Operations Research of the 'Real World' Effectiveness of Multi-Month Dispensing of ART for Stable Patients in CARGs in Zimbabwe

This study will use a cluster (facility) randomized trial with 3 arms; patients receiving 3 monthly ART supply at facilities-standard of care (3MF) (Control group), patients receiving 3 monthly ART supply at newly formed CARGs (3MC) and patients receiving 6 monthly ART supply at newly formed CARGs (6MC). Participants will be followed up for 24 months. A total of 30 study clusters (facilities) will be randomization, into the three study arms; 10 clusters per arm stratified 2 urban, 8 rural.

Study Setting and study population The study population is HIV-infected adults over the age of 18 years recruited from health facilities in U.S. President's Emergency Plan for AIDS Relief (PEPFAR) Zimbabwe scale-up districts (using existing national records of patients in the ART registers) that have been stable on a standard first-line ART treatment regimen for at least 6 months.

Facilities from Chitungwiza Municipality, Matebeleland South Province, Masvingo Province and Mberengwa district will be purposively selected to be the urban/periurban and rural settings in the study. These areas were selected as they have PEPFAR Zimbabwe scale-up priority facilities, high numbers of patients on ART and have prioritized implementation of CARGs. These locations are also viewed to be good representations of PEPFAR Zimbabwe scale-up priority facilities, hence are likely to improve the generalizability of the study findings to the study population. As contexts differ between rural and urban/periurban settings, it is important to determine the effectiveness of MMD within CARGs within both contexts.

Sampling and Sample size Sample sizes were determined for a cluster-randomized noninferiority trial, using PASS v.14 (TM) software. Sample size estimates were calculated for the primary outcome of retention in care at 12 months. The probability of patient attrition twelve months after study enrolment in the control group (3MF) is assumed to be 5%, derived from the relative difference in patient retention between 12 and 24 months from the 2015 Zimbabwe UNAIDS Global AIDS Response progress report. An intracluster correlation coefficient (ICC) of 0.01 for patient retention/attrition amongst stable ART patients is assumed. The noninferiority limit is specified as 3.25%. Assuming α=0.05, power of 85%, using one-sided Z-test (unpooled) test statistic, 182 participants per facility will be required with a total sample size of 5460 participants. A sample size re-estimation during accrual will increase the total sample size to approximately 5760; 192 enrolled participants per study site and 1,920 participants per arm. As retention in care is the primary outcome, no adjustment for loss to follow-up needs to be made.

Study procedures As this is operational research, the study will not interfere with routine patient management and will follow national guidelines. The health care worker assesses each patient's clinical notes to establish stability and study inclusion. Once informed consent is obtained, a baseline questionnaire will be applied to record demographic characteristics and the patients clinical and ART history. Patients will receive care based on their study ARM and relevant SOP. Patient files and CARG tools will be reviewed and data collected at each ART refill. Viral loads will be repeated annually.

Data Collection The study will use both qualitative and quantitative methods. The data sources will include patient clinic notes, electronic patient management system (ePMS), CARG monitoring data collection tools. In addition, variables to control for confounding in the analysis for outcomes of interest will be collected at baseline through a short cross-sectional survey. A pre-and-24-month post survey time-flow evaluation of patient waiting times to assess gains of facility decongestion will be conducted. Feasibility of MMD of ART within CARGs will be measured prospectively at baseline, 12 and 24 months. To understand the acceptability of CARGs by patients and service providers; patient satisfaction, and improvement in quality of care at the facilities, 20 patient FGDs and 20 heath care provider KII will be conducted at baseline, 12 and 24 months. A micro-costing approach supplemented by a macro-costing approach of fixed costs will be utilized to measure cost outcomes. A resource use form will collect information per patient through retrospective review of clinic records. Start up costs will be included under each arm. Data to assess patient level costs will be collected from a randomly selected sub-sample of 365 patients per arm at baseline, 12 months and 24 months.

Data analysis Quantitative data will be analyzed using STATA 14. An intention-to-treat analysis will be performed for the primary outcome of retention in care, in which all patients and clinics will be analyzed in the group to which they were originally assigned. Risk differences between arms will be compared with binomial population averaged generalized estimating equations using an identity link and an exchangeable correlation matrix, specifying for clustering by facility. Viral load suppression analyzed using log-binomial regression, generalized estimating equations will be used to estimate risk ratios between the study arms, specifying for clustering by facility and using an exchangeable correlation matrix. Average patient waiting times will be compared between study arms using linear regression.

Content qualitative analysis will be conducted using Atlas Ti, A cost outcomes analysis will be conducted from the provider perspective. Average cost per yield rates (cost per client retained in care, cost per client virally suppressed, cost per patient per year) will be calculated. An incremental cost-effectiveness ratio (ICER) for the will be calculated; ICER threshold comparison as recommended by World Health Organisation (WHO) will be employed. Average patient costs will be from the patient's perspective. Startup costs for each service delivery model will also be analyzed.

RISKS / BENEFITS TO PARTICIPANTS There may be no direct benefit to participants who take part in this study, beyond benefits related to their routine ART medication that would have been achieved otherwise. Information learned in this study may be of benefit to participants and others in the future, particularly information that may lead to optimized treatment guidelines for HIV-infected stable patients on ART.

As this is operational research, study procedures are routine clinical care associated with minimal risk to participants. All patients will attend routine clinic visits to be assessed for new Opportunistic infections (OIs), pregnancy, drug-drug interactions or side effects of medications. This will ensure that patients who develop co-morbidities or become pregnant during the study receive standard care, although they will not be excluded from the study.

REIMBURSEMENTS AND COMPENSATION As participants of FGDs may use transport money to the discussion venue, this will be reimbursed. Those who are part of a FGD will receive snacks after the discussion. CARGs will receive a bag to carry drugs from the clinic back to their communities. This will be made clear to all study participants during the informed consent process.

CONFIDENTIALITY ASSURANCES Participant information will not be released without written permission to do so except as necessary for review, monitoring, and/or auditing. All study-related information will be stored securely. Participant research records will be stored in locked areas with access limited to study staff. case record files (CRFs), and other documents that may be transmitted off-site will be identified by patient identifier only. Likewise, communications between study staff and the investigators regarding individual participants will identify participants by patient identifier only. The patient identifier will be an anonymized key to identify unique subjects and will be used for data pooling, transfer, storage, manipulation and analysis. Personal identifying data will not be available to the investigators. Study sites will store study records that bear participant names or other personal identifiers separately from records identified by patient identifier. Lists, logbooks, appointment books, and any other documents that link the patient identifier to personal identifying information at the facilities should be stored in a separate, locked location in an area with limited access. Electronic data will be stored on password secured servers, which will be backed-up on a regular basis. Data will be presented in aggregate form only.