GEO-ADDITIVE LOGISTIC MODELLING OF CHILDHOOD ANAEMIA IN MALAWI

Abstract

Anaemia during childhood remains a major public health challenge in Malawi. Understanding anaemia risk factors can help health professionals identify the groups that are more vulnerable to the morbidity. Epidemiological studies in Malawi on child anaemia have neglected the community spatial effect to childhood anaemia. Neglecting community spatial effect in models ignores the influence of unobserved or unmeasured contextual variables, and at the same time the resultant model may under estimate model parameter standard errors which can result in over significance of covariates. This study therefore aimed at investigating risk factors of childhood anaemia in Malawi with focus on geographical spatial effect. The study used the 2010 Malawi demographic healthy survey data. Child anaemia status was defined as binary and fitted the binary logistic model for the two categories (anaemia (Hb < 11), and no anaemia (Hb ≥ 11) where no anaemia was the base line category. Similary, a cumulative logit model was fitted for the three categories outcome (no anaemia (Hb ≥ 11), moderate to mild anaemia (7 ≤ Hb ≤ 10.9) and severe anaemia (Hb < 7), where severe anaemia was the base line category. A geo-additive model estimated by empirical Bayes method was adopted to simultaneously account for residual spatial effects, non linear effect of continuous covariates, and fixed effects of childhood anaemia. In both models, continuous covariates were modelled by the penalized splines, spatial effects were smoothed by the two dimensional penalized spline, and fixed effects were assigned a diffuse prior. The results showed evidence of distinct residual spatial patterns that were not explained by the socioeconomic and demographic factors. In particular Nsanje, Chikwawa, Salima, Nkhota-Kota, Mangochi and Machinga increased the risk of childhood anaemia. Karonga, Chitipa, Rumphi, Mzimba, Ntchisi, and Chiradzulu reduced the risk of childhood anaemia. Socioeconomic and demographic determinants generally confirmed what is known in the literature. Maternal anaemia, child stunting, and child fever, had a positive effect on child anaemia. Childhood anaemia decreased with child age. It also decreased with wealth index. There was a U relationship between child anaemia and mother age. Strategies in childhood anaemia control should be tailored to local conditions, taking into account the specific etiology and prevalence of anaemia.