Peer-reviewed article
Co-occurrence of anaemia and malaria among children under five in Uganda: A joint modeling analysis
Grace Kakaire, Robert Too, Gregory Kibet Kerich, Mathew Kosgei ·
Summary
Written by AI, not yet AICS's view · claude-sonnet-5-5This study used data from a Ugandan Malaria Indicator Survey covering 5,755 children under five to look at how anaemia and malaria occur together. It used a joint statistical model instead of treating the two illnesses as unrelated. About 13.8% of children had both illnesses at the same time. The two illnesses were positively correlated (ρ ≈ 0.19). Some risk factors changed in effect once this link was accounted for, so the authors suggest integrated health interventions.
Key points
- The sample was 5,755 children under five from the most recent Malaria Indicator Survey.
- 13.8% of children had both anaemia and malaria at the same time.
- The copula correlation between anaemia and malaria was significantly positive (ρ ≈ 0.19).
- The association between maternal anaemia and child malaria risk reversed direction once interdependence was accounted for.
- Joint modeling gave more stable and plausible estimates than separate models.
Limitations
This is a cross-sectional survey analysis, so it cannot show that one illness causes the other or that any intervention works. The abstract does not report the survey year, the full list of risk factors or effect sizes for them.
Abstract
Childhood illnesses such as anaemia and malaria continue to pose significant health risks for children under five in sub-Saharan Africa. These conditions often co-occur, reflecting complex interdependencies driven by shared biological and environmental factors. Traditional analytical approaches typically assume independence between health outcomes, potentially overlooking meaningful associations. This study aims to identify the dependence structure among anaemia and malaria in children under five using a joint modeling approach. Data were drawn from the most recent Malaria Indicator Survey (MIS), comprising 5,755 children. A bivariate probit model was estimated using a Gaussian copula framework with probit margins for each outcome. This approach allowed for the simultaneous modeling of the marginal probability of each illness and the latent correlations among them. The model incorporated a range of child-, household-, and maternal-level covariates. Average Marginal Effects (AMEs) were calculated to facilitate interpretation. The copula correlation (ρ) between anaemia and malaria from the copula structure was estimated to quantify the dependence between the two illnesses. Descriptive analyses revealed substantial co-occurrence: 13.8% of children experienced both illnesses simultaneously. The joint copula model detected significant positive copula correlations (ρ ≈ 0.19) between anaemia and malaria. Several covariates displayed differential effects in the joint versus separate models. For instance, the association between maternal anaemia and child malaria risk reversed direction once interdependence was accounted for. Joint modeling yielded more stable and plausible estimates across outcomes. This study highlights the critical importance of accounting for outcome dependence in the analysis of child morbidity. The significant correlations among malaria and anaemia suggest that these illnesses are influenced by common or interacting risk factors. Joint modeling using a Gaussian copula framework enhances both the precision of marginal effect estimates and the understanding of comorbidity patterns. These findings support the design of integrated health interventions that address the interconnected nature of childhood illnesses in high-burden settings.