Longitudinal model of infant hemoglobin: evidence of chained mediation in a peruvian cohort

In Peru, the persistence of childhood anemia despite interventions reveals gaps in understanding its longitudinal progression. Objective: To evaluate hemoglobin evolution in children under three years of age using a chain mediation model, identifying the role of intermediate measurements on subsequent values. Patients and Method: Analytical longitudinal study using secondary data from hemoglobin screenings in children under three years from Junín, Peru. A total of 2,564 children with four consecutive hemoglobin measurements (baseline, 1, 3, and 6 months) were included. Hemoglobin (Hb) concentrations were the primary variables, with age as a covariate. A chain mediation model was applied with 5,000 bootstrap resamples. A directed acyclic graph was developed to repre-sent the proposed causal structure. Results: Baseline Hb was significantly associated with subsequent measurements. It strongly predicted Hb at the first month (B = 0.8743; 95%CI: 0.8482–0.9004), and along with this, explained Hb levels at three months (B = 0.3796 and 0.5296; p < 0.001). At six months, all three prior measurements were significant predictors, with Hb carrying the greatest weight at three months (B = 0.3672; 95%CI: 0.3296–0.4048). Age showed a positive effect at all stages. The chain mediation model revealed both direct and indirect effects of baseline Hb on Hb levels at six months (total effect = 0.4508), highlighting key mediation pathways through Hb at 1 and 3 months. Conclusions: Baseline hemoglobin predicts levels at six months through direct and indirect path-ways, underscoring the value of sequential monitoring and the third month as a critical point for intervention.