Balancing adaptive delivery and implementation fidelity over time: a longitudinal mixed-methods study of a community health worker–led adolescent nutrition intervention in rural Uganda

Abstract

Abstract Background Adolescents and young adults (AYAs) aged 10–24 years in sub-Saharan Africa experience low nutrition literacy and poor diet quality. Community health worker (CHW)-led interventions offer a delivery platform, yet little is known about how implementation proceeds under routine conditions or how fidelity is maintained alongside adaptive delivery adjustments. We assessed implementation fidelity over time and examined how contextual influences and supervision-guided adaptive delivery adjustments shaped delivery processes and supported preservation of core intervention functions during the nine-month implementation of a CHW-led intervention in Uganda. Methods This longitudinal mixed-methods implementation study was nested within a nine-month randomized controlled trial in rural Eastern Uganda. Fidelity was assessed across early, mid, and late implementation using Carroll’s framework. Qualitative data were collected through repeated focus group discussions with CHWs, AYAs, and parents. Data were analysed using an inductive–deductive approach: adaptive delivery processes were identified inductively and organized and interpreted using the Consolidated Framework for Implementation Research. Quantitative and qualitative findings were integrated using joint displays to compare fidelity findings with qualitative contextual influences, supervision, and adaptive delivery adjustments. Results Fidelity scores were high across implementation periods, with modest variation across domains. On a 1–5 scale, overall comparable fidelity scores were 4.18 (SD 0.12), 4.21 (SD 0.07), and 4.16 (SD 0.08) during early, mid, and late implementation, respectively. Adherence remained stable while quality of delivery was slightly higher during mid-implementation. Five adaptive delivery processes were identified: flexible scheduling and revisits; contextualized reinforcement of complex content; resource-sensitive counselling adjustment; workload-responsive micro-planning; and network-supported participant follow-up. Qualitative findings indicated that participant availability, comprehension challenges, household resource constraints, CHW workload, and participant mobility shaped delivery. Supervisory reflection and debrief provided a forum for reviewing challenges, sharing practical solutions, harmonizing delivery practices, and agreeing on context-responsive adjustments. These adjustments modified delivery forms, including visit timing, communication approaches, locally feasible examples, visit sequencing, and follow-up arrangements, while preserving core intervention functions. Conclusion Adaptive delivery and implementation fidelity can be managed as interdependent implementation processes rather than competing priorities. Fidelity monitoring, supportive supervision, and guided problem-solving supported context-responsive delivery adjustments while supporting maintenance of core intervention functions. Trial registrations The parent randomized controlled trial was prospectively registered with the Pan African Clinical Trials Registry (PACTR202406699274475) on 7 June 2024. https://pactr.samrc.ac.za/TrialDisplay.aspx?TrialID=30558 .