Causal Inference in the Association Between Maternal Anaemia and Child Stunting in Nigeria: A Mendelian Randomisation Study

Observational studies have consistently reported associations between maternal anaemia during pregnancy and child stunting, but the causal nature of this relationship is difficult to establish due to shared socioeconomic confounders and reverse causality. This study applies Mendelian randomisation as an instrumental variable approach to test whether maternal anaemia during pregnancy exerts a causal effect on child stunting outcomes in a Nigerian population, using genetic variants associated with haemoglobin concentration as instruments. A two-sample Mendelian randomisation analysis was designed using data from two independent sources: a sample of 2,000 Nigerian women of reproductive age enrolled in the Nigeria Genomic and Epidemiological Network study for genetic instruments relating to haemoglobin and iron status, and a separate child cohort study of 1,500 mother-child pairs from Kano, Lagos, and Cross River States with prospective stunting measurements at 12, 18, and 24 months of age and maternal haemoglobin data from pregnancy. Single nucleotide polymorphisms associated with haemoglobin concentration in West African populations were identified and validated. Two-stage least squares estimation with robust standard errors was used for the primary causal analysis. Sensitivity analyses including weighted median, MR-Egger, and MR-PRESSO were applied to test for and address potential violation of instrument validity assumptions. A dose-response analysis examined whether the causal effect varied by severity of anaemia. Findings will provide the first Mendelian randomisation-based causal evidence on the maternal anaemia-child stunting relationship in a Nigerian population, with significant implications for prenatal anaemia intervention prioritisation. Keywords: Mendelian randomisation, maternal anaemia, child stunting, causal inference, Nigeria.