Informing scale and sustainment of a systems-based approach to optimize PMTCT in central Mozambique: a multi-methods evaluation of SAIA-SCALE implementation

Abstract

Prevention of mother-to-child transmission of HIV (PMTCT) programs are essential to improve the health of and prevent vertical HIV transmission from pregnant and breastfeeding mothers to their children. In Mozambique, although 90% of pregnant women living with HIV initiate antiretroviral therapy (ART) in antenatal care, less than 40% are retained in care by six months postpartum, and consequently, vertical HIV transmission is estimated at 10%. Improved patient follow-up, increased early infant PCR testing, and reduced incident HIV among children are possible by optimizing and sustaining service efficiency across the PMTCT care cascade using systems-engineering strategies like the Systems Analysis and Improvement Approach (SAIA). SAIA facilitates cycles of cascade analysis, process mapping, and continuous quality improvement (CQI) among healthcare teams to make improvements along the PMTCT care cascade. In line with global implementation science priorities, research on SAIA has evolved to test methods for scale and sustainment, with the goal of long-term uptake by national entities. From 2018-2021, the SAIA-SCALE trial evaluated a more pragmatic delivery model in Manica Province, Mozambique, using district management teams rather than research staff. This three-wave stepped-wedge cluster randomized trial included two phases of implementation, an intensive phase characterized by research staff mentorship, and a maintenance phase with limited external support. While the SAIA-SCALE model serves as a foundation for national scale-up for wider improvements to PMTCT service efficiency and coverage, it is susceptible to implementation decay in the absence of adequate strategies and planning for sustainment. The objective of this dissertation was to understand implementation experiences at scale and identify potential needs and threats to sustainment in order to inform adaptations for testing with future iterations of SAIA for sustained implementation and improved performance at scale. The three studies conducted were embedded within the SAIA-SCALE trial. Our first study (Chapter 2) measured and assessed implementation fidelity (i.e., protocol adherence) of SAIA at scale to understand changes in fidelity over time and contextualize trial findings. We operationalized Carroll and colleagues’ conceptual framework for implementation fidelity to assess adherence to content, coverage, and frequency of an ideal SAIA cycle, then determined composite high vs. low fidelity for 828 facility-month of SAIA-SCALE implementation. High fidelity was observed in 62% facility-months. Fidelity elements revealed strong adherence to content and frequency; coverage was the weakest element (82%). Fidelity improved with successive randomization waves, potentially reflecting gained experience and skills in practice facilitation as the program matured. Monthly trends in fidelity revealed that high fidelity improved during the intensive phase of SAIA-SCALE implementation when research staff provided mentorship, but decreased by 92% (OR: 0.08; 95% CI: 0.04-0.17; p<0.001) upon withdrawal of mentorship, with subsequent gradual recovery during maintenance phase, in which district managers facilitated SAIA cycles independently. These patterns reflect the program drift that frequently occurs during scale-up and underscores the need for proactive planning to maintain fidelity during transitions to scale and longer-term sustainment. Fidelity measurements from our first study enabled subsequent analyses in our second study (Chapter 3) to examine whether and to what degree fidelity drove program outcomes. We evaluated the association between SAIA fidelity and service- and patient-level outcomes, and whether SAIA-SCALE effectiveness was moderated by cumulative fidelity performance. High fidelity was associated with greater improvements within each implementation phase; each additional month of high-fidelity per facility-quarter was associated with significantly lower HIV transmission in intensive phase and significantly higher proportions of infants tested for HIV in maintenance phase. In addition, significant improvements from pre-implementation baseline were observed in early infant HIV testing among low cumulative fidelity facilities (<66% high-fidelity months) in intensive phase and overall HIV transmission among high cumulative fidelity facilities (≥66% high-fidelity months) in intensive phase and among low cumulative fidelity facilities in maintenance phase. However, cumulative fidelity did not consistently moderate the effect of SAIA-SCALE overall. Together, these results indicate that once implementation has begun, additional months of high fidelity can be associated with improved service- and patient-level outcomes, but that the magnitude of improvement from pre-implementation baseline may not depend on fidelity alone. For our third study (Chapter 4), we qualitatively analyzed 30 interviews with SAIA facilitators and attendees to understand readiness, determinants of readiness, and efforts for sustainment one month following the end of the SAIA-SCALE trial. Methods were structured using the Consolidated Framework for Implementation Research (CFIR 1.0) and a sustainment-focused application of the Theory of Organizational Readiness for Change embedded with CFIR constructs. Facilitators of organizational readiness for sustainment included perceived relative advantage of the SAIA strategy and continued access to the mobile PMTCT cascade analysis tool (PCAT) application. Barriers cited included withdrawal of mentorship by researchers, perceived lack of access to the mobile PCAT application, and a lack of funds for transportation, meeting materials (markers and paper), and snack incentives to encourage attendance. Commitment to sustain SAIA-SCALE was consistent among participants; however, collective efficacy to sustain varied by participant role, geographic location, and perceived availability of the PCAT application. Overall, district manager participants expressed readiness for SAIA-SCALE sustainment in health facilities in district capitals and all participants expressed a lack of sustainment readiness in peripheral facilities outside the district capitals. Participants proposed approaches to overcome perceived barriers to sustainment in peripheral health facilities and sustained implementation for long-term benefits to PMTCT services. This research addresses important questions to ensure implementation strategies such as SAIA maintain their health benefits as implementation progresses from scale to sustainment and on to real-world delivery. This dissertation contributes to the literature and offers implementers and policymakers an understanding of needs, informing efforts to sustain the SAIA-SCALE program including potential sustainment strategies and adaptations for improved fit to context. Furthermore, these findings will support future uptake, program planning, and dissemination of the SAIA strategy by the Ministry of Health in Mozambique, with implications for national programs in similar contexts.