Analysing outlier communities to child birth weight outcomes in Malawi: application of multinomial logistic regression model diagnostics

Abstract

Studies have reported significant effect of geographically shared variables on new-born baby weight. Although there is growing use of community-based child health interventions in public health research, such as through provinces, schools, or health facilities, there has been less interest by researchers to study outlying communities to child birth weight outcomes. We apply multinomial logistic regression model diagnostics to identify outlier communities to child birth weight in Malawi. We use a random sample of 850 clusters, each with at least 7 households based on 2015-16 Malawi demographic and health survey data. There were a total of 11,680 children with measured birth weight, that was categorised as either low (< 2,500 grams), normal (2,500 - 4,000 grams) or high (> 4,000 grams). The analyses were done in STATA version 15 and R version 3.6.3. Based on a multinomial logit model with various socio-demographic factors associated with child birth weight, the results showed that two clusters from rural parts of Southern region of Malawi had overly influence on estimated effects of the factors on birth weight. Both clusters had normal to high birth weight babies, with no child having low birth weight. There could be some desired motherhood practices applied by mothers in the two rural clusters worth learning from by policy makers in the child healthcare sector.