An Original Contribution to Epidemiological Methodology: Adapting Causal Inference Methods for Community Health Research in Resource-Limited Nigerian Settings

Causal inference in epidemiology requires methodological approaches that account for confounding, selection bias, and measurement error in ways that standard regression approaches inadequately address, yet the application of modern causal inference methods (directed acyclic graphs, propensity score methods, instrumental variables) to community health research in resource-limited Nigerian settings faces distinctive data quality, study design, and contextual validity challenges requiring original methodological adaptation. This study developed an original methodological contribution adapting causal inference methods for community health research in resource-limited Nigerian settings. A methodological research design was employed across three phases. Phase one conducted a systematic methodological review of causal inference approaches applied in Nigerian and comparable sub-Saharan African community health research from 2015 to 2024 (44 publications), identifying methodological gaps, frequent bias sources, and adaptation challenges. Phase two applied and stress-tested six causal inference methods (DAG-based confounding assessment, propensity score matching, inverse probability weighting, instrumental variable analysis, difference-in-differences, and regression discontinuity design) to datasets from three Nigerian community health studies with known data quality challenges, evaluating method performance under realistic data limitations. Phase three developed the Nigerian Community Health Causal Inference Guide (NCHCIG) synthesising recommendations from phases one and two. The NCHCIG provides decision rules for method selection based on study design, data quality indicators, and policy inference requirements; adaptations for DHIS2 routine data analytical bias sources; approaches to handling ethnic cluster confounding in multi-community studies; and supplementary validation methods for instrument credibility in IV analyses.