Missing data practices in DHS-based child health research in sub-Saharan Africa: a methodological review of diarrhea and acute respiratory infection studies

Demographic and Health Survey (DHS) data are important for determining risk factors associated with childhood diarrhea and acute respiratory infections (ARI) in sub-Saharan Africa (SSA); however, missing data from DHS can bias analyses, and no systematic review has assessed this practice. To address this, peer-reviewed DHS/MICS studies from SSA with childhood diarrhea or ARI as health outcomes were systematically retrieved from OpenAlex, PubMed, AJOL, and Dimensions. Screening and data extraction were conducted independently by two reviewers using a checklist developed from the Joanna Briggs Institute approach for assessing risk of bias. Empirical missingness was analyzed using data from the Nigeria DHS 2018 (N = 33,924 under-five children). From a total of 2,148 articles, 103 fulfilled our inclusion criteria (41 on diarrhea, 62 on ARI). Only three diarrhea articles (7.3%) and no ARI articles (0.0%) reported missing data. Complete-case analysis (CCA) was used in 80.5% of diarrhea articles and 100% of ARI articles. Multiple imputation (MI) was used in two diarrhea studies (4.9%) and no ARI articles. The empirical missingness rate for outcomes in both diseases was 9.8% across the full analytical sample. Clean cooking fuel was associated with reduced risk of ARI (CCA OR = 0.82, MI OR = 0.81, bias = -0.9%) and diarrhea (CCA OR = 0.66, MI OR = 0.64, bias = -3.5%). Inadequate missing data handling exists in both fields, but the challenge is more pronounced for ARI, where no effective alternative has been adopted to address the issue. Therefore, journals must enforce missing data reporting, and researchers should prioritize MI.