Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized © International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington DC 20433 Internet: www.worldbank.org; Telephone: 202 473 1000 This work is a product of the staff of The World Bank with external contributions. Note that The World Bank does not necessarily own each component of the content included in this work. The World Bank therefore does not warrant that the use of the content contained in the work will not infringe on the rights of third parties. The risk of claims resulting from such infringement rests solely with you. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or other partner institutions or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 Unported licence (CC BY 3.0) http://creativecommons.org/licences/by/3.0. Under the Creative Commons Attribution license, you are free to copy, distribute and adapt this work, including for commercial purposes, under the following conditions: Attribution – Please cite the work as follows: Evaluation of the National Adherence Guidelines for Chronic Diseases in South Africa Using Routinely Collected Data. Second Enrollment Report. Washington DC: World Bank. License: Creative Commons Attribution CC BY 3.0 Translations – If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in its translation. All queries on rights and licenses should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington DC, 20433, USA; fax: 202-522-2625; email: pubrights@worldbank.org. Evaluation of the Adherence Guidelines for Chronic Diseases in South Africa Using Routinely Collected Data SECOND ENROLLMENT REPORT 31 JANUARY 2017 National Department of Health authors: Mokgadi Phokojoe, Tshepo Malapo, Lillian Diseko, and Yogan Pillay National Health Laboratory Service authors: Sergio Carmona University of Witwatersrand authors: Sophie Pascoe, Amy Huber and Joshua Murphy World Bank authors: Nicole Fraser, Zara Shubber and Marelize Görgens Boston University authors: Matthew Fox and Sydney Rosen This page is for collation purposes. TABLE OF CONTENTS 4.1 General inclusion/Exclusion criteria............................................................................... 3 4.2 Cohort specific inclusion/exclusion criteria ................................................................ 3 6.1 Methods for identifying eligible individuals ................................................................ 6 6.2 Specific Procedures for Intervention Sites.................................................................... 7 6.3 Control sites ............................................................................................................................... 8 6.4 Cohort Specific Enrollment ................................................................................................. 8 6.5 Enrollment process/Identification of cohorts .......................................................... 13 6.6 Strategies to Overcome Barriers .................................................................................... 15 8.1 Timing of cohort initiation................................................................................................ 17 8.2 Enrollment by cohort .......................................................................................................... 18 8.3 Enrollment over time .......................................................................................................... 20 8.4 Enrollment by facility/district ........................................................................................ 22 8.5 Ineligible Patients ................................................................................................................. 23 v TABLE OF CONTENTS 10.1 Data collection for short-term endpoint for Fast Track Initiation Counselling .......................................................................................................................................... 32 FIGURES 1 Screening and enrollment by cohort over time compared to the target total through December 16th, 2016 ................................................................... 21 2 Enrollment over time by province and clinic .......................................................................... 22 3 Enrollment by District and Study Cohort.................................................................................. 23 4 Gantt Chart of Timeline for Project ............................................................................................. 34 TABLES 1 Population data (facility headcount and total active patients) at each facility and total numbers eligible by intervention (I) and control (C) for each intervention .................................................................................................... 5 2 Assumptions and sample sizes for each cohort ..................................................................... 10 3 Assumptions and sample sizes for each cohort ..................................................................... 16 4 Timing of cohort initiation by site and cohort ........................................................................ 17 5 Enrollment by cohort as of October 6th...................................................................................... 18 6 Ineligible subjects by cohort and reason for exclusion ....................................................... 24 7 Baseline characteristics of the cohorts ...................................................................................... 25 8 Adherence clubs .................................................................................................................................. 26 9 Decentralized Medicine Delivery (DMD) .................................................................................. 27 10 Enhanced Adherence Counselling (EAC) .................................................................................. 28 11 Early Tracing (TRIC).......................................................................................................................... 29 12 Tuberculosis, Hypertension and Diabetes - Screening Cohort (TBHD) ....................... 30 13 Tuberculosis, Hypertension and Diabetes - Diagnosed Cohort ....................................... 31 14 Study evaluation outcomes for protocol 1 ............................................................................... 32 15 Short-term outcomes (ART initiation within 30 days) for those eligible for FTIC cohort..................................................................................................................... 33 vi ACRONYMS AIDS Acquired Immune Deficiency Syndrome ART Antiretroviral treatment ARV Antiretroviral CBO Community based organization CRF Electronic case report form DMD Decentralized medicine delivery EAC Enhanced adherence counseling FSW Female sex worker FTIC Fast track initiation counseling HIV Human Immunodeficiency Virus NCD Non-communicable diseases NGO Non-Governmental Organization KP Key Population MSM Men who have sex with men NDoH National Department of Health NHLS National Health Laboratory Services PHC Primary health care TB Tuberculosis TBHD TB, Hypertension and Diabetes TRIC Tracing and retention in care TROA Total remaining on ART vii This page is for collation purposes. ACKNOWLEDGEMENTS This report is part of the impact evaluation led by the South African National Department of Health, in collaboration with the NDOH’s National Health Laboratory Service, World Bank, Boston University and the University of Witwatersrand. The impact evaluation has three Principal Investigators: Mokgadi Phokojoe (National Department of Health), Nicole Fraser (World Bank) and Matthew Fox (Boston University). The evaluation was funded by the Government of South Africa, the World Bank and the UK Department for International Development. The following individuals are part of the evaluation team and contributed to this report: National Department of Health: Mokgadi Phokojoe (Director: Care and Support), Tshepo Malapo (M&E Manager), Lillian Diseko (Deputy Director for HIV Care and Treatment) and Yogan Pillay (Deputy Director-General for Health) National Health Laboratory Service: Sergio Carmona (NHLS Johannesburg, and Department of Molecular Medicine and Haematology, Faculty of Health Sciences, University of the Witwatersrand, Johannesburg) World Bank: Nicole Fraser, Zara Shubber, Marelize Görgens Boston University: Matthew Fox and Sydney Rosen University of Witwatersrand (Health Economics and Epidemiology Research Office (HE2RO)): Sophie Pascoe, Amy Huber and Joshua Murphy The South Africa National Department of Health and the rest of the research team would like to thank all the health care managers and providers in the study provinces, districts and facilities for their participation and support in this evaluation. All graphics in the report were designed by Theo Hawkins, product designer and Knowledge Management Officer at the World Bank HIV team ix This page is for collation purposes. EXECUTIVE SUMMARY Now that enrollment has been completed for the HIV cohorts, this report describes the complete enrollment into the HIV cohorts for protocol 1 for the Evaluation of the National Adherence Guidelines for Chronic Diseases in South Africa Using Routinely Collected Data. The study is evaluating short-term and long-term effects of five interventions being implemented by the National Department of Health (NDoH) in South Africa to improve adherence to HIV care and chronic disease care in general: Fast track initiation counselling (FTIC), decentralized medicine delivery (DMD), adherence clubs (AC), early patient tracing (TRIC) and enhanced adherence counselling (EAC). The study uses a randomized evaluation design to compare health facilities where the intervention was rolled out with facilities where it was not. After completing data enhancement activities at each of the 24 sites (12 intervention and 12 control sites in 4 provinces in South Africa), the teams began enrollment into 7 study cohorts (5 HIV intervention evaluation cohorts and 2 TB, hypertension and Diabetes (TBHD) observational cohorts). Enrollment began on 20 June 2016 and was stopped as of 16 December 2016 for the HIV cohorts. Enrollment for the TBHD cohorts is ongoing (with expected completion by March 2017). While the team faced several challenges to completing enrollment the team developed approaches to overcoming the enrollment obstacles for nearly all cohorts by working with the World Bank and NDoH co-principal investigators. In addition to the barriers reported in the first enrollment report, the team encountered some additional barriers: 1) additional delays in rollout of the interventions; 2) an insufficient number of patients provided with the interventions at some sites; and 3) delays in implementation of the interventions which meant we could not continue to prospectively enroll subjects and still meet our endpoints within the appropriate time frame. To overcome these barriers and achieve the targeted sample size for all but one cohort (EAC), the team employed three main strategies: 1) increasing the duration of enrollment through nearly the end of December 2016; 2) overenrolling patients at some sites; and 3) shifting to including some retrospective enrollment, whereby sites with additional numbers of eligible patients for enrollment prior to December 2016 overenrolled patients. While these strategies mean we do not have even enrollment across sites, this allowed us to achieve or nearly achieve our total sample size in nearly all cohorts. The exception to this is the EAC cohort, where either not enough subjects were offered the intervention or the intervention was offered and not documented sufficiently to allow completion of enrollment within this cohort. xi EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA Passive follow up has now begun for patients in the HIV cohorts through monitoring data collection systems (TIER.net and NHLS) and patient files for study outcomes. Much of the work by the local team has now shifted to focus on protocol 2 where qualitative data is being collected to supplement the quantitative analyses that will result from protocol 1. xii BACKGROUND For antiretroviral therapy (ART) for HIV and treatments for other chronic diseases to be effective, patients must remain in care for longer periods of time, initiate treatment as early as allowed under prevailing guidelines, consistently achieve high levels of adherence to their treatment regimen and, as a result, exhibit low and stable monitoring test results and/or treatment completion. In the case of HIV, treatment is lifelong and requires consistent, nearly complete adherence to sustain an undetectable viral load. Numerous studies and reviews have indicated that retention in care and adherence to ART in South Africa are sub-optimal and pose a serious threat to the long-term success of the national HIV response. Although there is less evidence on hand, these same problems almost certainly also pertain to tuberculosis (TB), for which treatment completion and cure rates do not approach global targets and to non-communicable diseases (NCDs), for which almost no treatment adherence data are available. To address this challenge, in 2014 the NDOH developed the “National Adherence Guidelines for Chronic Diseases (HIV, TB and NCDs)” with rollout in 2015/2016. The guidelines address the provision of a minimum package of interventions to increase linkage to care, retention in care, and adherence to treatment. The minimum package of interventions includes five interventions that are being evaluated under this study: 1) fast track initiation counseling (FTIC); 2) enhanced adherence counseling for unstable patients (EAC); 3) adherence clubs (AC); 4) decentralized medicine delivery (DMD); and 5) early tracing of all missed appointments to improve retention in care (TRIC). The study was designed as a matched cluster randomized study in 24 clinics, 12 of which would receive early implementation of the minimum package and 12 would delay implementation and serve as control sites. Clinics were matched on clinic characteristics: total remaining on ART, clinic size, setting, location and viral suppression. This design was achieved for all of the interventions except for DMD, where a national decanting strategy meant that many of the control sites implemented DMD. STUDY AIM AND OBJECTIVES The overall aims of this study are to assess the impact of a subset of the National Adherence Guidelines’ (AGL) minimum package of interventions on HIV patients’ treatment outcomes at public sector clinics; estimate the costs of the interventions; and describe the cascade of care for TB, hypertension, and diabetes at these same clinics. The study has 8 specific aims which are detailed in the two study protocols entitled “Evaluation of the National Department of Health's National Adherence Guidelines for Chronic Diseases in South 1 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA Africa Using Routinely Collected Data” and “Process Evaluation of the National Department of Health's National Adherence Guidelines for Chronic Diseases in South Africa”. The objectives are: ► Among HIV-infected patients newly eligible for antiretroviral therapy, evaluate the impact of Fast Track Treatment Initiation Counselling on initiation and viral suppression. ► Among HIV-infected patients who are stable on antiretroviral therapy, evaluate the impact of Adherence Clubs on ART adherence and viral suppression. ► Among HIV-infected patients who are stable on antiretroviral therapy, evaluate the impact of decentralized medicine delivery on ART adherence and viral suppression. ► Among HIV-infected patients who have poor adherence to antiretroviral therapy, evaluate the impact of EAC on treatment adherence and viral suppression. ► Among HIV-infected patients in antiretroviral therapy programs who miss a scheduled appointment by 5 days or more, evaluate the impact of Early Patient Tracing on retention in care. ► For each clinic included in the study, evaluate the overall impact of the Adherence Guidelines on patient outcomes. ► Estimate the incremental and total cost of each of the interventions listed above compared to standard of care. ► Describe the current status of the cascade of care and adherence to treatment for tuberculosis, hypertension, and diabetes, for the purpose of tailoring the minimum package of interventions to these conditions in the national rollout. PURPOSE OF THE REPORT We previously wrote an enrollment report which covered enrollment up through October of 2016. Since that time enrollment has closed for all of the HIV cohorts. The purpose of this report is to describe the complete enrollment for protocol 1 which was completed for the HIV cohorts at the end of December 2016 (December 16th). Enrollment is still ongoing for the TB, Hypertension and Diabetes cohorts and this report provides updated enrollment data for those cohorts as well. In addition, we describe the additional challenges we faced since the last report to completing enrollment and document the strategies that we used to complete our enrollment for the HIV cohorts. As with the previous report, we describe the data sources used to collect baseline information on the cohorts, review the eligibility criteria for each cohort and describe the methodology used to identify eligible subjects for each cohort. Finally, we review the final enrollment data within each HIV cohort and describe the cohorts in terms of their basic clinical and demographic characteristics. 2 SECOND ENROLLMENT REPORT ELIGIBILITY CRITERIA For all cohorts, we created general eligibility criteria that focused the study on non-pregnant adults. All cohorts used these general eligibility criteria, including the TBHD cohorts. Below we describe the general inclusion and exclusion criteria which applied to all cohorts within the study. 4.1 General inclusion/Exclusion criteria Inclusion criteria ► ≥ 18 years old ► Meet the inclusion criteria for one or more of the cohorts Exclusion criteria ► Not resident in the facility’s catchment area ► Recorded intention to transfer care to a different facility within 12 months ► Pregnant and eligible for prevention of mother to child transmission (PMTCT) 4.2 Cohort specific inclusion/exclusion criteria Within each cohort, specific inclusion/exclusion criteria were developed to ensure enrollment of appropriate subjects. The enrollment criteria followed the December 2014 national guidelines for HIV care and ART and November 2015 National Adherence Guidelines for Chronic Disease (HIV, TB and NCDs) with some updates based on the August 2016 version of the Guidelines. Below we describe inclusion/exclusion criteria applied to each cohort. Cohort 1: Fast track treatment initiation inclusion criteria (Patients newly eligible for ART) ► Determined to be eligible to start ART under prevailing national guidelines (before September 2016, CD4 count < 500, WHO Stage 3 or 4 condition, no TB or cryptococcal meningitis; September 2016 and after, anyone HIV positive and considered eligible by the sites) Cohorts 2 and 3: Adherence clubs or decentralized medicine delivery (Patients stable on ART) ► On same ART treatment regimen for at least 12 months ► Most recent viral load taken in past 3 months ► Two consecutive viral loads undetectable (<400 copies/ml3) 3 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA Cohort 4: Enhanced adherence counseling (Patients on ART with poor adherence) ► On first-line ART for at least 3 months ► Poor adherence as indicated by an elevated viral load (>400 copies/ml3) Cohort 5: Early patient tracing (Patients lost from ART programs) ► Initiated ART ► Failed to return for a scheduled appointment within 5 to 90 days from their scheduled appointment date Cohort 6: TBHD cohort (Tuberculosis, hypertension, and diabetes patients) ► Screened for TB, hypertension or diabetes at last visit during enrollment period June- December 2016 or diagnosed with TB or hypertension between June-December 2016, or diabetes between January-December 2016. ► Does not have existing TB, hypertension (diagnosed before June 2016) or diabetes (diagnosed before January 2016) OVERVIEW OF STUDY SITES The study is being conducted at 24 primary health care clinics (PHCs) in South Africa. In consultation with the NDoH, the study team chose six clinics from one district each in Gauteng, KwaZulu Natal, Limpopo, and North West Provinces giving a total of 24 sites. These provinces were chosen because in most cases they are high HIV burden provinces with high burden districts and high volume clinics. Each site was required to meet the following criteria: 1) a high volume site as defined by having its Total Remaining on ART (TROA) being above 1000 patients; 2) not already be a National Health Insurance pilot site; 3) generating computerized TIER.net Phase 2 (all HIV patient data captured electronically including back capture of historical data) essential for the evaluation outcomes; and 4) not participating in any other adherence-related studies or pilots. Sites were chosen so that they could be matched with another site within a district roughly on TROA, proportion of patients virally suppressed, setting (rural/urban/formal/informal) and location (sites near each other). After choosing the 12 pairs of sites (24 total), NDoH, World Bank, Boston University and HE2RO randomly allocated one site within each pair to be an intervention site and the other to be a control site. Table 1 below describes the study populations at each facility using data from TIER.net and DHIS on patients who were active in care (i.e. had visited the clinic within the last four months) and 4 SECOND ENROLLMENT REPORT who would have been eligible for enrollment by virtue of having met our eligibility criteria. Catchment population and PHC headcount are from DHIS while the eligibility data come from TIER.net data. Table 1 Population data (facility headcount and total active patients) at each facility and total numbers eligible by intervention (I) and control (C) for each intervention 2015 Total PHC Total active FTIC AC/DMD** catchment headcount, on ART on eligible eligible EAC eligible TRIC eligible population monthly 2015 30 June 2016 June 1-30 30 June 2016 June 1-30 30 June2016 Facility average N N N (%)* N (%) N (%) EKURHULENI I: Motsamai Clinic 12681 2701 1563 43 486 (31%) 53 (3%) 86 (6%) C: Tamaho Clinic 18686 3980 1741 18 366 (21%) 8 (0%) 374 (21%) I: Phola Park CHC 70047 14920 3434 17 1422 (41%) 97 (3%) 259 (8%) C: Ramokonopi CHC 76257 16242 3502 63 999 (29%) 146 (4%) 190 (5%) I: Khumalo Clinic 26255 5592 2587 22 808 (31%) 66 (3%) 185 (7%) C: Zonkizizwe 1 Clinic 21856 4655 1614 35 528 (33%) 55 (3%) 121 (7%) EKURHULENI TOTAL 225782 48090 14441 198 4609 (32%) 425 (3%) 1215 (8%) MOPANI I: Grace Mugodeni CHC 22533 5 212 2061 15 982 (48%) 31 (2%) 158 (8%) C: Motupa Clinic 17263 4 511 1708 23 546 (32%) 17 (1%) 204 (12%) I: Giyani CHC 25982 7 149 2184 30 621 (28%) 60 (3%) 363 (17%) C: Dzumeri Clinic 23644 5 753 1597 10 598 (37%) 16 (1%) 253 (16%) I: Tzaneen Clinic 17258 4265 1851 19 700 (38%) 2 (0%) 334 (18%) C: Nkowankowa CHC 22629 4 881 1132 10 233 (21%) 6 (1%) 97 (9%) MOPANI TOTAL 129309 31771 10533 107 3680 (35%) 132 (1%) 1409 (13%) BOJANALA I: Letlhabile CHC 69554 9825 3855 43 1729 (45%) 31 (1%) 487 (13%) C: Wonderkop Clinic 21236 3200 1848 49 875 (47%) 28 (2%) 272 (15%) I: Hebron Clinic 32352 4301 1714 29 870 (51%) 34 (2%) 130 (8%) C: Majakaneng Clinic 21539 3259 1431 24 751 (52%) 12 (1%) 136 (10%) I: Tlhabane CHC 79290 15130 5202 86 1330 (26%) 50 (1%) 667 (13%) C: Bafokeng CHC 62149 10810 3113 37 1252 (40%) 36 (1%) 520 (17%) BOJANALA TOTAL 286120 46525 17163 268 6807 (40%) 191 (1%) 2212 (13%) UTHUNGULU I: King Dinizulu Clinic 24456 6058 2528 28 1152 (46%) 22 (1%) 420 (17%) C: Nkwalini Clinic 10434 2573 1089 11 609 (56%) 11 (1%) 232 (21%) I: Thokozani Clinic 42678 10657 3875 65 220 (6%) 15 (0%) 722 (19%) C: Nseleni CHC 71060 18273 6218 83 2118 (34%) 6 (0%) 1021 (16%) I: Buchanana Clinic 21944 3574 1323 12 722 (55%) 10 (1%) 239 (18%) C: Ntambanana Clinic 19103 3323 1416 17 769 (54%) 8 (1%) 99 (7%) UTHUNGULU TOTAL 189675 44458 16449 216 5590 (34%) 72 (0%) 2733 (17%) TOTAL 830 886 170 844 58586 789 20686 (35%) 820 (1%) 7569 (13%) Note: *Percent of total active on ART, 30 June 2016; **DMD has different intervention and control sites. As can be seen, by design the sites were very different in terms of geographic location, size and number of eligible subjects within each of the intervention cohorts. For most cohorts, this did not pose any major challenges. Over the time period described, for most cohorts and sites a sufficient number of subjects became eligible to allow us to reach our site specific targets and certainly our overall targets. At the same time, this table illustrates some of the challenges the team faced in enrolling subjects into some of the cohorts: there were few new eligible subjects in the EAC cohort 5 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA over the entire enrollment period, particularly in uThungulu and parts of Mopani and Ekurhuleni. This meant that we did not have sufficient numbers of subjects to enroll at several sites as will be shown below and led to the decision not to reach the increased sample size agreed to (discussed below in the section on sample size). ENROLLMENT 6.1 Methods for identifying eligible individuals To enrol subjects into the study, the team identified patients who met the enrollment criteria for each cohort (described below). The procedures were different depending on whether or not the site was an intervention or control site. In control sites, because the intervention had not been implemented, the team had to find subjects who met the counterfactual definition of a control (i.e. they would have been eligible for the interventions had they gone to an intervention site). These subjects could typically be identified by using TIER.net. At intervention sites, we only included subjects who actually got the interventions. Identifying which patients got the interventions turned out not to always be straightforward. In part, this was because sites didn’t always keep good records on who got the interventions for some cohorts (the most striking example being those in the EAC cohort). In other cases, it was because the interventions were not given to those who should have been eligible. This was common with the tracing (TRIC) cohort as in many cases sites chose to apply TRIC to all patients missing appointments, not just those who recently missed them as called for the guidelines. In other cases, it was difficult to identify eligible subjects because sites shifted priorities and moved patients from one intervention to another (e.g. from ACs to DMD) as a result of implementation of the decanting/decongestion strategy. In cases where registers were not sufficient to allow us to identify which patients got the interventions within a site we developed alternative methods to identify the cohorts such as: reviewing DMD script sheets for DMD; checking patient files for evidence of FTIC or EAC counselling and completed Patient Adherence Plans, or AC visits; and collaborating with implementing partners to access their tracing registers where those partners had taken over the responsibility for tracing. An exception to how we enrolled subjects into the cohorts at control sites was for the DMD and the AC cohorts. To create the counterfactual control group we needed to find patients who were eligible for these interventions. Because the eligibility criteria for these two cohorts were identical, we sought to prevent enrollment of the same subject into two cohorts at a control site. We note, however, that there would not have been bias introduced if we had, we simply did this for logistical reasons to increase the probability that we would get roughly balanced patient characteristics predictive of the outcomes within each cohort. Therefore to enrol into these two cohorts, we identified patients eligible for these two interventions at each control site and allocated them 6 SECOND ENROLLMENT REPORT randomly to one of the two cohorts. This was further complicated by the fact that the DMD analysis was observational as described below. For certain cohorts at specific sites we were unable to prospectively enrol the required number of patients before closing enrollment on the 16 December 2016 (e.g. EAC and TRIC in Mopani, AC and DMD in Ekurhuleni). For these cohorts we are continuing to review cohort registers and patient files during follow up visits to double check that eligible patients were not missed during the enrollment period. If these patients are identified they will be enrolled retrospectively and followed as part of the cohort. 6.2 Specific Procedures for Intervention Sites As noted, at intervention sites we sought to identify individuals who were both eligible for and received specific interventions. We note that in some cases, sites did not give all patients eligible for an intervention the intervention and in other cases they gave the interventions to subjects who were not eligible. Accordingly, we required both eligibility for and receipt of the intervention as enrollment criteria. To achieve this, we used one of three approaches described below. Our goal was to spread out enrollment into each cohort over a period of months (June to December). We did this to prevent overenrolling subjects into the cohorts during the early stages of the rollout when sites were still learning how to deliver the interventions. This was not always successful as in some cases site enrollment could be spaced over the enrollment period because rollout of the interventions occurred early and quickly, while in others the interventions were not rolled out early enough for us to enrol over many months. We identified potentially eligible subjects by first downloading the most recent TIER.Net dispatches from each facility and running STATA code to check for patients who should have received the interventions as described in the AGL for each specific HIV cohort. Prior to each facility visit that the team made, the Data Manager sent lists of eligible patients for each cohort to the Provincial Study Coordinators. A password protected ‘Cohort Creation List’ was sent along with targets for visit enrollment for each cohort. Then one of three approaches was used at each intervention site: 1. List to Register Approach. In cases where intervention specific registers (e.g. FTIC register, AC register, etc.) were kept, our Cohort Creation Lists were compared against each one to identify which of the eligible patients had actually received the intervention. Provincial teams checked each patient on the cohort creation list until the target number of cohort patients to be enrolled was identified. 2. List to File Approach. In cases where intervention registers were absent, did not contain enough information to verify a patient got the interventions, or were deemed to be of poor quality and therefore might miss patients who actually got the interventions, files of patients on the Cohort Creation Lists were identified and used to determine whether or not the patient had in fact received the intervention for which they were eligible. Evidence of receiving the intervention varied by cohort (e.g. record of specific AGL counselling sessions, presence of a patient adherence plan, etc.). This approach was used in later rounds of 7 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA recruitment mainly to complete enrollment of FTIC and EAC cohorts and predominantly in Ekurhuleni. 3. Direct from Register Approach. If patients on the Cohort Creation List could not be found using the List to Register Approach then rather than using the cohort creation lists to identify eligible patients, patients were identified directly from the intervention specific registers and their files found to confirm they were eligible according to the AGL criteria. Regardless of the method used to identify patients, patient files were reviewed and information extracted using an electronic case report form (CRF) to confirm patients did meet all eligibility criteria for that cohort (including receipt of the intervention). 6.3 Control sites For controls sites, the Data Manager created Cohort Creation Lists of eligible individuals as described above using TIER.net data and sent each list to the Provincial Study Coordinators. Each list was reviewed and the team pulled files for each patient until the cohort enrollment target was achieved (i.e. all patients enrolled in control cohorts were found using the List to File Approach). Files were then reviewed and information extracted using the electronic CRF to confirm eligibility. We note that many control sites did implement some form of the interventions (e.g. adherence counselling is routinely done at sites even if not enhanced, tracing is often done even if not targeted at those recently missing visits). At control sites, patients were enrolled based on eligibility for the interventions, while receipt of interventions at these control sites was disregarded for enrollment purposes. 6.4 Cohort Specific Enrollment Below we describe the approach to identifying eligible patients for each of the cohorts. For all cohorts at both intervention and control sites we first produced Cohort Creation Lists from the most recent TIER dispatch for each facility identifying patients who on the date of the site visit by the data team: ► Were active in care (visit within last 124 days and no record that the patient was listed as “died”, “lost to follow up” or “transferred/moved out”) ► Were greater than 18 years of age ► Had their pregnancy status at last visit was not listed as “pregnant” Cohort lists were refined using TIER.net to select patients who met the general inclusion criteria. We then proceeded differently for intervention and control subjects and followed specific procedures for each cohort as described below. 8 SECOND ENROLLMENT REPORT Cohort 1: Fast Track Initiation Counselling (Patients newly eligible for ART) ► Control Subjects: Control subjects were identified and enrolled using the ‘List to File’ approach. Patient files for eligible patients were found and screened using the electronic CRF to ensure FTIC eligibility criteria were met. ► Intervention Subjects: For intervention subjects, we first used a ‘List to Register’ approach then a ‘Direct from Register’ then the ‘List to File’ approach. In some facilities (predominantly in Ekurhuleni) FTIC registers were absent or inadequately completed, hence a ‘List to File’ approach was used. Eligible files were reviewed for evidence of FTIC counselling sessions recorded on clinical stationery and/or presence of a completed or partially completed AGL Patient Adherence Plan on the patient file. All selected files were then screened using the electronic CRF to confirm all eligibility criteria that could not be verified through TIER.net (i.e. resident in the facility catchment area, no intention to transfer to another facility, no cryptococcal meningitis diagnosis at ART initiation). Once eligibility had been confirmed an FTIC barcode was assigned and the patient enrolled. Cohort 2: Adherence Clubs (Patients stable on ART) AC enrollment was complicated by the fact that DMD was rolled out at some of the control facilities and because at some sites AC patients were shifted to DMD to meet DMD targets, necessitating additional checks. ► Control Subjects: Control subjects were identified and enrolled using the ‘List to File’ approach and patient files screened to ensure all AC eligibility criteria were met. Further checks were made to ensure selected patients were not currently receiving their medication through DMD if it was in use at a control site. ► Intervention Subjects: For intervention subjects, first a ‘List to Register’ approach was used to identify patients recorded on the facility AC register(s). If the sample size could not be achieved using this approach, the ‘Direct from Register’ approach was used to identify and find AC patient files. A ‘List to File’ approach was not used for this cohort. Eligible patient files were screened using the electronic CRF to confirm patients’ eligibility (including confirmation of enrollment in an AC for intervention sites), and ensuring that patients had no record of picking up their medications separately from a DMD pick-up-point. All patients who eligible for the AC cohort were assigned an AC barcode. Cohort 3: Decentralized Medicine delivery (Patients stable on ART) As noted, DMD was not rolled out at all intervention sites and was rolled out at some control sites (Table 2). Therefore, we had to tailor our approach to enrollment for this cohort. For DMD, cohort lists were developed as described above for the AC cohort, although in the case of DMD we did not have a way to confirm enrollment into DMD using TIER.net as no field captures this information. 9 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA ► Control Subjects: Unlike the other HIV cohorts, because DMD was rolled out in some randomized control sites and not in all randomized intervention sites, control DMD subjects were enrolled from any site that was not implementing DMD during the enrollment period (June-December) regardless of whether or not the site was randomized to intervention or control. Control subjects were identified and enrolled using the ‘List to File’ approach. Given that we did enroll DMD control subjects at some intervention sites, we instituted further checks to ensure that selected cohort patients were not also enrolled in an AC. At control sites where DMD was implemented during the enrollment period, we checked the IDs of all patients enrolled in the AC cohort and replaced any patient in that cohort who received DMD. ► Intervention Subjects: For intervention subjects (as noted, enrolled at any site currently implementing DMD irrespective of randomization assignment), the ‘List to register’ approach was used to identify patients recorded on DMD register(s) as having been decanted to a DMD (CCMDD/CDU) medicine pick up point. The relevant patient file numbers were recorded and those patient files found. If the targeted sample size of DMD patients could not be enrolled at a site, then the ‘Direct from register’ approach was used and patient files found. As with ACs, a ‘List to file’ approach was not used for this cohort at DMD intervention sites. At intervention sites that implemented DMD while we were enrolling, any patient in the AC cohort who was transferred to DMD was replaced in the AC cohort. Patient files were found and then screened using the electronic CRF to confirm patients’ eligibility including confirmation that the patient was not also enrolled in an AC. Assuming all criteria were met the patient was enrolled and a DMD barcode assigned. Table 2 Assumptions and sample sizes for each cohort DMD Intervention/ Control DMD Intervention/ Control Facility Status Facility Status gp Motsamai Clinic DMD Intervention nw Letlhabile CHC DMD Intervention gp Tamaho Clinic DMD Intervention nw Wonderkop Clinic DMD Control gp Phola Park CHC DMD Control nw Hebron Clinic DMD Intervention gp Ramokonopi CHC DMD Control nw Majakaneng Clinic DMD Intervention gp Khumalo Clinic DMD Control nw Tlhabane CHC DMD Intervention gp Zonkizizwe 1 Clinic DMD Control nw Bafokeng CHC DMD Intervention lp Grace Mugodeni CHC DMD Control kz King Dinizulu Clinic DMD Intervention lp Motupa Clinic DMD Control kz Nkwalini Clinic DMD Control lp Giyani CHC DMD Control kz Thokozani Clinic DMD Intervention lp Dzumeri DMD Control kz Nseleni CHC DMD Intervention lp Tzaneen Clinic DMD Control kz Buchanana Clinic DMD Control lp Nkowankowa CHC DMD Control kz Ntambanana Clinic DMD Control 10 SECOND ENROLLMENT REPORT Cohort 4: Enhanced Adherence Counseling (Patients on ART with poor adherence) For EAC, cohort lists were first developed using TIER.net. ► Control Subjects: Control subjects were identified and enrolled using only the ‘List to file’ approach. ► Intervention Subjects: For intervention subjects, first the facility EAC register(s) were used in a ‘List to register’ approach. If the targeted sample size for a site for the EAC cohort could not be enrolled, the ‘Direct from Register’ approach was used. In some intervention facilities where EAC registers were absent or inadequately completed (predominantly Ekurhuleni) the ‘List to File’ approach was used. Patient files were reviewed for evidence of EAC sessions and/or presence of a completed or updated AGL Patient Adherence Plan. Once eligibility had been confirmed the patient was enrolled and an EAC barcode assigned. Cohort 5: Early patient tracing (TRIC) For TRIC, cohort lists were again refined using TIER. ► Control Subjects: Control subjects were identified using just the ‘List to File’ approach. Eligible files were screened using the electronic CRF to ensure that patients had in fact missed an appointment by 5-90 days as indicated on TIER (occasionally a TIER backlog could result in patients being identified as having missed a visit when in fact they had not). ► Intervention Subjects: For intervention subjects, a ‘List to Register’ approach was used to identify patients from the TRIC registers who needed tracing or who had been traced. The relevant patient files were then found. If the targeted TRIC sample size for a site could not be enrolled, then the ‘Direct from Register’ approach was used to identify TRIC eligible patients and their patient files. In some facilities (e.g. some Mopani and uThungulu clinics), TRIC registers were not held by the facility but by implementing partners or Ward Based Outreach Teams (WBOT) teams and information was not sent back to the facility. This led to delays in enrolling this cohort at some intervention sites. When possible, visits were made to partner offices and meetings with WBOTs were organized in order to gain access to the registers and tracing information. This is similar to the ‘Direct from Register’ approach. The names and file numbers (if recorded) of patients recorded on these lists were noted and taken back to the facility by the team and these patient files found. The ‘List to File’ approach was not used at intervention sites for this cohort. Patient files were screened to confirm eligibility (including confirmation that they had missed a scheduled appointment by 5-90 days). Once eligibility was confirmed a TRIC barcode was assigned. 11 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA Cohort 6: Tuberculosis, Hypertension, and Diabetes Cohort (TBHD patients) TBHD Screening Cohort (Cohort 6A) At each facility visit the study team reviewed patient files from the previous or current day to exclude patients < 18 years and pregnant women. All remaining patients are then eligible for the TBHD Screening cohort (Group A). Baseline data for these patients is extracted from their files using the electronic CRF. As we require 100 eligible subjects per site for Group A, we continued this process until the targeted sample size was reached. After the patient information was captured in the CRF, it is reviewed by the research team to determine the status of each patient: ► Not screened or eligible for screening (i.e. not screened for tuberculosis, hypertension or diabetes; no evidence of TB, hypertension or diabetes/no known disease/screened negative; prevalent case of tuberculosis or hypertension (diagnosed more than 1 month ago), prevalent case of diabetes (diagnosed more than 6 months ago), MDR or XDR TB) ► Screened positive for tuberculosis (at last visit) ► Screened positive for hypertension (at last visit) ► Screened positive for diabetes (at last visit) ► Newly diagnosed with tuberculosis (in last month) ► Newly diagnosed with hypertension (in last month) ► Newly diagnosed with diabetes (in last 6 months) At this point the Data Manager returned a list of file numbers of all patients who had screened positive for tuberculosis, hypertension or diabetes to the Provincial Teams and these files were found and marked with a barcode label to signify enrollment in the TBHD Screening cohort to be followed up. TBHD Diagnosed Cohort (Cohort 6B) There were three ways in which patients are identified as eligible for and enrolled in the TBHD Diagnosed cohort: 1. Incident cases from the TBHD screening cohort. Patients from the TBHD screening cohort who are diagnosed with tuberculosis, hypertension or diabetes during the TBHD cohort enrollment period (June-December 2016) are eligible for inclusion in the TBHD diagnosed cohort (Cohort 6B). These file numbers are confirmed by the Data Manager and then marked with a Cohort 6B (TBHD diagnosed) barcode. 2. Cases identified as recently/newly diagnosed while enrolling the TBHD screening cohort. Cases of tuberculosis and hypertension diagnosed at their last site visit (June-December 12 SECOND ENROLLMENT REPORT 2016), or diabetes cases diagnosed after Jan 2016 that were identified during the creation of the TBHD Screening cohort (Cohort 6A) were also eligible. These were identified by the Data Manager from the baseline CRF data previously collected. A list of file numbers is then returned to the provincial teams so that patient files could be marked with a Cohort 6B (TBHD Diagnosed) barcode. 3. Register identified cases. Tuberculosis patients (diagnosed after May 2016) were identified from the tuberculosis suspect register or the tuberculosis register, and hypertension (diagnosed after May 2016) and diabetes patients (diagnosed after Jan 2016) were identified from the PHC tick register. File numbers were noted and patient files pulled to verify the diagnosis is within the required time period. If eligibility was confirmed files were marked with the Cohort 6B (TBHD Diagnosed) barcode label. 6.5 Enrollment process/Identification of cohorts While enrollment generally went smoothly after initiation of enrollment procedures began on June 20th, 2016 the team did identify some barriers to completing the process, and in some cases, starting the enrollment process. This was often cohort specific, with sites having difficulty enrolling into one or two of the cohorts, while the others were able to enrol without problem. Barriers for each cohort, therefore required different solutions. The initial barriers to enrollment fell into four general categories: 1) delays in implementation of the interventions at the sites and delays in implementation per the NDOH’s Standard Operating Procedures of the AGL; 2) changes in rollout of the intervention due to the pressure to “decant” patients; 3) incomplete or not used registers for the interventions; and 4) delay in enrollment at control sites due to delays in implementation of the interventions at intervention sites. These have each been discussed in the previous enrollment report. While these challenges were largely solved or the impacts mitigated, some new challenges arose to completing full enrollment. In addition to the barriers reported in the first enrollment report, the team encountered: 1) additional delays in rollout of the interventions; 2) an insufficient number of patients provided with the interventions at some intervention sites; and 3) delays in implementation of the interventions meant we could not continue to enroll subjects and still meet our endpoints within the appropriate time frame. Below we discuss each barrier. Delays in Implementation or not implemented as per AGL As with the previous report, we note that the main barrier to enrollment resulted from delays in the implementation of the rollout of some of the interventions in the AGL or failure to implement interventions as specified in the AGL. DMD and ACs were the most likely to be affected and this was largely due to the implementation during the study period of a National Decanting Strategy which sought to decongest clinics. The Decanting Strategy meant that many sites, whether control or intervention, needed to quickly implement DMD, preventing us from achieving a “pure” counterfactual. Because of this, we made the shift to analyse this cohort using methods for 13 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA observational research rather than randomized trials. Still, the shift also had the impact of slowing recruitment because we could not be certain that subjects recruited into the AC cohorts were not being decanted to DMD. The issue was most commonly seen in Gauteng and uThungulu. Reluctance to implement interventions as specified in the AGL SOPs further hampered enrollment as on occasion this meant that those enrolled in the interventions were not eligible or if they were eligible had not been enrolled in sufficient numbers. This issue was most commonly seen in uThungulu, particularly where ACs had previously existed prior to the AGL and had used slightly different enrollment criteria (e.g. only one viral load required). Other examples included EAC in Buchanana, uThungulu where staff in charge had seemingly been enrolling patients who were virally supressed rather than unsuppressed; EAC in Grace Mugodeni in Mopani where clinicians were reluctant to refer patients for EAC; and also TRIC at Grace Mugodeni where a decision had been made to focus tracing on patients who had just missed a visit (i.e. by 1 day) rather than early and late missed appointments. These issues were all escalated to managers and support partners but unfortunately were not rectified quickly enough or are still ongoing. Incomplete, poorly completed or not used registers for the interventions While in most cases, our data enhancement plans allowed us to work with sites to ensure improved record keeping and data recording, not all sites had the capacity to use the registers for documenting who got the interventions. This issue occurred most often in Mopani and Ekurhuleni for the Early Adherence Counselling and TRIC interventions and was often the result of implementing partners choosing to keep separate registers for interventions, which were not kept at the sites, or a shortage of staff capacity and some resistance by staff to complete these registers. Poor completion and use of registers and no record of intervention delivery in patient files was why recruitment into the Early Adherence Counselling cohort fell well below target at Khumalo, Tzaneen and Thokozani and the reason we were unable to achieve TRIC target numbers in Tzaneen. Insufficient numbers of patients enrolled onto the interventions While the decanting strategy resulted in delays in enrollment, in other sites, some interventions simply did not enrol a sufficient number of patients into the interventions or they enrolled patients without using the AGL inclusion criteria, which meant some subjects who got the interventions were not eligible for enrollment into the evaluation. The most common examples of this were for DMD and EAC. For example, while DMD numbers at Motsamai in Ekurhuleni were initially encouraging, the number of patients enrolled and picking up their medications at the DMD pick-up-point began to diminish quite quickly after implementation with clients preferring to pick their medications up at the facility. The reluctance by clinicians to refer patients for Early Adherence Counselling at Grace Mugodeni in Mopani resulted in too few patients being enrolled for this intervention. Delays in Implementation Prevented Enrollment Due to the agreed Endpoint Timeframes 14 SECOND ENROLLMENT REPORT Not all sites that delayed rollout of the interventions (or slow rollout once they began the interventions). However, as the evaluation’s timeline has already been shifted several times due to delays in AGL implementation, and the long-term outcomes would take 12 months of cohort follow- up, there was a limit to the timeframe of possible cohort enrollment. The intention was to close HIV cohort enrollment in December 2016 to obtain the data on AGL effectiveness by December 2017. Therefore, it was challenging in some sites to achieve the full sample size for each HIV cohort by December 2016. The most common examples of this were for ACs in Ekurhuleni (Motsamai and Khumalo) in Gauteng. 6.6 Strategies to Overcome Barriers The team enacted three main strategies to achieve the targeted sample size in the face of the issues described above. Each is described below. Increasing the duration of enrollment through 16 December 2016 We had initially hoped to complete enrollment by the end of November of 2016. In consultation with the World Bank, we made the decision to extend the enrollment period and only close enrollment on December 16th, 2016. This increased duration allowed us to meet or nearly meet the increased sample size targets in 4 of the 5 cohorts with only Early Adherence Counselling not approximately reaching the targeted sample size. Because not a sufficient number of subjects received EAC and met the eligibility criteria (EAC was often provided to those who did not meet the criteria according to the guidelines), we were not able to achieve our targeted sample size in the required time and the strategies below were also not entirely sufficient to achieve the target. Overenrolling patients at some sites While we had initially hoped for a perfectly balanced design in which all sites enrolled the same number of subjects for each intervention, this was not possible at all sites. As we had to under enroll at some sites, we increased enrollment at others where there were sufficient numbers of patients to do so. This was easier to do in control sites than in intervention sites because control site subjects were not required to have actually gotten the intervention, only to have been eligible to do so. This meant more patients could be found who met the inclusion criteria. The most common examples of this were in FTIC, ACs and Early Adherence Counselling although some over enrollment occurred at specific sites for DMD and TRIC as well. The sites where over enrollment occurred can be seen in Table 5 below. Shifting to including some retrospective enrollment While we did need to close enrollment as of December 16th 2016 in order to reach the long-term endpoints by end 2017, this did not prevent us from enrolling subjects from sites that we had previously not enrolled all eligible subjects. This retrospective approach to enrollment meant that the site teams were not looking for newly eligible subjects on their visits but rather were looking for subjects who had been eligible but had not already been enrolled. This strategy was used 15 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA sparingly as it was not possible at many sites as either all eligible subjects were enrolled or because we didn’t want to have a strongly unbalanced design in which we had a large sample size at one site and few at another. The most common examples of this were in EAC, DMD and TRIC. SAMPLE SIZE In our initial protocol, our total sample size for the study was estimated to be 2,880 subjects in the five intervention cohorts and 4,800 in the TBHD cohort, for a total sample size of 6,680. However, as we noted there was crossover between subjects in the AC and DMD cohorts, in consultation with the co-Principal Investigators we decided to increase the sample size by 20% in each of the HIV cohorts. While this sample size was likely not necessary for most of the cohort, it would give us sufficient power to detect differences even if some subjects needed to be excluded from the final analysis there was no reason not to. The table below shows the sample size that was estimated to be required for each cohort initially and after we increased the target numbers. We determined each sample size to be sufficient to measure the short term outcome for that cohort. For all the cohorts except the TBHD cohort, calculations assumed a site-clustered design with the clinic as the cluster and 24 clusters evenly randomized between intervention and comparison groups. They also assumed a coefficient of variation of 0.1, 80% power; and an alpha of 0.05. In Table 3 below we describe the remaining assumptions behind the sample size for each cohort. Table 3 Assumptions and sample sizes for each cohort Increased Detection Cohort Initial Sample Size Sample Size Assumed % with the outcome in control arm limit Fast Track ART Initiation 600 patients (25 per 720 patients Counseling clinic) (30 per clinic) 60% of patients will initiate ART 15% change 480 patients (20 per 576 patients Adherence Clubs clinic) (24 per clinic) 80% of patients will make all medication pickups 15% change Decentralized Medicine 480 patients (20 per 576 patients delivery clinic) (24 per clinic) 80% of patients will make all medication pickups 15% change Enhanced Adherence 1008 patients 52% of patients with a detectable viral load will re- Counseling 840 (35 per clinic) (42 per clinic) suppress after one session 15% change Early tracing of patients 480 patients 576 patients 30% of patients will be loss to follow up without lost to follow up (20 per clinic) (24 per clinic) intervention 15% change TB, hypertension, and Descriptive in nature, no specific sample size No specified diabetes 4800 patients 4800 patients calculation was done change 16 SECOND ENROLLMENT REPORT ENROLLMENT 8.1 Timing of cohort initiation As noted, cohort enrollment began on June 20th, 2016. Table 4 below describes the timing of initiation of site assessments, and data collection for each cohort. The table also shows matched pairs of clinics where the start of cohort enrollment was 2-4 weeks (yellow) or one month or more (red) apart. Table 4 Timing of cohort initiation by site and cohort Cohort 6 Cohort 1 Cohort 2 Cohort 3 Cohort 4 Cohort 5 (TBHD): Site (FTIC): Data (AC): Data (DMD): Data (EAC): Data (TRIC): Data Data assessment collection collection collection collection collection collection Facility date start date start date start date start date start date start date gp Motsamai Clinic 28-Oct-15 23-Jun-16 Pending 23-Jun-16 16-Aug-16 23-Jun-16 23-Jun-16 gp Tamaho Clinic 25-Nov-15 20-Jun-16 31-Aug-16 21-Jun-16 20-Jun-16 20-Jun-16 20-Jun-16 gp Phola Park CHC 10-Nov-15 25-Aug-16 20-Jul-16 20-Jul-16 26-Aug-16 15-Aug-16 20-Jul-16 gp Ramokonopi CHC 23-Nov-15 29-Aug-16 21-Jul-16 30-Aug-16 29-Aug-16 29-Aug-16 20-Jul-16 gp Khumalo Clinic 17-Nov-15 18-Aug-16 29-Jun-16 29-Jun-16 19-Aug-16 19-Aug-16 30-Jun-16 gp Zonkizizwe 1 Clinic 16-Nov-15 05-Aug-16 04-Jul-16 04-Jul-16 05-Aug-16 08-Aug-16 01-Jul-16 lp Grace Mugodeni CHC 16-Nov-15 24-Jun-16 27-Jun-16 27-Jun-16 13-Jul-16 12 Oct-16 27-Jun-16 lp Motupa Clinic 20-Nov-15 28-Jun-16 14-Jul-16 15-Jul-16 28-Jun-16 28-Sep-16 28-Jun-16 lp Giyani CHC 17-Nov-15 30-Jun-16 11-Aug-16 21-Jul-16 30-Jun-16 Pending 30-Jun-16 lp Dzumeri Clinic 14-Dec-15 04-Jul-16 11-Aug-16 05-Jul-16 05-Jul-16 04-Oct-16 04-Jul-16 lp Tzaneen Clinic 19-Nov-15 21-Jun-16 06-Jul-16 20-Jun-16 20-Jul-16 06-Oct-16 20-Jun-16 lp Nkowankowa CHC 18-Nov-15 22-Jun-16 22-Jun-16 22-Jun-16 08-Jul-16 23-Sep-16 22-Jun-16 nw Letlhabile CHC 24-Nov-15 28-Jun-16 28-Jun-16 17-Aug-16 29-Jun-16 30-Jun-16 28-Jun-16 nw Wonderkop Clinic 25-Nov-15 27-Jun-16 27-Jun-16 02-Sep-16 28-Jun-16 14-Jul-16 27-Jun-16 nw Hebron Clinic 25-Nov-15 20-Jun-16 21-Jun-16 26-Aug-16 20-Jun-16 10-Aug-16 20-Jun-16 nw Majakaneng Clinic 24-Nov-15 22-Jun-16 22-Jun-16 16-Aug-16 22-Jun-16 15-Jul-16 22-Jun-16 nw Tlhabane CHC 26-Nov-15 25-Jul-16 22-Aug-16 06-Sep-16 08-Aug-16 25-Jul-16 24-Jun-16 nw Bafokeng CHC 26-Nov-15 12-Jul-16 12-Jul-16 29-Aug-16 12-Jul-16 12-Jul-16 23-Jun-16 kz King Dinizulu Clinic 09-Dec-15 28-Jun-16 07-Sep-16 06-Sep-16 28-Jun-16 09-Nov-16 28-Jun-16 kz Nkwalini Clinic 24-Dec-15 30-Jun-16 22-Aug-16 26-Sep-16 21-Jul-16 30-Jun-16 30-Jun-16 kz Thokozani Clinic 27-Nov-15 04-Aug-16 17-Aug-16 17-Aug-16 21-Sep-16 14-Sep-16 04-Aug-16 kz Nseleni CHC 21-Dec-15 25-Aug-16 26-Aug-16 26-Aug-16 25-Aug-16 13-Oct-16 04-Aug-16 kz Buchanana Clinic 11-Dec-15 21-Jun-16 21-Oct-16 21-Jun-16 28-Jul-16 22-Jun-16 20-Jun-16 kz Ntambanana Clinic 10-Dec-15 23-Jun-16 31-Aug-16 23-Jun-16 24-Jun-16 27-Jun-16 23-Jun-16 Note: More than one month between enrollment at control and enrollment at intervention site. 2-4 weeks between enrollment at matched pair. 17 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA 8.2 Enrollment by cohort The team has tracked enrollment into each cohort over time to ensure progress towards the sample size targets and to ensure we did not exceed the total sample size allowed. Table 5 below shows the results of that tracking, demonstrating accrual into each cohort at each site through December 16th, 2016. The table is also stratified by evaluation facility and district. In the control arm alone, all five HIV cohorts reached the target sample size. In the intervention arm alone, four HIV cohorts did not reach the full sample size: the DMD cohort attained 96% (231 enrolled/240 target), AC cohort 95% (275/288), EAC cohort 71% (360/504) and TRIC 94% (272/288). Three of these four cohort met their initial sample size before it was increased by 20% (DMD, AC, TRIC) with only EAC falling below the initial sample size. Table 5 Enrollment by cohort as of October 6th Enrolled Enrolled Enrolled Enrolled Enrolled Enrolled Enrolled Cohort 1 Cohort 2 Cohort 3 Cohort 4 Cohort 5 Cohort 6a Cohort 6b Facility (FTIC) (AC) (DMD) (EAC) (TRIC) (TBHD) (TBHD) Total Target per facility 30 24 24 42 24 100 100 344 GAUTENG Motsamai Clinic 28 23 8 43 24 100 36 262 Tamaho Clinic 29 24 27 42 24 100 21 267 Phola Park CHC 31 28 24 49 25 100 62 319 Ramokonopi CHC 30 24 25 42 24 100 36 281 Khumalo Clinic 28 8 26 10 24 100 18 214 Zonkizizwe 1 Clinic 30 24 24 41 23 100 16 258 LIMPOPO Grace Mugodeni CHC 30 24 24 10 15 100 5 208 Motupa Clinic 35 24 24 42 23 100 3 251 Giyani CHC 30 24 24 37 24 100 2 241 Dzumeri 30 24 24 41 24 100 3 246 Tzaneen Clinic 27 24 24 18 15 100 2 210 Nkowankowa CHC 30 25 24 44 28 100 4 255 NORTH WEST Letlhabile CHC 32 24 22 42 26 100 51 297 Wonderkop Clinic 30 24 19 42 25 100 18 258 Hebron Clinic 31 24 25 42 21 100 29 272 Majakaneng Clinic 30 24 24 42 24 100 36 280 Tlhabane CHC 30 24 24 42 24 100 21 265 Bafokeng CHC 30 25 24 42 24 100 14 259 KWAZULU NATAL King Dinizulu Clinic 34 24 24 42 24 100 56 304 Nkwalini Clinic 34 27 27 43 28 100 40 299 Thokozani Clinic 37 24 26 15 26 100 30 258 Nseleni CHC 30 25 27 43 29 100 21 275 Buchanana Clinic 31 24 33 10 24 100 28 250 Ntambanana Clinic 30 24 26 43 24 100 25 272 18 SECOND ENROLLMENT REPORT Table 5 Enrollment by cohort as of October 6th (continued) Cohort 1 Cohort 2 Cohort 3 Cohort 4 Cohort 5 Cohort 6a Cohort 6b District (FTIC) (AC) (DMD) (EAC) (TRIC) (TBHD) (TBHD) Total Ekurhuleni 176 131 134 227 144 600 189 1601 Mopani 182 145 144 192 129 600 19 1411 Bojanala 183 145 138 252 144 600 169 1631 uThungulu 196 148 163 196 155 600 200 1658 Study Total 737 569 579 867 572 2400 577 6301 Study Target 720 576 576 1008 576 2400 2400 8256 Percent of target achieved 102% 99% 101% 86% 99% 100% 24% 76% Intervention enrollment 369 275 231 360 270 1200 340 3047 Control enrollment 368 294 348 507 302 1200 237 3254 Two factors were mainly responsible for the underachievement of the sample size for DMD: DMD was implemented at only 10 clinics while the remaining 14 sites were control sites. In addition, Motsamai (DMD intervention site) only had 8 patients for enrollment. Other DMD implementation issues also posed enrollment challenges including: a) Accessibility of the DMD registers and the ability to identify patients when they were picking up medicines through DMD; b) cases where DMD was being implemented for hypertension/diabetes but not for patients with HIV; c) Incorrect labelling of medicines and spelling of patient names; and d) Lack of documentation from CCMDD providers to update TIER. The last was perhaps the biggest barrier as we saw very little documentation of DMD in TIER. The chief reason for not meeting the enrollment target in the EAC cohort was the low number of patients who were eligible for enrollment in the following clinics: Khumalo (GP), Grace Mugodeni (LP), Tzaneen (LP), Thokozani (KZN), and Buchanana (KZN). This occurred because of the following EAC implementation challenges: a) Poor recording on registers and in patient files (difficulty identifying those receiving the EAC intervention); b) Registers not being updated due to absence of Lay Counsellor, or lost registers; and c) Confusion over which patients are eligible for EAC (virally suppressed patients have been receiving EAC but were not meant to be enrolled in the evaluation cohort). The evaluation team has also observed that EAC (as well as FTIC) registers are primarily driven to completion through the request of the Evaluation Team and not for patient management reasons. Two clinics contributed to under-achievement of the targeted sample size in the TRIC cohort as they had insufficient numbers of eligible patients: Grace Mugodeni and Tzaneen Clinic. The main implementation challenges for TRIC at the intervention clinics were: a) lack of airtime or access to phones within clinics; b) Difficulty linking information between facility, support partner, WBOTs, and CHWs; c) Wrong targeting of TRIC (defaulters, unsuppressed, appointments missed 1 day ago only); d) Absence of Lay Counsellor (delays tracing and completion of registers); and e) incomplete registers. The AC cohort achieved 95% of its targeted sample size in the intervention cohort. This was mainly due to insufficient numbers of eligible patients for cohort enrollment at Khumalo Clinic. Other sites 19 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA under-enrolled by one or two subjects, and this was only because on data cleaning some patients enrolled were found to not have met the inclusion criteria and had to be excluded. Thus, overall this cohort did not experience many issues, though as with the other cohorts, completion of registers was an issue. 8.3 Enrollment over time While we initially wanted each site to recruit the same number of subjects per cohort, due to logistical constraints at some sites and, more commonly, due to delays in implementation of the interventions at the sites, enrollment was not evenly distributed across time with some cohorts enrolling faster than others. Figure 1a-g below demonstrate this for the six cohorts. These figures show that the FTIC, ACs and DMD cohorts were the fastest to enrol, reaching targets at some sites in about 16 weeks and at all sites within 24 weeks. While enrollment was largely steady week on week in these cohorts, enrollment accelerated between week 8 and 9 for the Adherence Clubs and DMD cohorts. This acceleration was largely due to enrollment at control sites in pairs where enrollment of the cohort at the intervention site was delayed and resolution of some register issues that had previously delayed enrollment. Each of these cohorts required screening more subjects than we ultimately enrolled. Subjects who were not eligible are described below. The EAC and TRIC cohorts took longer to enrol (up to 26 weeks) and by the end of enrollment on 16 December 2016 it had not been possible to enrol the full EAC cohort. The most prominent reasons enrollment was slower or incomplete for these cohorts was related to: 1) interventions not being implemented to AGL specifications (e.g. focussing EAC on defaulters rather than unsuppressed patients, or only tracing pre-ART or patients who had defaulted rather than those who had early or late missed appointments); and 2) incomplete or unavailable registers at intervention sites. We also noted that for the TRIC cohort, a larger number of patients who were screened were found to be ineligible compared to other cohorts. This was due in large part to the intervention not being delivered as per the AGL specifications. As anticipated, the Tuberculosis, Hypertension and Diabetes cohorts are taking the longest to complete. We have currently identified all 2400 eligible patients for the TBHD Screening cohort and are in the process of identifying those who screen positive who will be followed as part of this cohort. By January 13, 2017, we had also identified 577 (24%) subjects as eligible for enrollment into the TBHD Diagnosed cohort. However as described on page 13, we have not yet initiated cohort enrollment via the third method for the TBHD diagnosed cohort but will be able to do this retrospectively once it is initiated. We anticipate that enrollment of these two cohorts will be completed by March 2017. 20 SECOND ENROLLMENT REPORT Figure 1 (Cohort 1‒6a, 6b) Screening and enrollment by cohort over time compared to the target total through December 16th, 2016 21 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA 8.4 Enrollment by facility/district In addition to variation in enrollment over time by cohort, we also encountered variation in enrollment by site and province. While overall all five intervention cohorts have nearly completed (with the possibility of some retrospective enrollment if feasible), Figure 2 shows that enrollment was uneven with facilities in North West province enrolling fastest and Limpopo and KwaZulu Natal taking somewhat longer. This was largely due to delays in being able to enrol for one or more cohorts because of implementation or register issues. Figure 2a‒d Enrollment over time by province and clinic a b c d Figure 3 below shows enrollment into each cohort within each district. Note that the large difference between 6A and 6B cohort numbers in Mopani is due to only having screening information for TB patients (for hypertension and diabetes cases, no records are created until patients are diagnosed). Because the third route of detecting the diagnosed patients has not been implemented yet, the number of diagnosed patients in Figure 3 only contains those Mopani patients diagnosed TB positive who were identified from those screened. This will change as we implement the third method of identifying diagnosed patients. 22 SECOND ENROLLMENT REPORT Figure 3 Enrollment by District and Study Cohort 8.5 Ineligible Patients As part of the enrollment process we have screened subjects for eligibility and identified subjects who were ineligible upon file review. Below in Table 6 we describe the number of ineligible subjects per cohort. We note that for all cohorts except DMD (where the intervention was delivered in control sites) more subjects were screened in the intervention cohorts than the control cohorts. This was expected as interventions subjects had additional eligibility criteria, namely getting the intervention. The most common reason for not being eligible in the intervention cohorts was a patient not actually receiving the intervention for which they were eligible. This was most common with FTIC where 89 patients (46% of all those excluded) were eligible for FTIC but did not receive it. For all cohorts, between 4% (Early Adherence Counselling controls) and 15% (FTIC intervention) of files were missing resulting in the need to exclude patients. Not being able to locate a patient file was more common in intervention sites than control sites, but this is largely due to the fact that more files needed to be screened in interventions sites due to the additional screening criteria. Very few patients were found ineligible due to the other general eligibility criteria (e.g. <18 years old, not resident, pregnant) suggesting that TIER.Net data was reasonable for screening patients for eligibility. Of note, in the TRIC cohort, the numbers screened were similar in the intervention and control arms. This was because many patients were found ineligible in this cohort because TIER.net data was not up to date given the time between the TIER.Net dispatch date and file screening, causing us to think a patient had missed a visit on our initial screen when clinic files confirmed they had not. 23 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA While this was more common in control clinics, this was balanced out in the intervention cohort by patients who did not receive the intervention. Table 6 Ineligible subjects by cohort and reason for exclusion Cohort 1: FTIC Cohort 2: AC Cohort 3: DMD Cohort 4: EAC Cohort 5: TRIC Control Intervention Control Intervention Control Intervention Control Intervention Control Intervention Reason not eligible N (%) N (%) N (%) N (%) N (%) N (%) N (%) N (%) N (%) N (%) Total screened (eligible + ineligible) 445 564 340 444 441 361 564 689 479 476 STUDY CRITERIA File not found 6 (8%) 29 (15%) 4 (9%) 12 (7%) 9 (10%) 15 (12%) 2 (4%) 36 (11%) 12 (7%) 27 (13%) Not 18yrs 0 (0%) 0 (0%) 1 (2%) 0 (0%) 11 (12%) 0 (0%) 9 (16%) 16 (5%) 6 (3%) 9 (4%) Not resident 6 (8%) 3 (2%) 0 (0%) 0 (0%) 0 (0%) 1 (1%) 1 (2%) 1 (0%) 2 (1%) 0 (0%) Intention to transfer 5 (6%) 2 (1%) 0 (0%) 2 (1%) 1 (1%) 1 (1%) 5 (9%) 1 (0%) 14 (8%) 4 (2%) Pregnant 8 (10%) 12 (6%) 2 (4%) 3 (2%) 6 (6%) 0 (0%) 6 (11%) 12 (4%) 11 (6%) 12 (6%) Unknown 0 (0%) 0 (0%) 2 (4%) 0 (0%) 2 (2%) 0 (0%) 1 (2%) 0 (0%) 2 (1%) 0 (0%) COHORT-SPECIFIC CRITERIA COHORT 1: FTIC Not HIV positive 3 (4%) 2 (1%) TB Diagnosis 7 (9%) 7 (4%) Not eligible in last 30d 28 (36%) 17 (9%) Not ART Naïve 14 (18%) 34 (17%) Not receiving intervention 0 (0%) 89 (46%) COHORT 2: AC & COHORT 3: DMD Not 12-36 mos on ART 6 (13%) 32 (19%) 5 (5%) 30 (23%) ART change in last 12mos 6 (13%) 7 (4%) 3 (3%) 7 (5%) Last 2 VL not suppressed 20 (43%) 41 (24%) 28 (30%) 37 (28%) No VL in last 3 mos 3 (7%) 21 (12%) 5 (5%) 9 (7%) Not in AC / DMD 2 (4%) 50 (30%) 4 (4%) 30 (23%) Receiving other RPS than randomized 0 (0%) 1 (1%) 19 (20%) 0 (0%) COHORT 4: EAC No ART 3 mos 7 (12%) 14 (4%) No VL > 400 18 (32%) 50 (15%) Not on EAC register 2 (4%) 192 (58%) No VL in last 3 mos 6 (11%) 7 (2%) COHORT 5: TRIC 7 (12%) 14 (4%) Not ART Initiated 5 (3%) 11 (5%) Didn't miss last visit* 116 (66%) 63 (31%) Did not receive tracing 9 (5%) 80 (39%) Total ineligible 77 195 46 169 93 130 57 329 177 206 Total enrolled 368 369 294 275 348 231 507 360 302 270 Note: * For this group, often these patients appear on TIER as eligible but by the time we visit the site, clinical data has been updated and/or the patient has returned and so they are no longer eligible. COHORT DESCRIPTIVES Now that enrollment has been nearly completed (with the possibility of some retrospective enrollment if feasible), we have baseline data on subjects in each of the cohorts. Data come from TIER.net and patient files and have been entered into an electronic database for descriptive analysis 24 SECOND ENROLLMENT REPORT in STATA. Below in Table 7 we describe each of the cohorts in terms of their baseline characteristics, both overall and by intervention arm. Table 7 Baseline characteristics of the cohorts COHORT 1: FAST TRACK INITIATION COUNSELLING FTIC Control FTIC Intervention FTIC Total N=368 N=369 N=737 Characteristic n (%) n (%) n (%) Age (n=737) 18-29 87 (24%) 97 (26%) 184 (25%) 30-39 142 (39%) 150 (41%) 292 (40%) 40-49 84 (23%) 79 (21%) 163 (22%) 50+ 55 (15%) 43 (12%) 98 (13%) Gender (n=737) Female 214 (58%) 218 (59%) 432 (59%) Male 154 (42%) 151 (41%) 305 (41%) CD4 Count (at ART initiation) (n=714) (median, IQR) 234 (123-357) 210 (108-358) 224 (117-357) TB status (n=737) Current TB diagnosis 0 (0%) 0 (0%) 0 (0%) No current TB diagnosis 368 (100%) 369 (100%) 737 (100%) Note: *Totals for individual variables may differ because some observations have been dropped due to missing or out-of-range values. These are undergoing review to determine the correct value. The FTIC cohort participants are mostly under age 40 (65%). As is typical of a public-sector HIV treatment population, the cohort is more likely to be female (59%) though slightly less than the 66% we typically see (see Egger et al. Int Journal Epidemiol 2011; 41: 1256-1264 Table 2 for example). The average CD4 count at ART initiation in this cohort is also barely above the initiation threshold when the ART program started (<200) with an average of 224 cells/ml3 at ART initiation, suggesting that our population is more immunocompromised than the general population of patients eligible for ART. This is expected as FTIC is likely to be prioritized for sicker patients even if not as per the SOP. Despite being a population that by definition was expected to be sicker than the general population, by design we did not enrol any patients with tuberculosis in this cohort. For FTIC, the intervention and control cohorts were largely balanced on baseline characteristics with only minor differences between groups, much like what would be expected in a moderate sized individually randomized trial. We did observe some small difference in the CD4 count medians, with higher CD4 counts among those in the control group although only by 24 cells. This is also not surprising as it is not possible to know who in the control group would have gotten the intervention had the intervention been delivered there and likely sites prioritized the sickest. Still, it is reassuring the differences are so small. In addition, we note that there were only 23 subjects (3%) for whom we could not find a baseline CD4 count, far less than we typically see in observational cohorts in South Africa (20% in the IeDEA cohort, Egger et al. 2011). 25 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA Table 8 Adherence clubs AC Control AC Intervention AC Total N=294 N=275 N=569 Characteristic n (%) n (%) n (%) Age (n=569) 18-29 61 (21%) 58 (21%) 119 (21%) 30-39 108 (37%) 100 (36%) 208 (37%) 40-49 68 (23%) 72 (26%) 140 (25%) 50+ 57 (19%) 45 (16%) 102 (18%) Gender (n=569) Female 204 (69%) 206 (75%) 410 (72%) Male 90 (31%) 69 (25%) 159 (28%) CD4 Count (at ART initiation) (n= 263 control; 240 intervention)* 282 (170-411) 255 (146-357) 268 (157-379) Viral Load (copies/ml) (median, IQR) (n=569) 50 (20-124) 50 (20-124) 50 (20-125) log10 Viral Load (copies/ml) (median, IQR) (n=569) 1.70 (1.30-2.09) 1.70 (1.30-2.09) 1.70 (1.30-2.09) TB status (n=569) Current TB diagnosis 1 (1%) 0 (0%) 1 (1%) No current TB diagnosis 293 (99%) 275 (100%) 568 (99%) Time on ART at enrollment (days) (median, IQR) (n=569) 577 (472-860) 839 (551-1163) 714 (506-938) Note: * We note that the adherence club cohort is not limited to those who were treatment naïve, so some patients could have been transfer-in patients without a baseline CD4 count. Others may have had a lost file and no record of the baseline CD4 count. The NHLS data we have access to for this study only contains lab results from April 2016 onward, so we are not able to find baseline CD4 counts for those in adherence clubs as we require them to be on ART for more than one year. For the adherence clubs, a little over half of patients were under age 40 (58%) and just above 70% were female. These patients were somewhat healthier at ART initiation than those in the FTIC cohort at an average of 268 cells, but with a fairly wide range (157-379). We had nearly complete viral load data in this cohort as this was necessary to confirm eligibility for the interventions. Baseline CD4 count was missing from 66 subjects (12%) closer to but still below what we observe in the literature for observational HIV cohorts (again see Egger et al. Int Journal Epidemiol 2011; 41: 1256-1264 Table 2 for example). As noted, these cannot be updated using the NHLS data because we only have data from April 2016 onwards and these patients are not new on ART. The intervention and control cohorts were similar with respect to the demographic variables (age and sex) and with respect to log viral load at time of eligibility, which is by design (patients must be supressed to be eligible for the intervention). We did however observe a difference between the cohorts in terms of CD4 count at treatment initiation with the control cohort being somewhat healthier at ART initiation (282 vs. 255). As with FTIC, the differences were small and unlikely to impact the results. We also saw differences between groups in the duration on treatment at enrollment with the control population on treatment for a shorter period of time than the intervention population. This suggests that the intervention sites were either targeting the ACs towards patients who were on treatment for a longer time than eligibility criteria required or that patients who were on treatment for longer were more interested in this intervention. 26 SECOND ENROLLMENT REPORT Table 9 Decentralized Medicine delivery (DMD) DMD Control DMD Intervention DMD Total N=348 N=231 N=579 Characteristic n (%) n (%) n (%) Age (n=579) 18-29 67 (19%) 38 (16%) 105 (18%) 30-39 117 (34%) 90 (39%) 207 (36%) 40-49 100 (29%) 69 (30%) 169 (29%) 50+ 64 (18%) 34 (15%) 98 (17%) Gender (n=579) Female 241 (69%) 168 (73%) 409 (71%) Male 107 (31%) 63 (27%) 170 (29%) CD4 Count (at ART initiation) (control n=297, 279 (142-387) 259 (133-346) 269 (139-366) intervention n=218) Viral Load (copies/ml) (median, IQR) (n=577) ** 42 (20-100) 124 (35-124) 50 (20-124) log10 Viral Load (copies/ml) (median, IQR) (n=577) 1.62 (1.30-2.22) 2.09 (1.54-2.09) 1.69 (1.30-2.09) TB status at study enrollment (n=574) Current TB diagnosis 0 (0%) 1 (1%) 1 (1%) No current TB diagnosis 343 (100%) 230 (99%) 573 (99%) Time on ART at enrollment (days) (median, IQR) (n=579) 756 (488-916) 797 (498-951) 769 (491-935) Note: ** 2 viral loads were not found. As this is an inclusion criteria, we will need to verify if these can be found and if not, they will not be included in the final dataset. The DMD cohort also had roughly half of the patients below age 40 (54%) and had many more females (71%) than males. As we would expect given the interventions had the same eligibility criteria, the cohort was similar to the Adherence Clubs cohort with respect to ART initiation CD4 count (269 cells), though we note that this is not the CD4 count at the time of enrollment into the cohort. As with Adherence Clubs, we were missing a baseline CD4 count on 64 (11%) of subjects, typical of but lower than, most HIV programs in South Africa. The median viral load at baseline was also very low as would be expected given the eligibility criteria for the intervention (i.e. stable patients). The intervention and control cohorts were well balanced with respect to sex and age but there were some small imbalances in CD4 count at ART initiation (279 vs. 259) and log viral load at eligibility (1.62 vs 2.09). These don’t appear too large, but it is something we will pay attention to when we conduct the analysis. Interestingly, here the time on treatment was very similar between groups, suggesting that unlike Adherence Clubs, DMD was not being targeted towards those on treatment longer or shorter than the eligibility criteria would require. The median time on ART for both groups was also much more similar to the intervention group in the Adherence Club cohorts, suggesting overall that patients who got one of the two interventions were on treatment a fair bit longer than required (a minimum of one year), likely as clinics processed a backlog of eligible patients as part of the Decanting Strategy. In addition, due to the Decanting Strategy some of our DMD control sites were evaluation intervention sites, and as such we would exclude any patient from our DMD group who had been enrolled in an Adherence Club. This likely would have excluded those who had been on ART for shorter periods of time. 27 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA Table 10 Enhanced Adherence Counselling (EAC) EAC Control EAC Intervention EAC Total N=507 N=360 N=867 Characteristic n (%) n (%) n (%) Age (n=867) 18-29 56 (11%) 56 (16%) 112 (13%) 30-39 184 (36%) 138 (38%) 322 (37%) 40-49 150 (30%) 103 (29%) 253 (29%) 50+ 117 (23%) 63 (18%) 180 (21%) Gender (n=867) Female 310 (61%) 212 (59%) 522 (60%) Male 197 (39%) 148 (41%) 345 (40%) CD4 Count (at ART initiation) (n=437 control; 323 intervention)*** 163 (79-270) 146 (79-261) 157 (79-269) Viral Load (copies/ml) (median, IQR) (n=864)** 3550 (914-36,000) 11712 (2010-67200) 5256 (1145-46,300) log10 Viral Load (copies/ml) (median, IQR) (n=864) 3.55 (2.96-4.55) 4.06 (3.30-4.82) 3.72 (3.05-4.67) TB status at study enrollment (n=866) Current TB diagnosis 2 (1%) 2 (1%) 4 (1%) No current TB diagnosis 505 (99%) 357 (99%) 862 (99%) Time between last viral load and enrollment (days) (median, IQR)* (n=867) 54 (34-77) 58 (33-92) 55 (34-83) Note: * Time between last viral load and enrollment. It is likely patients are only enrolled in EAC at their next visit after becoming eligible through an elevated viral load unless there is a specific intervention at the site to follow up patients with high viral loads (if the patient is only returning every other month to pick up medication then these times seem reasonable, as the EAC SOP says “an EAC identified file should trigger referral for EAC as soon as the patient comes back to the facility”). It is p ossible that an EAC patient may be contacted in between telling them they need EAC but this is not recorded; ** We note that 3 viral loads were not found. As this is an inclusion criteria, we will need to verify if these can be found and if not, they will not be included in the final dataset; *** We note that the enhanced adherence counselling cohort is not limited to those who were treatment naïve, so some patients could have been transfer-in patients without a baseline CD4 count. Others may have had a lost file and no record of the baseline CD4 count. The NHLS data we have access to for this study only contains lab results from April 2016 onward, so we are not able to find baseline CD4 counts for those in adherence clubs as we require them to be on ART for more than one year. The EAC cohort was very evenly divided between those below and above 40 (50%) but slightly fewer females (60%) than we typically see in ART cohorts, possibly suggesting that males may be more likely to have elevated viral loads in these cohorts than in the general population. The cohort was quite sick at ART initiation with an average CD4 count below 200 cells/ml3 (157 cells/ml3). The cohort was eligible for the intervention for a median of 55 days at enrollment, suggesting a little over a one month delay between eligibility and enrollment. We believe most patients who get EAC only get it after returning to the clinic from a previous visit where a blood draw for viral load monitoring found them to have met the eligibility criteria. As with the previous two interventions we had near complete data on viral load at enrollment as this was necessary to determine eligibility. We were missing CD4 counts at ART initiation for 107 (12.4%) still below the 20% we typically observe in our observational cohorts. The intervention and control cohorts were well balanced with respect to age and sex and tuberculosis enrollment. There were small differences in CD4 count (163 vs 146 cells/ml3). Log viral load was also somewhat different and lower in the control cohort (log10 3.55 vs. 4.06). This suggests that despite the eligibility criteria, EAC was being targeted towards those with the highest viral loads while those with lower, but still eligible viral loads are not being given the same priority. 28 SECOND ENROLLMENT REPORT Table 11 Early Tracing (TRIC) TRIC TRIC Control TRIC Total Intervention N=302 N=270 N=572 Characteristic n (%) n (%) n (%) Age (n=572) 18-29 55 (18%) 67 (25%) 122 (21%) 30-39 102 (34%) 111 (41%) 213 (37%) 40-49 82 (27%) 63 (23%) 145 (25%) 50+ 63 (21%) 29 (11%) 92 (16%) Gender (n=572) Female 192 (64%) 182 (67%) 374 (65%) Male 110 (36%) 88 (33%) 198 (35%) CD4 Count (at ART initiation) (control n=268, 211 (114-331) intervention n=244) 209 (124-327) 215 (108-335) Last Viral Load (copies/ml) (median, IQR) (control n=232, intervention n=179) 100 (20-159) 100 (20-270) 100 (20-200) log10 last Viral Load (copies/ml) (median, IQR) (control n=232, intervention n=179) 2.00 (1.30-2.19) 2.00 (1.30-2.43) 2.00 (1.30-2.30) TB status (n=567) Current TB diagnosis 2 (1%) 2 (1%) 4 (1%) No current TB diagnosis 296 (99%) 267 (99%) 563 (99%) Time between last missed visit and enrollment (days)* (median, IQR) (n=566) 32 (17-58) 85 (29-131) 45 (21-99) Note: * Most patients are only enrolled at their next visit between last missed visit and enrollment (visit can be missed by up to 90 days). The patients need to be eligible for tracing from May through October. Therefore if someone missed a scheduled visit in February, they would be eligible for tracing through May, and only enrolled in October, resulting in 243 days between missed visit and enrollment. The TRIC cohort was also roughly divided between those below (58%) and above 40 (42%) and very close to what we expect for percent females at 65% suggesting those that receive TRIC are similar to the general clinic populations. As with previous cohorts the patients were quite sick at ART initiation with an average CD4 count of 211 cells/ml3, though higher than the EAC cohort. In this cohort 60 (10.4%) patients had a missing CD4 count at ART initiation, lower than what we typically observe in our HIV cohorts. We note here that there was a substantial amount of missing viral load data, but this is explained by the fact that this cohort is not required to have a viral load to be eligible, and many patients in this cohort have left care and are therefore expected to not have had a viral load. In addition, patients may also have not reached the point of needing a viral load (e.g. if they missed an appointment between initiation and 6 months). Overall the last viral load was low in this cohort, suggesting that many of the patients lost and being traced (or eligible to be traced) had not been experiencing an elevated viral load just prior to missing a visit. The cohort was eligible for the intervention for an average of 45 days at enrollment (days between last missed visit and enrollment to the cohort, not time between missed visit and tracing attempt). The intervention and control cohorts were very well balanced with respect to CD4 at ART initiation and log of last viral load. However they did show some differences, with younger patients in the intervention cohort and the intervention cohort being eligible for the intervention substantially longer than the control cohort (32 vs. 85 days). This likely reflects the fact that we did retrospective enrollment into the cohorts rather than reflecting any meaningful differences. 29 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA Table 12 Tuberculosis, Hypertension and Diabetes - Screening Cohort (TBHD) HIV Negative HIV Positive Total TBHD- Screening N=1385 N=1015 N=2400 Characteristic n (%) n (%) n (%) Age 18-29 271 (20%) 180 (18%) 451 (19%) 30-39 198 (14%) 326 (32%) 524 (22%) 40-49 218 (16%) 280 (28%) 498 (21%) 50-59 289 (21%) 158 (16%) 447 (19%) 60+ 409 (30%) 71 (7%) 480 (20%) Gender Female 895 (65%) 670 (66%) 1565 (65%) Male 490 (35%) 345 (34%) 835 (35%) Any TB Screening at last visit Yes 292 (21%) 794 (78%) 1086 (45%) No 1093 (79%) 221 (22%) 1314 (55%) Any Diabetes screening at last visit Yes 51 (4%) 6 (1%) 57 (2%) No 1334 (96%) 1009 (99%) 2343 (98%) Any Hypertension screening at last visit Yes 117 (8%) 131 (13%) 248 (10%) No 1268 (92%) 884 (87%) 2152 (90%) Screening at last visit- combined No screening 962 (69%) 183 (18%) 1145 (48%) TB screening 260 (19%) 697 (69%) 957 (40%) Diabetes screening 39 (3%) 3 (0%) 42 (2%) Hypertension screening 89 (6%) 35 (3%) 124 (5%) TB and Diabetes screening 7 (1%) 1 (0%) 8 (0%) TB and Hypertension screening 23 (2%) 94 (9%) 117 (5%) Hypertension and Diabetes screening 3 (0%) 0 (0%) 3 (0%) TB, Hypertension, and Diabetes screening 2 (0%) 2 (0%) 4 (0%) Any recent TB diagnosis Yes 13 (1%) 17 (2%) 30 (1%) No 1372 (99%) 998 (98%) 2370 (99%) Any recent Diabetes diagnosis Yes 91 (7%) 3 (0%) 94 (4%) No 1294 (93%) 1012 (100%) 2306 (96%) Any recent Hypertension diagnosis Yes 182 (13%) 36 (4%) 218 (9%) No 1203 (87%) 979 (96%) 2182 (91%) Recent diagnosis- combined (enrolled in cohort 6a and 6b) No diagnosis 1151 (83%) 961 (95%) 2112 (88%) TB diagnosis 13 (1%) 17 (2%) 30 (1%) Diabetes diagnosis 39 (3%) 1 (1%) 40 (2%) Hypertension diagnosis 130 (9%) 34 (3%) 164 (7%) Hypertension and Diabetes diagnosis 52 (4%) 2 (1%) 54 (2%) Overall, the TBHD cohort was older than the previous cohorts (only 41% under age 39) but their sex distribution looked exactly like an ART cohort, suggesting that ART cohorts are much like the general population of patients seeking care at a PHC. TB screening was common in this cohort (with 45% receiving TB screening at last visit) with only 10% getting hypertension screening and only 2% receiving diabetes screening. This differed strongly by HIV status, however, as would be expected. Those with HIV were much more likely to get TB screening at last visit than those who did not have HIV (78% vs 21%). Overall, those with HIV were much less likely not to receive any screening (18% with HIV vs 69% without did not receive any screening, though again this was 30 SECOND ENROLLMENT REPORT largely driven by TB screening). Still even among hypertension there was some suggestion that screening was better among those with HIV (13%) compared to those without (8%), though the numbers were small overall. The opposite was true for diabetes however, with the HIV cohort receiving almost no screening (4% vs 1%). Among all patients, 79% had no diagnosed condition at last visit, though this included the 48% of patients who received no screening. Diagnosis of all conditions was rare, with 1% diagnosed with TB, 2% diagnosed with diabetes and 7% diagnosed with hypertension. The HIV negative cohort was less likely to have no condition diagnosed than the HIV positive cohort (83% vs 95%). This might seem counterintuitive, but likely reflects the fact that more patients in the HIV cohort would already have been diagnosed with a condition. The most common condition diagnosed among the HIV negatives was hypertension (9%). Table 13 Tuberculosis, Hypertension and Diabetes - Diagnosed Cohort HIV Negative HIV Positive Total TBHD- Diagnosed N=479 N=98 N=577 Characteristic n (%) n (%) n (%) Age 18-29 23 (5%) 11 (11%) 34 (6%) 30-39 50 (10%) 34 (35%) 84 (15%) 40-49 72 (15%) 22 (22%) 94 (16%) 50-59 121 (25%) 21 (21%) 142 (25%) 60+ 213 (44%) 10 (10%) 223 (39%) Gender Female 305 (64%) 52 (53%) 357 (62%) Male 174 (36%) 46 (47%) 220 (38%) Screening at last visit No screening 343 (72%) 55 (56%) 398 (69%) TB screening 105 (22%) 36 (37%) 141 (24%) Diabetes screening 12 (3%) 3 (3%) 15 (3%) Hypertension screening 13 (3%) 4 (4%) 17 (3%) TB and Hypertension screening 5 (1%) 0 (0%) 5 (1%) Hypertension and Diabetes screening 1 (0%) 0 (0%) 1 (0%) Recent diagnosis TB diagnosis 33 (7%) 47 (48%) 80 (14%) Diabetes diagnosis 65 (14%) 1 (1%) 66 (11%) Hypertension diagnosis 305 (64%) 48 (49%) 353 (61%) Hypertension and Diabetes diagnosis 76 (16%) 2 (2%) 78 (14%) The diagnosed cohort is still in recruitment, so it is difficult to interpret the data. However, of the 577 subjects enrolled, very few were young with only 21% being under 40 years of age. This would be consistent with the age distribution of these comorbid conditions. In addition, more males have been diagnosed than females, again consistent with what we know about tuberculosis rates. 31 EVALUATION OF THE NATIONAL DEPARTMENT OF HEALTH'S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA FOLLOW UP PROCESSES With the HIV cohorts now nearing completion of enrollment (with only possible retrospective enrollment remaining), we will continue to follow up patients through passive record review while continuing to work with the sites to maintain quality data and to monitor data systems for study outcomes. The short term and long-term outcomes are described in Table 8 below. Table 14 Study evaluation outcomes for protocol 1 Cohort Short-term Outcome Long-term Outcomes Fast track ART initiation % who initiate ART within 30 days of becoming ART % of patients virally suppressed (< 400 copies/ml3) within counseling eligible 9 months of ART eligibility Adherence clubs % who receive all medications within the first four % virally suppressed (< 400 copies/ml3) at twelve months months after eligibility after club eligibility Decentralized medicine % virally suppressed suppression (< 400 copies/ml3) 12 delivery % who receive all medications within 3 months months Enhanced adherence % who resuppress their viral load (< 400 copies/ml3) % who resuppress their viral load (< 400 copies/ml3) counseling within 3 months of eligibility within 12 months of eligibility Early tracing of patients lost to follow up % who return to care within 3 months of eligibility % who return to care within 12 months of eligibility TB, hypertension, and For each condition, % of patients who have 80% visit For each condition, % of patients who achieve disease diabetes compliance in first 3 months after diagnosis control at the six-month visit after diagnosis We will continue to monitor TIER.net and work with the sites to collect follow up data for patients enrolled in the study. The Gantt chart below in Figure 4 show the timeline for follow up and completion of data collection for protocol 1. 10.1 Data collection for short-term endpoint for Fast Track Initiation Counselling The short-term outcome for FTIC requires a 30-day follow up for each patient to assess the percentage who initiate ART. FTIC follow up occurs through TIER.net through identification of the initiation visit with verification through patient registers if needed. FTIC follow up has been completed for 81% of patients who have currently been enrolled. All subjects are expected to reach a short-term outcome by February 2017. At that point we will need to verify data against patient files for any subjects with missing data in TIER (e.g. should a backlog occur) with an expected date of having all short-term outcomes by March 2017. Below are outcomes for the patients who have reached a primary short term outcome for FTIC. Currently 84% of patients in intervention sites and 81% of patients in control sites initiated ART within 30 days of eligibility. 32 SECOND ENROLLMENT REPORT Table 15 Short-term outcomes (ART initiation within 30 days) for those eligible for FTIC cohort INTERVENTION CONTROL Initiated Initiated Eligible for within % Eligible for within % Facility Total outcome 30 days initiated Facility Total outcome 30 days initiated Motsamai Clinic 28 25 18 72% Tamaho Clinic 29 14 10 71% Phola Park CHC 31 8 6 75% Ramokonopi CHC 30 23 17 74% Khumalo Clinic 28 26 26 100% Zonkizizwe 1 Clinic 30 21 11 52% Grace Mugodeni CHC 30 29 23 79% Motupa Clinic 35 25 20 80% Giyani CHC 30 25 22 88% Dzumeri Clinic 30 28 25 89% Tzaneen Clinic 27 24 15 63% Nkowankowa CHC 30 27 19 70% Letlhabile CHC 32 26 23 88% Wonderkop Clinic 30 30 25 83% Hebron Clinic 31 27 23 85% Majakaneng Clinic 30 30 26 87% Tlhabane CHC 30 30 26 87% Bafokeng CHC 30 29 26 90% King Dinizulu Clinic 34 29 26 90% Nkwalini Clinic 34 32 29 91% Thokozani Clinic 37 26 22 85% Nseleni CHC 30 27 22 81% Buchanana Clinic 31 16 13 81% Ntambanana Clinic 30 26 24 92% Total 369 291 243 84% Total 368 312 254 81% 33 Figure 4 Gantt Chart of Timeline for Project EVALUATION OF THE NDOH’S NATIONAL ADHERENCE GUIDELINES FOR CHRONIC DISEASES IN SOUTH AFRICA 34 2016 2017 Project Management (Protocol 1&2) June July August SeptemberOctober NovemberDecember January February March April May June July August SeptemberOctober November December Activities, Deliverables and Milestones % Complete 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4 Key Milestones Secure all permissions and approvals 100% Recruit and train provincial teams 100% Site assessments 100% Data enhancement 100% Finalize Protocol 1&2 data collection tools 100% Protocol 1: Enrolment HIV Cohorts 96% Protocol 1: Enrolment TBHD Cohort 62% Protocol 1: Cohort Follow-up 8% Protocol 2: Patient interviews 46% Protocol 2: Patient FGDs 16% Protocol 2: Provider interviews 30% Protocol 2: Data collection for objective 1 0% Analysis and Reporting 60% CONCLUSIONS Enrollment into the HIV cohorts has now completed with enrollment into the TBHD cohort expecting completion in March 2017. While there have been challenges to getting the cohort enrolled, including delays in implementation of the interventions and changes to the way the interventions have been implemented, the team has been able to enrol subjects into the cohorts. These challenges have been overcome through a combination of support from the World Bank and NDoH PIs to working with the sites to ensure accurate record keeping and access to registers. The cohorts are now entering the follow up phase for the HIV cohorts and work will shift to a mix of maintaining control over the quality of the cohort data, completing enrollment into the TBHD cohorts and to continue with protocol 2 where we will collect largely qualitative data from patients and providers to better understand how the rollout is occurring and the impacts that it is having. 35