Balancing adaptations and fidelity of implementation strategies to optimize primary healthcare in Ethiopia: Lessons from embedded implementation research

Abstract

BACKGROUND: Optimizing Ethiopia's Health Extension Program (HEP) has been critical to enhancing primary healthcare (PHC) service delivery, particularly in remote and underserved settings. Embedded implementation research (EIR) strengthens PHC systems by positioning implementers and local managers at the center of addressing operational bottlenecks, enabling adaptive, context-specific refinement of interventions and strategies while maintaining high fidelity to their core components. This study evaluates the implementation fidelity, adaptation of strategies, and lessons to address challenges in equitable access, quality, and accountability through the "Improve Primary Health Care Service Delivery" project. METHOD: A participatory pragmatic EIR approach was applied to co-design and pressure test the implementation strategies to optimize the HEP roadmap (2020-2035). Fourteen woredas from agrarian and pastoral contexts were selected to pressure test HEP optimization strategies. Continuous adaptation tracking and fidelity monitoring were conducted throughout the project implementation period from April 2022 to September 2024. The HEP optimization implementation strategies, adaptations, and modifications were guided by frameworks such as Framework for Reporting Adaptations and Modifications to Evidence-based Implementation Strategies to ensure interventions fit local contexts. RESULTS: The adaptations of implementation strategies resulted in implementation fidelity, which improved maternal and child health service coverage, strengthened community health program units, enhanced referral systems through Networks of Care (NoCs), and contextualization of service delivery for pastoralist communities. While most strategies were retained and repackaged for scale, others-including performance-based incentives and governance restructuring-were deprioritized because of feasibility, policy, and sustainability considerations. Adaptations were guided through iterative learning and stakeholder engagement and clustered into four broad areas: contextualization of service delivery models, strengthening community engagement, streamlining quality improvement and NoCs approaches, and refining accountability mechanisms. Stakeholder engagement was pivotal in balancing fidelity with contextual adaptations, fostering trust and sustainability. Despite challenges such as resource constraints, sociocultural barriers, and infrastructure limitations, systematic monitoring and iterative learning processes facilitated the refinement of scale-up strategies. CONCLUSIONS: Adaptive implementation science effectively optimizes health programs in complex and dynamic contexts. Participatory co-design, stakeholder engagement, and systematic frameworks facilitated the development and refinement of strategies that improved service quality, access, and accountability in diverse PHC settings.