 366 (19.5) 500–749 cells/mL 429 (22.6) 396 (25.4) 314 (16.7) 750–3000 cells/mL 270 (14.2) 267 (17.1) 210 (11.2) Not recorded 575 (30.3) 441 (28.3) 932 (49.6) Weight, kg, median (IQR) (n = 5289) 60 (53.2–68.7) 59 (52.0–69.8) 61.5 (54.0–71.0) WHO stage, n (%) (n = 5285) I 1841 (97.7) 1505 (98.2) 1809 (96.8) II 43 (2.3) 28 (1.8) 59 (3.2) Marital status, n (%) (n = 5327) Married 1051 (55.5) 726 (46.7) 958 (51.0) Not married* 843 (44.5) 828 (53.3) 921 (49.0) Current employment, n (%) (n = 5325)† Employed 676 (35.7) 464 (29.9) 877 (46.7) Not employed 1218 (64.3) 1088 (70.1) 1002 (53.3) Smoking, n (%) (n = 5316) Never smoked 1631 (86.1) 1412 (91.3) 1669 (89.0) Current or previous smoker 263 (13.9) 135 (8.7) 206 (11.0) Current alcohol consumption, n (%) (n = 5320) No 1447 (76.4) 1175 (75.9) 1511 (80.5) Yes 446 (23.6) 374 (24.1) 367 (19.5) Distance to facility, n (%) (n = 5317) ,4 km 496 (26.2) 515 (33.3) 455 (24.2) 4–9 km 542 (28.6) 550 (35.6) 622 (33.1) .9 km 527 (27.8) 346 (22.4) 405 (21.6) Unknown 329 (17.4) 135 (8.7) 395 (21.0) Time from ART initiation until study enrollment, mo, median (IQR) (n = 5464) 44.3 (20.1–86.3) 66.1 (31.8–103.6) 38.3 (18–77) Time from HIV diagnosis till ART initiation, mo, median (IQR) (n = 4899) 1.5 (0.3–11.8) 2.3 (0.6–15.3) 1.2 (0–10.5) Disclosed HIV status, n (%) (n = 5327) Yes 1790 (94.5) 1428 (92.0) 1790 (95.2) No 104 (5.5) 125 (8.1) 90 (4.8) Facility location, n (%) (n = 5336) Rural 1443 (76.0) 1125 (72.2) 1443 (76.8) Urban 455 (24.0) 433 (27.8) 437 (23.2) Health care level, n (%) (n = 5336) Clinic 1443 (76.0) 1277 (82.0) 1665 (88.6) Hospital 455 (24.0) 281 (18.0) 215 (11.4) District, n (%) (n = 5336) Maseru 1225 (64.5) 586 (37.6) 444 (23.6) Mafeteng 280 (14.8) 671 (43.0) 1049 (55.8) Mohale’s Hoek 393 (20.7) 301 (19.3) 387 (20.6) *Includes those never married, divorced, or widowed. †Employment status was determined by self-report and recorded as a binary variable. Self-employment was considered employed. Tukei et al J Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 284 | www.jaids.com Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. TABLE 2. Analysis of Primary Outcome (Patient Retention) After 12 Months by Intention-to-Treat in the Multimonth Dispensing of ART Cluster-Randomized Trial in Lesotho* Baseline Factor Enrolled (N) Retained (n) Retained (%) Unadjusted Analysis Adjusted Analysis† Risk Difference 95% CI P Risk Difference 95% CI P Arm (vs. 3MF) 3MF (control) 1898 1842 97.1 Reference — — Reference — — 3MC 1558 1504 96.5 20.2% 21.9 to 1.6 0.845 20.1% 21.6 to 1.5 0.924 6MCD 1880 1781 94.7 22.3% 24.1 to 20.4 0.016 21.3% 23.0 to 0.5 0.149 6MCD vs. 3MC 3MC 1558 1504 96.5 Reference — — Reference — — 6MCD 1880 1781 94.7 22.1% 23.9 to 20.3 0.023 21.2% 22.9 to 0.5 0.168 Gender Male 1797 1733 96.4 0.4% 20.8 to 1.7 0.508 Female 3539 3394 95.9 Reference — — Age category 18–24 yrs 175 159 90.9 25.2% 29.3 to 21.1 0.013 24.7% 28.6 to 20.7 0.020 25–49 yrs 3324 3196 96.2 Reference — — Reference — — $50 yrs 1837 1772 96.5 0.3% 20.8 to 1.4 0.621 0.4 20.8 to 1.6 0.501 CD4 cell count ,200 cells/mL 175 168 96.0 20.8% 24.0 to 2.4 0.624 ,200–499 cells/mL 1327 1287 97.0 Reference — — 500–749 cells/mL 1139 1101 97.0 20.2% 21.8 to 1.4 0.781 $750 cells/mL 747 719 96.3 20.7% 22.7 to 1.4 0.526 Not recorded 1948 1852 95.1 21.7% 23.6 to 0.1 0.073 WHO stage I 5155 4958 96.2 Reference — — II 130 124 95.4 20.7% 24.3 to 2.9 0.700 Marital status Married 2735 2633 96.3 Reference — — Not married 2592 2486 96.0 20.3% 21.2 to 0.6 0.497 Current employment Employed 2017 1940 96.2 Reference — — Not employed 3308 3117 96.0 20.4% 21.5 to 0.7 0.477 Smoking Never smoked 4712 585 96.9 Reference — — Current/previous smoker 604 4525 96.0 1.0% 20.7 to 2.8 0.253 Alcohol consumption No 4133 3965 95.9 Reference — — Yes 1187 1147 96.6 0.8% 20.4 to 2.1 0.175 Distance to facility ,4 km 1466 1409 96.1 Reference — — 4–9 km 1714 1640 95.7 0.0% 22.0 to 1.9 0.974 .9 km 1278 1229 96.2 0.7% 21.4 to 2.7 0.524 Unknown 859 831 96.7 0.8% 22.4 to 3.9 0.629 Disclosed HIV status Yes 5008 4811 96.1 Reference — — No 319 308 96.6 0.7% 21.4 to 2.7 0.525 Facility location Urban 1325 1262 95.3 21.0% 23.1 to 1.2 0.377 21.0 22.8 to 0.7 0.240 Rural 4011 3865 96.4 Reference — — Reference — — Health care level Clinic 4385 4227 96.4 Reference — — Hospital 951 900 94.6 22.3% 26.0 to 1.9 0.219 (continued on next page) Community-Based Differentiated ModelsJ Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. www.jaids.com | 285 Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. 2017 vs. October 2017 in arms 3MF and 3MC). Participants’ follow-up was closed on July 31, 2019. At baseline, imbalance between the arms was apparent with respect to age (3MC participants were older, and 6MCD had a higher proportion of participants aged 18–24 years); 3MF participants were more likely to be enrolled at hospital- based facilities; and variation by district was apparent (3MF participants were less likely to be from Mafeteng) (Table 1). The median duration of follow-up (from study enroll- ment to the last ART receipt) was 17.7 months [interquartile range (IQR): 15.1–19.1], 15.3 months (IQR: 12.1–18.2), and 12.4 months (IQR: 11.8–15.7) in arms 3MF, 3MC, and 6MCD, respectively. Enrollment in the 6MCD arm was slower compared with the other arms because of structural challenges and thus follow-up time was lower in this arm. Twelve months after enrollment, 1842 (97.1%), 1504 (96.5%), and 1781 (94.7%) participants enrolled in 3MF, 3MC, and 6MCD remained in ART care, respectively (Table 2). The measured intracluster correlation coefficient for retention in ART care was 0.01. Retention in 3MC did not TABLE 2. (Continued ) Analysis of Primary Outcome (Patient Retention) After 12 Months by Intention-to-Treat in the Multimonth Dispensing of ART Cluster-Randomized Trial in Lesotho* Baseline Factor Enrolled (N) Retained (n) Retained (%) Unadjusted Analysis Adjusted Analysis† Risk Difference 95% CI P Risk Difference 95% CI P District Maseru 2255 2195 97.3 Reference — — Reference — — Mafeteng 2000 1894 94.7 22.5% 24.4 to 20.7 0.007 22.1% 23.5 to 20.6 0.006 Mohale’s Hoek 1081 1038 96.0 21.2% 23.3 to 0.8 0.241 21.1% 22.4 to 0.3 0.127 The measured intracluster correlation coefficient for retention was 0.01. *Outcome analyses were by intention-to-treat using population-averaged generalized estimating equations specified for clustering by facility and using robust standard errors. All models were adjusted for randomization stratum (rural/urban). †Estimates adjusted for age category, district, and randomization stratum. WHO, World Health Organization. TABLE 3. Analysis of Secondary Outcome (Retention in the Study Arm) After 12 Months by Intention-to-Treat in the Multimonth Dispensing of ART Cluster-Randomized Trial in Lesotho* Baseline Factor Enrolled (N) Retained (n) Retained (%) Unadjusted Analysis Adjusted Analysis† Risk Difference 95% CI P Risk Difference 95% CI P Arm (vs. 3MF) 3MF (control) 1898 1802 94.9% Reference — — Reference — — 3MC 1558 1487 95.4% 0.9% 21.3 to 3.1 0.445 1.1% 20.6 to 2.8 0.192 6MCD 1880 1754 93.3% 21.6% 23.6 to 0.3 0.093 20.6% 22.4 to 1.1 0.495 6MCD vs. 3MC 3MC 1558 1487 95.4% Reference — — Reference — — 6MCD 1880 1754 93.3% 22.5% 24.4 to 20.5 0.014 21.9% 23.6 to 20.2 0.032 Gender Male 1797 1715 95.4% 1.3% 20.2 to 2.9 0.088 Female 3539 3328 94.0% Reference — — Age 18–24 yrs 175 149 85.1% 29.4% 214.5 to 24.3 ,0.0001 29.0% 214.0 to 24.0 ,0.0001 25–49 yrs 3324 3143 94.6% Reference — — Reference — — $50 yrs 1837 1751 95.3% 0.7% 20.7 to 2.1 0.309 0.6% 20.9 to 2.1 0.438 CD4 cell count ,200 cells/mL 175 164 93.7% 21.2% 25.3 to 2.9 0.565 ,200–499 cells/mL 1327 1263 95.2% Reference — — 500–749 cells/mL 1139 1085 95.3% 0.1% 21.4 to 1.6 0.866 $750 cells/mL 747 708 94.8% 20.3% 22.4 to 1.8 0.762 Not recorded 1948 1823 93.6% 21.5% 23.3 to 0.4 0.115 WHO stage I 5155 4877 94.6% Reference — — II 130 121 93.1% 21.6% 26.6 to 3.5 0.539 Tukei et al J Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 286 | www.jaids.com Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. differ compared with control 3MF in both unadjusted and adjusted analyses, and the noninferiority limit was achieved, adjusted RD = 20.1% [95% confidence interval (CI): 21.6% to 1.5%]. Retention in 6MCD was reduced compared with control 3MF and 3MC in unadjusted analyses, RD = 22.3% (95% CI: 24.1% to 20.4%) and RD = 22.1% (95% CI: 23.9% to 20.3%), respectively. However, 6MCD had a higher proportion of participants aged 18–24 years and a higher proportion from Mafeteng district, and both factors were significantly associated with lower retention. After adjustment for age category and district, the noninferiority limit was achieved for both 6MCD vs. control 3MF and 6MCD vs. 3MC, adjusted RD = 21.3% (95% CI: 23.0% to 0.5%) and adjusted RD = 21.2% (95% CI: 22.9% to 0.5%), respectively. After 12 months, 1802 (94.9%), 1487 (95.4%), and 1754 (93.3%) participants enrolled to 3MF, 3MC, and 6MCD continued to receive ART in their original arm, respectively (Table 3). The number of participants who transitioned off the arms due to needing increased fre- quency of ART receipt was low and similar between arms, 19 (1.0%), 14 (0.9%), and 13 (0.7%) in 3MF, 3MC, and 6MCD, respectively. In both unadjusted and adjusted analyses, retention in the arm did not differ between 3MC vs. control 3MF and also between 6MCD vs. control 3MF. However, retention in the study arm was lower in 6MCD vs. 3MC in the unadjusted analysis, RD = 22.5% (95% CI: 24.4% to 20.5%), and remained lower after adjusting for baseline imbalance, adjusted RD = 21.9% (95% CI: 23.6% to 20.2%). TABLE 3. (Continued ) Analysis of Secondary Outcome (Retention in the Study Arm) After 12 Months by Intention-to-Treat in the Multimonth Dispensing of ART Cluster-Randomized Trial in Lesotho* Baseline Factor Enrolled (N) Retained (n) Retained (%) Unadjusted Analysis Adjusted Analysis† Risk Difference 95% CI P Risk Difference 95% CI P Marital status Married 2735 2587 94.4% Reference — — Not married 2592 2448 94.6% 20.1% 21.1 to 0.9 0.812 Current employment Employed 2017 1910 94.7% Reference — — Not employed 3308 3123 94.4% 20.5% 21.8 to 0.7 0.415 Smoking Never smoked 4712 4447 94.4% Reference — — Current/previous smoker 604 579 95.9% 1.7% 20.3 to 3.6 0.102 Alcohol consumption No 4133 3899 94.3% Reference — — Yes 1187 1130 95.2% 1.0% 20.6 to 2.5 0.223 Distance to facility ,4 km 1466 1387 94.6% Reference 4–9 km 1714 1612 94.1% 20.2% 22.4 to 2.0 0.879 .9 km 1278 1206 94.4% 0.3% 22.0 to 2.5 0.820 Unknown 859 820 95.5% 1.0% 22.5 to 4.5 0.580 Disclosed HIV status Yes 5008 4730 94.5% Reference — — No 319 305 95.6% 1.2% 21.1 to 3.6 0.309 Facility location Urban 1325 1239 93.5% 21.2% 23.6 to 1.2 0.320 21.3% 23.4 to 0.7 0.202 Rural 4011 3804 94.9% Reference — — Reference — — Health care level Clinic 4385 4161 94.9% Reference — — Hospital 951 882 92.7% 23.0% 26.8 to 0.9 0.138 District Maseru 2255 2167 96.1% Reference — — Reference — — Mafeteng 2000 1868 93.4% 22.9% 24.5 to 21.3 ,0.0001 22.6% 24.1 to 21.0 0.001 Mohale’s Hoek 1081 1008 93.3% 22.7% 24.7 to 20.7 0.009 22.5% 24.2 to 20.8 0.004 *Outcome analyses were by intention-to-treat using population-averaged generalized estimating equations specified for clustering by facility and using robust standard errors. All models were adjusted for randomization stratum (rural/urban). †Estimates adjusted for age category, district, and randomization stratum. WHO, World Health Organization. Community-Based Differentiated ModelsJ Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. www.jaids.com | 287 Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. TABLE 4. Viral Suppression 12 Months After Enrollment Enrolled (N) VL Due* (N) VL Done (N) VL Suppressed (n) VL Completion† VS (ITT) (%)‡ VS Among Those with VL Results§ VS by ITT (N = 5336)║ VS Among Those With VL Results (N = 3914)#Unadjusted Adjusted¶ RRR 95% CI P RRR 95% CI P RR 95% CI P Arm 3MF 1898 1821 1503 1482 82.5% 78.1% 98.6% Ref — — Ref — — Ref — — 3MC 1558 1501 1126 1104 75.0% 70.9% 98.1% 0.70 0.35–1.41 0.316 1.15 0.54–2.47 0.707 1.00 0.98–1.01 0.449 6MCD 1880 1767 1285 1263 72.7% 67.2% 98.3% 0.84 0.36–1.93 0.676 1.30 0.57–2.93 0.525 1.00 0.98–1.01 0.729 6MCD vs. 3MC 1.19 0.47–3.03 0.710 1.12 0.47–2.65 0.788 1.00 0.99–1.02 0.776 Facility location Urban 1325 1250 982 966 78.6% 72.9% 98.4% Ref — — Ref — — Ref — — Rural 4011 3839 2932 2883 76.4% 71.9% 98.3% 0.98 0.40–2.37 0.956 1.57 0.79–3.10 0.191 1.00 0.99–1.02 0.806 District Maseru 2255 2175 1797 1777 82.6% 78.8% 98.9% Ref — — Ref — — Ref — — Mafeteng 2000 1890 1343 1304 71.1% 65.2% 97.1% 0.37 0.18–0.72 0.004 0.30 0.17–0.55 ,0.0001 0.98 0.97–0.99 0.002 Mohale’s Hoek 1081 1024 774 768 75.6% 71.0% 99.2% 1.49 0.65–3.41 0.341 1.50 0.56–3.96 0.82 1.00 1.00–1.01 0.395 *Enrolled less died, LTFU, and TFO. †Viral load done (N)/viral load due (N). ‡Viral load suppressed (n)/enrolled (N). §Viral load suppressed (n)/VL done (N). ║Estimates from a multinomial logit regression model specifying for clustering by facility and using unsuppressed viral load as the base category, adjusted for randomization stratum (urban/rural). ¶Adjusted for district and randomization stratum. #Estimates using log-binomial regression with generalized estimating equations specifying for clustering by facility and adjusted for randomization stratum, including only participants with available viral load results. RRR, relative risk ratio; RR, risk ratio; TFO, transferred out. T ukei et al J A cq u ir Im m u n e D efi c S yn d r � V o lu m e 8 5 , N u m b e r 3 , N o ve m b e r 1 , 2 0 2 0 288 | w w w .ja id s.co m C opyright © 2020 W olters K luw er H ealth, Inc. A ll rights reserved. C opyright © 2020 W olters K luw er H ealth, Inc. U nauthorized reproduction of this article is prohibited. After 12 months, viral load completion was 1503 (82.5%), 1126 (75.0%), and 1285 (72.7%) in arms 3MF, 3MC, and 6MCD, respectively (Table 4). VS by ITT in 3MC and 6MCD was not different from 3MF; adjusted relative risk ratio (aRRR) = 1.30 (95% CI: 0.57 to 2.93) and aRRR = 1.12 (95% CI: 0.47 to 2.65), respectively. Adjusted models were adjusted for the randomization stratum and study district. Among participants with available viral load results after 12 months, 1482 (98.6%), 1104 (98.1%), and 1263 (98.3%) were virally suppressed in arms 3MF, 3MC, and 6MCD, respec- tively. No differences in viral suppression were found between any arms in analyses among participants with available viral load results (Table 4). After 12 months, 7 (0.4%), 12 (0.8%), and 18 (1.0%) participants were recorded as having died in 3MF, 3MC, and 6MCD, respectively (Table 5). A further 49 (2.6%), 42 (2.7%), and 81 (4.3%) were LTFU in these same arms, respectively. The proportion of participants recorded as having died among those LTFU was slightly lower in 3MF than that in 3MC and 6MCD: 7 of 56 (12.5%), 12 of 54 (22.2%), and 18 of 99 (18.2%), respectively (P = 0.23). There was a borderline increase of mortality in 6MCD vs. control 3MF in the unadjusted analysis, RD = 0.5% (95% CI: 0.0% to 1.0%; P = 0.07). However, the difference in risk was small and was not significantly different at the 5% level both before and after adjustment, adjusted RD = 0.5% (95% CI: 20.1% to 1.0%; P = 0.07). Mortality in 6MCD was not higher when compared with 3MC. There were no differences in LTFU between the intervention arms vs. control. LTFU in 6MCD was higher compared with 3MC in the unadjusted analysis, but there was no difference after adjustment for age and district, adjusted RD = 0.6% (95% CI: 21.0% to 2.2%). Viral suppression was high (.98%) with no differences between arms. DISCUSSION This study is one of the first cluster-randomized trials to evaluate the outcomes of extended dispensing intervals of TABLE 5. Mortality and Loss to Follow-up 12 Months After Enrollment* Arm Enrolled (N) Died, n (%) LTFU, n (%) Mortality Loss to Follow-up Unadjusted Adjusted† Unadjusted Adjusted‡ RD 95% CI P RD 95% CI P RD 95% CI P RD 95% CI P 3MF 1898 7 (0.4) 49 (2.6) Ref. — — Ref. — — Ref. — — Ref. — — 3MC 1558 12 (0.8) 42 (2.7) 0.4% 20.1 to 0.8 0.090 0.3% 20.1 to 0.6 0.162 20.3% 22.0 to 1.4 0.702 20.2% 21.4 to 1.0 0.739 6MCD 1880 18 (1.0) 81 (4.3) 0.5% 0.0 to 1.0 0.070 0.5% 20.1 to 1.0 0.079 1.8% 20.3 to 3.8 0.099 0.4% 21.3 to 2.1 0.650 6MCD vs. 3MC 0.1% 20.5 to 0.7 0.763 0.2% 20.4 to 0.8 0.471 2.1% 0.1 to 4.1 0.037 0.6% 21.0 to 2.2 0.479 *Outcome analyses were by intention-to-treat using population-averaged generalized estimating equations specified for clustering by facility and using robust standard errors. All models were adjusted for randomization stratum (urban/rural location). †Estimates adjusted for age category, gender, and urban/rural location. ‡Estimates adjusted for age category, district, and urban/rural location. Ref, reference category. FIGURE 2. Study design. Community-Based Differentiated ModelsJ Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. www.jaids.com | 289 Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. ART using community-based distribution in sub-Saharan Africa. This study showed the community-based models to have similar outcomes of retention and viral load suppression as the standard facility-based model, with no significant differences between the arms by adjusted ITT analyses. The results are congruent with a similar study, which was undertaken in Zimbabwe.21 It is anticipated that the interven- tions will be cost-effective, as costs are expected to be reduced with extended intervals of community-based distri- bution vs. facility-based ART dispensing. Community-based differentiated models of ART deliv- ery (DMAD) in low-income countries were primarily de- signed to adopt HIV services to client needs in order to minimize barriers to adherence and retention on ART by bringing the services closer to the client and to reduce the rapidly increasing burden on the health systems caused by increasing numbers of patients initiating ART.6,12,14,22–26 Various observational studies have suggested that DMAD may offer other benefits beyond system and patient cost savings such as improved patient outcomes including lower death rates and LTFU, improved retention, improved health systems’ operational efficiency, improved accessibility of services, shorter patient waiting times, patients’ social support, and patient and provider satisfaction.4,19,27–34 Studies further suggest that when a health system adopts client- centered DMAD, the limited facility-based resources are focused on the populations most at risk of adverse outcomes, thus providing better access to improved quality of care and treatment in a more effective way.16,35 Studies have discussed the benefits of DMAD but have not rigorously investigated potential differences in patient outcomes between these models, with a lack of randomized studies to adequately assess the effects of DMAD within community ART dispensing models. Our study adds important public health evidence to the findings of other studies by demonstrating that community DMAD did not compromise the outcomes of retention and viral suppression for stable HIV patients. In our study, nonavailability of viral load results was the most common reason for noneligibility for the study (Fig. 1). This was due to prolonged turnaround time of about 3 months. For a successful MMSD program, we recommend strengthening of the laboratory services to reduce viral load turnaround times and improve access to timely viral load results. In addition, as our study found similar outcomes among longer ART dispensing intervals (6 vs. 3 months), we recommend national policy change to allow stable HIV patients, including migrant populations, access to longer ART dispensing intervals (6 months) to minimize barriers to adherence and retention in care (Fig. 2). The study limitation includes lower recruitment in 3MC than other arms because of limited availability of potentially eligible patients in the clusters randomized to the arm, and thus, study power for comparisons involving 3MC was lower than anticipated. As cluster allocation was not stratified by district, baseline imbalance with respect to district was apparent. Although baseline imbalance was controlled for in adjusted analyses, adjustment may not have fully corrected for baseline differences. Participant outcomes beyond 12 months after enrollment were not assessed as follow-up closed at 12 months. Caution should be taken when generalizing interpretation of our findings to other settings because there are varying definitions of stable patients. The sample of participants aged 18–24 years in our study was small, and thus, overall results may not be generalizable to this age group. In our study and similar to previous studies, retention among youth was low, and DSD models may need to be better tailored to this age group to achieve optimal outcomes.36 Despite these limitations, this study’s robust design is the first such study in Lesotho to explore the outcomes of retention and viral suppression among community-based MMSD models of ART. CONCLUSIONS The study shows that it is feasible to implement community-based differentiated models of MMSD of ART at 3- and 6-month intervals outside the standard facility model with high retention, minimum loss to follow-up, and high viral load suppression. Further evaluations should include longer participant follow-up to ascertain longer-term out- comes of community-based MMSD of ART. ACKNOWLEDGMENTS The authors acknowledge the Lesotho Ministry of Health and Christian Health Association of Lesotho (CHAL), the PEPFAR, USAID EQUIP Health number AID-OAA-A- 15-00070USAID/PEPFAR, the Elizabeth Glaser Pediatric AIDS Foundation (EGPAF), Lesotho Network of AIDS Services Organizations (LENASO), patients, and health facility staff. REFERENCES 1. WHO guidelines approved by the guidelines review committee. In: Guideline on when to Start Antiretroviral Therapy and on Pre-exposure Prophylaxis for HIV. Geneva, Switzerland: World Health Organization Copyright © World Health Organization 2015; 2015. 2. Loeliger KB, Niccolai LM, Mtungwa LN, et al. Antiretroviral therapy initiation and adherence in rural South Africa: community health workers’ perspectives on barriers and facilitators. AIDS Care. 2016;28: 982–993. 3. Tiruneh YM, Galárraga O, Genberg B, et al. Retention in care among HIV-infected adults in Ethiopia, 2005- 2011: a mixed-methods study. PLoS One. 2016;11:e0156619. 4. Prust ML, Banda CK, Callahan K, et al. Patient and health worker experiences of differentiated models of care for stable HIV patients in Malawi: a qualitative study. PLoS One. 2018;13:e0196498. 5. Organization WH. Retention in HIV programmes: defining the chal- lenges and identifying solution: meeting report, 13–15 September, 2011. 2011. 6. Tuller DM, Bangsberg DR, Senkungu J, et al. Transportation costs impede sustained adherence and access to HAART in a clinic population in southwestern Uganda: a qualitative study. AIDS Behav. 2010;14: 778–784. 7. Amanyire G, Wanyenze R, Alamo S, et al. Client and provider perspectives of the efficiency and quality of care in the context of rapid scale-up of antiretroviral therapy. AIDS Patient Care STDS. 2010;24: 719–727. 8. Fay H, Baral SD, Trapence G, et al. Stigma, health care access, and HIV knowledge among men who have sex with men in Malawi, Namibia, and Botswana. AIDS Behav. 2011;15:1088–1097. 9. Teck R. Reducing frequency of ARV pick-ups for HIV patients stabilised on ARV treatment. 2015. Paper presented at: 18th International Tukei et al J Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 290 | www.jaids.com Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited. Conference on AIDS and STI in Africa; November 29 to December 4, 2015; Harare, Zimbabwe. 10. Lesotho Go. National Guidelines on the Use of Antiretroviral Therapy for HIV Prevention and Treatment. Maseru, Lesotho: 2016. 11. Sharp J, Wilkinson L, Cox V, et al. Outcomes of patients enrolled in an antiretroviral adherence club with recent viral suppression after experi- encing elevated viral loads. South Afr J HIV Med. 2019;20:905. 12. Wilkinson L, Harley B, Sharp J, et al. Expansion of the Adherence Club model for stable antiretroviral therapy patients in the Cape Metro, South Africa 2011–2015. Trop Med Int Health. 2016;21:743–749. 13. Faturiyele IO, Appolinare T, Ngorima-Mabhena N, et al. Outcomes of community-based differentiated models of multi-month dispensing of antiretroviral medication among stable HIV-infected patients in Lesotho: a cluster randomised non-inferiority trial protocol. BMC Public Health. 2018;18:1069. 14. Rasschaert F, Decroo T, Remartinez D, et al. Sustainability of a community-based anti-retroviral care delivery model—a qualitative research study in Tete, Mozambique. J Int AIDS Soc. 2014;17: 18910. 15. Grimsrud AT, Cornell M, Egger M, et al. Impact of definitions of loss to follow-up (LTFU) in antiretroviral therapy program evaluation: variation in the definition can have an appreciable impact on estimated proportions of LTFU. J Clin Epidemiol. 2013;66:1006–1013. 16. Lesotho Go. Lesotho CARG tool kit. Maseru, Lesotho: Health Mo; 2015. 17. Rasschaert F, Decroo T, Remartinez D, et al. Adapting a community- based ART delivery model to the patients’ needs: a mixed methods research in Tete, Mozambique. BMC Public Health. 2014;14:364. 18. Fairall L, Bachmann MO, Lombard C, et al. Task shifting of antiretroviral treatment from doctors to primary-care nurses in South Africa (STRETCH): a pragmatic, parallel, cluster-randomised trial. Lancet. 2012;380:889–898. 19. Pedroza C, Truong VT. Performance of models for estimating absolute risk difference in multicenter trials with binary outcome. BMC Med Res Methodol. 2016;16:113. 20. Huang S, Fiero MH, Bell ML. Generalized estimating equations in cluster randomized trials with a small number of clusters: review of practice and simulation study. Clin Trials. 2016;13:445–449. 21. Fatti G, Ngorima-Mabhena N, Mothibi E, et al. Outcomes of three- versus six-monthly dispensing of antiretroviral treatment (ART) for stable HIV patients in community ART refill groups: a cluster-randomized trial in Zimbabwe. J Acquir Immune Defic Syndr. 2020;84:162–172. 22. Duncombe C, Rosenblum S, Hellmann N, et al. Reframing HIV care: putting people at the centre of antiretroviral delivery. Trop Med Int Health. 2015;20:430–447. 23. Grimsrud A, Bygrave H, Doherty M, et al. Reimagining HIV service delivery: the role of differentiated care from prevention to suppression. J Int AIDS Soc. 2016;19:21484. 24. Bemelmans M, Baert S, Goemaere E, et al. Community-supported models of care for people on HIV treatment in sub-Saharan Africa. Trop Med Int Health. 2014;19:968–977. 25. Roy M, Bolton Moore C, Sikazwe I, et al. A review of differentiated service delivery for HIV treatment: effectiveness, mechanisms, targeting, and scale. Curr HIV/AIDS Rep. 2019;16:324–334. 26. Decroo T, Koole O, Remartinez D, et al. Four-year retention and risk factors for attrition among members of community ART groups in Tete, Mozambique. Trop Med Int Health. 2014;19:514–521. 27. Prust ML, Banda CK, Nyirenda R, 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(suppl 4):21650. 28. Myer L, Iyun V, Zerbe A, et al. Differentiated models of care for postpartum women on antiretroviral therapy in Cape Town, South Africa: a cohort study. J Int AIDS Soc. 2017;20(suppl 4):21636. 29. Phillips A, Shroufi A, Vojnov L, et al. Sustainable HIV treatment in Africa through viral-load-informed differentiated care. Nature. 2015;528:S68–S76. 30. Siapka M, Remme M, Obure CD, et al. Is there scope for cost savings and efficiency gains in HIV services? A systematic review of the evidence from low- and middle-income countries. Bull World Health Organ. 2014;92:499–511ad. 31. Tagar E, Sundaram M, Condliffe K, et al. Multi-country analysis of treatment costs for HIV/AIDS (MATCH): facility-level ART unit cost analysis in Ethiopia, Malawi, Rwanda, South Africa and Zambia. PLoS One. 2014;9:e108304. 32. Menzies NA, Berruti AA, Berzon R, et al. The cost of providing comprehensive HIV treatment in PEPFAR-supported programs. AIDS. 2011;25:1753–1760. 33. Ware NC, Wyatt MA, Geng EH, et al. Toward an understanding of disengagement from HIV treatment and care in sub-Saharan Africa: a qualitative study. PLoS Med. 2013;10:e1001369. discussion e1001369. 34. Okoboi S, Ding E, Persuad S, et al. Community-based ART distribution system can effectively facilitate long-term program retention and low-rates of death and virologic failure in rural Uganda. AIDS Res Ther. 2015;12:37. 35. Casale M, Carlqvist A, Cluver L. Recent interventions to improve retention in HIV care and adherence to antiretroviral treatment among adolescents and youth: a systematic review. AIDS Patient Care STDS. 2019;33:237–252. 36. Thin K, Frederix K, McCracken S, et al. Progress toward HIV epidemic control in Lesotho. AIDS. 2019;33:2393–2401. Community-Based Differentiated ModelsJ Acquir Immune Defic Syndr � Volume 85, Number 3, November 1, 2020 Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved. www.jaids.com | 291 Copyright © 2020 Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.