Early ART initiation among HIV-positive pregnant women in central Mozambique: A stepped wedge randomized controlled trial of an optimized Option B+ approach

Background: Despite effective prevention strategies and increasing investments in global health, maternal to child transmission (MTCT) of HIV remains a significant problem globally, especially in sub-Saharan Africa. In 2012, there were 94,000 HIV-positive pregnant women in Mozambique. Approximately 15% of these women transmitted HIV to their newborn infants, resulting in nearly 14,000 new pediatric HIV infections that year. To address this issue, in 2013, the Mozambican Ministry of Health implemented the World Health Organization-recommended “Option B+” strategy in which all newly diagnosed HIV-positive pregnant women are counseled to initiate combination anti-retroviral therapy (ART) immediately upon diagnosis regardless of CD4 count and to continue treatment for life. Given the limited experience with Option B+ in sub-Saharan Africa, few rigorous pragmatic trials have studied this new treatment strategy. Methods: This study utilizes an initial formative research process involving patient and health care provider interviews and focus groups, workforce assessments, value stream mapping, and commodity utilization assessments to understand the strengths and weaknesses in the current Option B+ care cascade. The formative research is intended to guide identification and prioritization of key workflow modifications and the development of an enhanced adherence and retention package. These two components are bundled into a defined intervention implemented and evaluated across six health facilities utilizing a stepped wedge randomized controlled trial study design. The overall objective of this trial is to develop and test a pilot intervention in central Mozambique to implement the new Option B+ guidelines with high fidelity and increase the proportion of HIV-positive pregnant women in target antenatal clinics (ANC) who start ART prior to delivery and are retained in care. Discussion: This pragmatic study utilizes research strategies that have the potential to meaningfully improve the Option B+ care cascade in central Mozambique and to decrease the MTCT of HIV. This trial is designed to identify critical low-cost improvement strategies that can be bundled into a defined intervention. If this intervention has a measurable impact, it can be rapidly scaled up to other ANC in Mozambique and sub-Saharan Africa.

The project utilizes an initial formative research process to understand inefficiencies in the current Option B+ care cascade, to guide identification and prioritization of key workflow modifications and the development of an enhanced adherence and retention package. These two components, described in detail below, are bundled into a defined intervention implemented and evaluated across six health facilities utilizing a stepped wedge randomized controlled trial (with the health facility as the unit of randomization). As a facility-level intervention, the proposed “test-and-treat” Option B+ intervention is implemented through a stepped wedge design (described below) in six high-volume health centers providing PMTCT and ART services in the Mozambican national health system. These health centers serve communities along the highly populated Beira highway and railway transport corridor that passes through Sofala and Manica provinces (Figure 1), from the port city of Beira on the Indian Ocean to the Zimbabwe border [21]. Map of Mozambique—provinces and provincial capitals [12]. Intervention sites include three health facilities in Sofala Province (Macarungo, Munhava, and Dondo) and three in Manica Province (Nhamaonha, 1° de Maio, and Gondola). All of these public health facilities provide the full range of PMTCT services, including HIV testing, access to CD4 testing (four out of six through transport to a central lab), and ART. ART has already been decentralized to ANC through MoH trainings. The intervention will be designed to enhance the new national MoH Option B+ ART policy in which all pregnant women attending their first ANC visit will be tested for HIV and, if positive, will be provided with first-line triple ART (TDF + 3TC + EFV, single daily fixed-dose combination) during that same visit, or within 14 days of the first visit and HIV test. In the central provinces of Sofala and Manica where the research is conducted, the provincial HIV prevalence rates of adults aged 15–49 years are among the highest in the country for women at 15.6% and 17.8%, respectively, and 14.8% and 12.6% for men [22]. The prevalence among pregnant women in 2009 was estimated at 18% [23,24]. HIV testing and treatment services in Manica and Sofala provinces have been scaled up with technical and financial support from Health Alliance International (HAI), a US non-profit affiliated with the University of Washington, School of Public Health. By early 2010, ART had been made available in 39 public sector sites [24]. In Manica, over 15,000 and, in Sofala, nearly 22,000 Mozambicans have started ART. In 2010, PMTCT services were offered in routine antenatal care in 93 health facilities in Manica and 110 facilities in Sofala. First ANC visit coverage is consistently high at over 90% for Sofala and Manica provinces [24,25]. HIV testing at the first ANC visit has also attained consistently high rates of 90% in both provinces through the current opt-out testing program [25]. However, on average, women arrive late, between 19 and 25 weeks mean gestational age (unpublished data). Prior to the introduction of Option B+, there were a number of documented challenges in the PMTCT care cascade: eligible HIV-positive pregnant women were referred to adult ART services but only about 20% started ART, only 29% of HIV-positive mothers received nevirapine during labor in central Mozambique, and only 60% of births occurred at a health care facility [22-24]. Recent preliminary studies (2009–2010) conducted in the proposed research sites by UW and HAI researchers confirmed these previously reported barriers to care and sources of LTFU. A recent study of ART initiation and the PMTCT treatment cascade in seven large health centers in Manica and Sofala, which include two of the target sites included in this proposal (Munhava highlighted in Figure 2 and Nhamaonha), showed major LTFU and very small proportions of eligible positive mothers initiating ART [20]. Munhava health center PMTCT patient flow. This observational study, led by Chris Dodd and Jennifer Einberg, examined PMTCT LTFU under the previous protocol. The proportion that started ART varied significantly across the sites, but was low for each. Only about 40% of HIV-positive women actually received their CD4 results within 30 days of the first ANC visit, and only 40% of those eligible managed to start ART. Overall, only 19.3% of eligible women started ART across all the sites (see Figure 3). Province-wide program data from 2010 suggest similar patterns; only about 10% (8% in Manica and 12% in Sofala) of HIV-positive pregnant women started ART, while data suggests that an estimated 40% were actually eligible based on the protocol for ART initiation at that time [25]. Aggregate flow cascade through seven health centers in Manica and Sofala provinces (2009). Three recent qualitative studies by HAI researchers that focus on patient experience further corroborate sources of LTFU from the perspective of patients. Patients reported that requirement for multiple visits prior to ART initiation created major barriers for women because of transport costs and fears of unintentional HIV status disclosure causing stigmatization [26,27]. The last study revealed major gaps in understanding among HIV-positive women about steps to follow for care, benefits of care, and lack of system tools to support patient follow-up to ART and through postpartum visits. Women reported confusion about integrated ANC services, the purpose of the various tests and treatments provided (i.e., syphilis testing and treatment, IPT for malaria, and HIV and CD4 testing), and the need for follow-up visits [28]. Together, these studies indicate that PMTCT services and ART referrals require excessive repeat visits, provide insufficient information to patients, and lack patient tracking tools to ensure follow-up. The CD4 testing process and subsequent preparatory visits were associated with major delays and drop-offs in initiating ART and LTFU. Based on this evidence, interventions to increase earlier ART initiation and subsequent treatment adherence will require major streamlining of the process coupled with improved adherence counseling [29]. The Option B+ approach provides a model to achieve this streamlining if appropriately designed to capitalize on the strengths and recognize the constraints in the Mozambique heath system. The introduction of the intervention in the study sites will occur in three steps following a stepped wedge design [30]. Prior to randomization, the six sites were stratified by province, and one site from each province (two total) was randomly selected to initiate the intervention at each of three stepped time points (months 5, 8, and 11; see Figure 4). Each step is separated by 3 months to allow an adequate number of people to be tested and initiate ART in each site (1.5 months, i.e., 1 month testing plus 14 days) and outcomes to be measured prior to the subsequent step (1.5 months, i.e., 45 days post-ART initiation). The implementation process includes staff training, integration of new adherence strategies, and intensive supervision and mentorship by the study staff and health facility managers to troubleshoot problems that arise. Stepped wedge design of the implementation of test-and-treat intervention in PMTCT programs in central Mozambique. During the first year of the project, researchers conducted formative research at the six selected sites to determine how to best adapt the Option B+ model to the six facilities. The research consists of these major activities: At the end of the formative research process in year 1, researchers developed key intervention materials together with the district and provincial directorates and the Beira Operations Research Center (CIOB), part of the Mozambican MoH, and review them for approval with health personnel at each site. Based on formative research findings produced in achieving the original specific aims 1 and 2, the following core components of the B+ study intervention have been designed and will be stepped in at the study sites to improve early retention in care and adherence: Option B+ study core components Facility clinical director together with maternal child health (MCH) nurses conducts meetings. Following the development of intervention components and before intervention implementation, in each facility as they are stepped in (designated as the “S” in Figure 4), a total of 25 staff (ANC nurses, receptionists, physicians, and physicians assistants or técnicos) participate in a 1-week training. The intervention bundle will be rolled out across the six intervention sites in three steps as described in Figure 4. The trainings focus on the MoH Option B+ protocol and the intervention core components as described above. Implementation trainings will include: a) refresher trainings on ART provision for MCH (SMI) nurses; b) new Option B+ policies, procedures, and drug regimes; c) laboratory testing protocols including CD4 count, hemograms, liver function testing, and biochemical panels as outlined by MOH PMTCT guidelines; d) new flow charts, job aids, treatment checklists, and data collection procedures; e) patient visit schedule for ANC and ART follow-up consultations through final postpartum visit; f) postpartum visit procedures for transfer and referrals of ART patients to outpatient services; and g) supervision procedures by physicians or physician assistants (técnicos) from outpatient services. Data will be collected as described below and will be evaluated in near real time to allow for concurrent impact evaluation. Throughout the intervention period, research teams will engage in process evaluation procedures to monitor the intervention and outcomes. These will include indicators for HIV testing, ART initiation, pharmacy refill rates, patient follow-up through visits and texts, adherence committee meetings, counseling sessions, and other key intervention activities. Researchers will conduct interviews with health workers to troubleshoot workflow challenges and bottlenecks and evaluate supervision practices. The process evaluation includes the following components: Year 1 focused primarily on formative research. Year 2 included designing the intervention and seeking IRB and MoH approvals, and in year 3, the intervention will be rolled out following the stepped wedge design with concurrent evaluation of impact as outlined in Figure 4. We will measure early retention by calculating scheduled 30-day visit rates for pharmacy refill (up to 45 days from ART initiation) among women who initiate ART under the study intervention model and compare these rates to women starting ART before the intervention, while controlling for trends over time. We will also calculate 90-day adherence rates among women who initiate ART under the study intervention model and compare these rates to women starting ART before the intervention, while controlling for trends over time. We hypothesize that the proportion of women initiating ART during the intervention phase who return for their 30-day refill visit within 45 days will increase from 50% to at least 70%. We also hypothesize that the proportion of women initiating ART during the intervention phase who have 90-day ART adherence rates ≥90% based on pharmacy refill data will increase from 40% to a least 60%. As a secondary, but critical, outcome, we will measure, monitor, and compare the proportion of HIV-positive pregnant women starting appropriate ART within 14 days of HIV testing before and after the intervention. Option B+ rollout data for ART initiation collected during the formative research period have varied widely among the six sites; however, MoH officials are confident that ART initiation rates will improve substantially and stabilize at a higher level using current protocols before the intervention begins, so follow-up remains the greater challenge. The intervention will seek to ensure that each site achieves at least 75% ART initiation (of mothers testing HIV-positive) within 14 days of HIV testing, and researchers will monitor and troubleshoot site performance throughout the research period. Study outcomes are summarized in Table 1. Option B+ stepped wedge study outcomes measures ART anti-retroviral therapy, HIV human immunodeficiency virus, PMTCT prevention of maternal to child transmission of HIV. Research teams will measure patient ART adherence by extracting data from routine health records. There are three sets of patient health records system that researchers will have access to for each subject: 1) the ANC registry that is completed for every woman in her initial ANC consultation, 2) the HIV treatment registry and chart that is used for each patient who initiates ART, and 3) the refill records of the pharmacy. The data extracted from each of these sources will be consolidated into a single Microsoft Access database. Once consolidated in the new data set, the patients will be identified using a new study-specific identification number in the data set for analysis. Data extraction from registries is performed by trained study team members in each facility and compared for consistency. In cases of inconsistency, data collection will be repeated. All data will be collected in Mozambique and analyzed using Stata SE (StataCorp, College Station, TX). We will compare outcomes in the pre- and post-intervention phases and will use an intention-to-treat analysis that adjusts for clustering by health facility and changes over time. Two analysis approaches will be considered. First, if each measurement period has a similar number of observations per period per site, each site will contribute four outcome measurements for 30-day ART pharmacy refill rates over time (one for each step period described in Figure 4), and the outcome measures (proportions) will be collapsed into continuous numeric variables. We will then perform a linear regression analysis for repeated measure panel data, specifying our outcome as the dependent (continuous) variable and intervention status (yes/no) as the main independent variable, while adjusting for time and first-level autocorrelation between successive measures. Health facility will be specified as the panel to control for clustering by clinic. If the number of observations does vary over time and between sites, we will do an alternative individual-level analysis that uses individual-level data with the outcome coded in a bivariate format (yes/no). We will then use logistic regression to determine the relationship between the outcome and the intervention period (pre vs. post), while controlling for time period and clustering by site. The analysis will also focus on the calculation of 90-day ART adherence dichotomized into good/poor as described above, among all patients starting ART in the post-intervention phase. To determine if over 60% of those initiating ART have good adherence, we will calculate 95% confidence intervals using the binomial method and determine whether this interval includes (or is above) the goal of 60%. We will compare the proportion of patients with good 90-day adherence in the pre- vs. post-intervention phase, using the same stepped wedge methodology described above. The power calculation for this study is based on the primary outcome (proportion of HIV-positive pregnant women starting ART who return within 45 days for the first 30-day refill) and based on published calculations for stepped wedge designs [30]. For this calculation, we estimate that at minimum 30 new HIV-positive women will initiate ART in each of the ANC centers for each 1-month measurement period and that the baseline proportion of women returning within 45 days is 50%, based on data from formative research. For a two-sided α of 0.05 and an estimated coefficient of variation (k) between facilities of 0.2, we would have 84.4% power to detect an expected increase in our post-intervention outcome of 20% (i.e., from 50% to 70%). The power changes minimally with varying coefficients of variation (i.e., 82.0% power with k = 0.4). To maintain at least 80% power, we would be still able to detect a difference of 19% from the baseline 50% (i.e., to 69%). To determine the precision of our estimate for 90-day adherence post-intervention, we estimate that an average of 30 new HIV-positive women who initiate ART will be identified per ANC site for each 1-month measurement period. As there are 12 post-intervention periods under consideration, we estimate that 720 (30 × 12) will initiate ART. This denominator would allow a calculation of adherence that is within ±4% for adherence rates >50%. The institutional review board of the Ministry of Health of Mozambique and the University of Washington approved this study. The trial is currently registered with ClinicalTrials.gov #{“type”:”clinical-trial”,”attrs”:{“text”:”NCT02371265″,”term_id”:”NCT02371265″}}NCT02371265. Formative research at all six sites in central Mozambique began in March 2013. The intervention bundle was finalized in 2014. Implementation of the stepped wedge randomized controlled trial is in process.

N/A

• Dima Health