Bringing women’s voices to PMTCT CARE: Adapting CARE’s Community Score Card© to engage women living with HIV to build quality health systems in Malawi

Background: Coverage of prevention of mother-to-child transmission of HIV (PMTCT) services has expanded rapidly but approaches to ensure service delivery is patient-centered have not always kept pace. To better understand how the inclusion of women living with HIV in a collective, quality improvement process could address persistent gaps, we adapted a social accountability approach, CARE’s Community Score Card To address these challenges, an inclusive quality improvement approach that empowers mothers living with HIV and their providers to systematically identify and overcome barriers to quality of care and identify joint solutions is needed. CARE’s Community Score Card©(CSC), a community-engagement approach that brings together service users and service providers at the local level to collectively share feedback and improve the quality of services, could potentially serve as one such approach. CARE’s CSC is grounded in the principles of social accountability, the belief that the mechanisms which allow citizens themselves to engage directly with duty bearers increase public officials’ accountability to their commitments and responsibilities, and in those of patient-centered care, a respect for a patient’s preferences, needs and values and a commitment to provide responsive, consultative care [25, 26]. The approach has been shown to improve service use and access, satisfaction with services, and accountability to patients’ needs and desires [27, 28]. CARE’s CSC approach is described more extensively elsewhere [27, 28]; however, briefly, the CSC consists of five-phases of implementation (see Fig. 1). Each of the five phases makes up a single cycle of the CSC process. Essential to the success of the CSC is the fact that these cycles are repeated on a regular basis, facilitating an ongoing quality improvement process and not a one-off event or activity. Progress on issues identified by those engaged in the process is assessed using score cards that track context-specific indicators, and action plans that document collective action to which participants in the process commit. CARE’s Community Score Card Process While CARE’s CSC has demonstrated positive impact on health services related to general maternal, neonatal and child health, and other health service domains [27, 28], at the time of the launch of this project it had not been adapted, implemented, or evaluated in a HIV health service delivery setting. As HIV remains a stigmatized health issue [29–31], people living with HIV too often are not given opportunities to voice their specific concerns and needs, and lack trust and confidence in the health system’s ability to provide confidential, tailored, and respectful care as a result [32, 33]. To address this gap, CARE partnered with the Malawi Ministry of Health through the support of the Elizabeth Glaser Pediatric AIDS Foundation (EGPAF) and the US Centers for Disease Control and Prevention to adapt and implement the CSC approach with PMTCT service providers and with mothers who use PMTCT services. Key considerations within this adaptation included operationalizing the approach through the clinic platform and 24-month PMTCT treatment cycle; alignment with the national HIV control strategy; and privacy of mothers living with HIV. Specific adaptations within each phase are described below. Typically, this phase includes the groundwork to identify scope, ensure buy-in from community and other stakeholders, and plan for implementation. Unlike previous iterations of CARE’s CSC that worked with communities directly to engage service users in the project, this adaptation relied on clinics to serve as the platform for both recruitment and implementation of the approach. Facilities with high maternal HIV-positive patient loads and poor PMTCT indicators were specifically targeted for implementation. Mothers initiating ART for PMTCT following ANC visits were invited to participate in the process. HIV-positive breastfeeding mothers were also identified and invited through outreach to existing support groups and via mother-infant-pair clinic days. Typically, the CSC trains community members to serve as co-facilitators in issue generation, scoring, and interface meetings at each site. For the PMTCT adaptation, expert clients1 and mentor mothers2 were prioritized as facilitators and trained to play this role, serving as a bridge between HIV-positive community members and the health system. In this stage, implementers bring together service users to identify priority issues, list and score indicators to measure progress, and generate suggestions for sustainable improvements. To promote a focus on PMTCT-specific issues, the adapted CSC restricted participation to PMTCT service users (HIV-positive pregnant and lactating mothers). As opposed to CARE’s previous CSC implementation experiences that were held publicly to promote accountability and broad-based collective action, the PMTCT CSC meetings were held at secluded off-site locations. The issue generation process also employed a PMTCT-specific discussion guide, created to surface issues specific to initiation and retention in treatment and uptake of early infant diagnosis services. The assumption was that by engaging only mothers who had used PMTCT services, and focusing discussions around barriers to these services specifically, the action plans and solutions identified through the CSC process would creatively address the barriers specific to PMTCT. Phase 3 is similar to phase 2 but focuses on service providers rather than service users. This phase allows service providers to share their own perspectives on the successes and challenges of delivering health services. Unlike in primary health centers where all clinic staff participated in issue generation and scoring exercises, in district hospitals participation was restricted to staff representatives from departments that specifically served HIV-positive pregnant and breastfeeding mothers: doctors, clinical officers, nurses and midwives from ANC and HIV/ ART clinical departments. To promote collective action and accountability, interface meetings convene not just service users and service providers but members of health advisory committees, district health management teams, and other relevant stakeholders to discuss the score cards and develop a joint action plan. Since interface meetings include these various groups, the adapted CSC for PMTCT included an option for HIV-positive participants to elect a representative to present their scores on their behalf, sometimes an expert client or mentor mother. To more rapidly address challenges of prevention of transmission to infants over a defined risk period, the implementation period for action plans under the adapted CSC was abbreviated from the typical 6 months to 3 months. The study was conducted at 11 health facilities across Dedza and Ntcheu districts, including nine health centers (primary care level) and two district hospitals (secondary level). Dedza and Ntcheu districts are in Malawi’s central region. The nine health centers were located in small towns or rural parts of the District whereas the two district hospitals were located in the more urban and peri-urban areas. Sites were selected purposively through review of routine PMTCT program monitoring data. Selected facilities met the following criteria: sufficient volume of newly identified HIV-positive pregnant mothers each year (minimum of 25); less than an 85% 6-month ART retention rate among mothers; and 6-week infant diagnosis performance that fell below the national average. Recruitment of PMTCT service users for participation in the CSC process occurred through two distinct avenues: support group-based recruitment and clinic-based recruitment. Clinic-based recruitment, conducted by clinic-staff, offered the most efficient way to recruit newly diagnosed HIV-positive mothers at the onset of their PMTCT journey. This recruitment was supplemented with a parallel recruitment exercise, conducted by project staff, among support group members as a way to include women who were already diagnosed and so would not be captured in the clinic-based recruitment process. Identifying, mapping and connecting with support groups for people living with HIV was part of the first phase of the adapted PMTCT CSC. Through these groups, pregnant and breastfeeding mothers living with HIV were invited to participate in the PMTCT CSC process. Newly diagnosed HIV-positive mothers and women who may not have yet joined a support group, were identified and recruited from the health facilities while accessing clinical services. Independent of the recruitment pathway, mothers who were interested in participating received a detailed written description of the process and consented to participate using a signature or thumbprint. A total of 822 mothers were recruited to participate. Service providers were recruited to participate in the CSC intervention in close collaboration with facility managers and the District Health Management Teams (DHMT). The project was introduced to the DHMT and then to in-charges and staff at each facility through in-person meetings and health workers were invited, at this time, to participate. Because a wide breadth of providers play a role in how women access and utilize PMTCT services, the program aimed to identify service providers from every level of service delivery in the primary health care facilities. In the two District Hospitals, representatives from ANC and ART clinics specifically participated due to operational limitations of including all clinic staff. Once health workers were invited and consented to participate, they were oriented on the importance of maintaining confidentiality during the CSC as part of the initial meetings and project start-up. A total of 64 health workers provided written consent. Select leaders and stakeholders from the broader community were engaged in the interface meetings and action planning phases of the CSC process. Engagement of these stakeholders increased accountability on issues identified through earlier phases of PMTCT CSC implementation. These leaders were identified through introductory meetings prior to the start of the project, and throughout its execution as different challenges and solutions emerged. These included religious leaders, DHMT members, leaders of governance structures (i.e., health advisory committees, village development committees, and village health committees), village chiefs, traditional authorities, and politicians (members of Parliament and councilors). PMTCT service users and providers were identified through the recruitment strategies described above and invited to attend each stage of the PMTCT CSC process. This included an initial issue generation meeting during the first cycle and then subsequent scoring, and interface meetings in all three cycles. To ensure that both issues and solutions addressed challenges unique to each facility, all meetings were facility specific, meaning they involved only service users and service providers associated with that particular facility. Each cycle took about 1 month to complete across all 11 health facilities, followed by a three-month action plan implementation and monitoring period. Together, all three cycles were conducted over a period of 12 months, from September 2017 to August 2018. Score card indicators were developed in a consultative, participatory process based on the issues identified during the issue generation meetings. Once issues were identified, an indicator development meeting was held. During these meetings, facilitators listed, reviewed, and discussed the priority recommendations that came out of the issue generation process. Major themes were classified into distinct domains and a perception-based indicator was created. For example, issues such as reluctance of male partners to get tested and low participation of male partners in ANC visits and decisions around infant care and testing; were classified into an indicator of “Level of male involvement on PMTCT Issues”. Once indicators were created, service users and service providers met separately across each of the 11 facilities to conduct scoring meetings. During scoring meetings, participants discussed each indicator and agreed on a perception-based score using a scale of 0 to 100. This process generated two separate score cards per health facility – one from the service users’ perspective and a second from the service providers’ perspective. These two score cards were presented and discussed during the Interface Meetings and informed the development of subsequent action plans. The same indicators were used across all 11 intervention sites (see Table 1). PMTCT CSC Indicators by locus of control and component of high-quality health system a PMTCT: prevention of mother-to-child transmission; HIV: human immunodeficiency virus; ART: antiretroviral therapy; HTC: HIV testing and counseling; EID: early infant diagnosis b Locus of control was assigned as either individual / patient, provider, health facility, community or health system based on where / who had the highest capacity and authority to effect change in the indicator c Component was assigned based on Kruk et al.’s high quality health system framework components During these meetings, data in the form of scores for each indicator were recorded using a paper template, posted on large poster paper so that all participants could view and confirm scores were recorded accurately. Upon completion of each round, scores for all 11 facilities were entered by CARE project staff into an Excel database. Quality assurance was conducted by the Technical Advisor supporting this work who examined the database after each month, identified any missing values or values outside of the plausible range, and through an audit of the paper records, made any necessary corrections. For our analysis we categorized the PMTCT CSC Indicators based on locus of control and by Kruk et al.’s domains and components of a high-quality health systems [16]. Locus of control was assigned across one of six categories as either individual or patient, provider, health facility, community, or health system based on where or who had the highest capacity and authority to effect change in the indicator. For example, availability of trained health workers was categorized as within the health system locus of control because, in this context, human resource allocations and training decisions are primarily made at the district level in accordance with national-level guidance and resourcing. Each indicator was also categorized based on one of the ten components of high-quality health systems defined by Kruk et al. [16]. We examined changes in scores aggregated across service provider and service user populations from first and last cycle. While scores were collected at individual sites, for this analysis we averaged the scores for the first and last cycle across the 11 sites to arrive at three distinct scores per time period, per indicator: a service user score and a service provider score, and a combined score that averaged across both service users’ and service providers’ perspectives. The analysis below presents the percent change from first to last cycle of the combined scores aggregated across all sites and across both scoring populations (service users and providers), rounded to the nearest whole number. We also compare the absolute percentage point difference between service user and service provider scores, aggregated across all sites, at first and last cycle. Differences were assessed using a Z test and p-values ≤ .05 were considered statistically significant. Analysis was conducted using Microsoft Excel [34].