Applications and Implementation Considerations for Stepped-Wedge Designs in Sub-Saharan Africa: A Systematic Review

Adesola Z. Musa^1,2Folahanmi T. Akinsolu^1,2,3*Abideen O. Salako^1,2,3Olunike R. Abodunrin^2,4Oluwabukola M. Ola^2,3Oliver C. Ezechi^1,2,3

• ¹Department of Clinical Sciences, Nigerian Institute of Medical Research (NIMR), Lagos, Lagos, Nigeria
• ²Lead City University, Ibadan, Nigeria
• ³Center for Reproduction and Population Health Studies, Nigerian Institute of Medical Research (NIMR), Lagos, Nigeria
• ⁴Department of Epidemiology, School of Public Health, Nanjing Medical University, Nanjing, Jiangsu Province, China

Stepped-wedge designs (SWD) have gained prominence as a flexible study design in public health and implementation science because they allow phased rollout of interventions while addressing ethical concerns about withholding potentially beneficial care. This systematic review evaluates how SWD has been implemented and used to generate evidence in sub-Saharan Africa (SSA), focusing on intervention types, outcomes assessed, and contextual factors that shape design feasibility and performance. A systematic review protocol was developed in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidance. We searched studies conducted in SSA that used an SWD from January 2000 to March 2024 using five electronic databases (PubMed, Web of Science, CINAHL, PsycINFO, and Cochrane Library). Eligible studies were those conducted in SSA settings and evaluating public health, clinical, or social interventions using an SWD approach. Data were extracted on study characteristics, SWD operationalisation (clusters, steps, timing), statistical methods, and sample size considerations. The 85 included studies covered diverse health domains, most commonly HIV/AIDS, maternal and child health, tuberculosis, and malaria, and were implemented across hospitals, clinics, communities, and schools. Most evaluations were facility-based (84.7%), with a smaller proportion implemented in community settings (15.3%). SWD structures varied widely: the number of clusters ranged from as few as four to as many as 54, and the number of steps ranged from two to twelve. Sample size calculations typically reflected anticipated changes in primary outcomes and frequently incorporated intra-cluster correlation assumptions to account for clustering. SWD was commonly selected due to ethical considerations, operational constraints, and resource limitations, but studies showed substantial heterogeneity in design choices, intervention complexity, outcome timing, and analytical approaches. Overall, SWD has been used effectively in SSA to evaluate a broad range of interventions in real-world settings, demonstrating strong adaptability to resource-limited contexts. However, challenges related to trial duration, implementation logistics, randomisation procedures, and statistical complexity underscore the need for careful planning, transparent reporting, and appropriate analytic methods. These findings provide practical insights for researchers and decision-makers seeking to optimise SWD use for ethical and robust evaluation of interventions in SSA.