On exploring and ranking risk factors of child malnutrition in Bangladesh using multiple classification analysis

BACKGROUND: Logistic regression analysis is widely used to explore the determinants of child malnutrition status mainly for nominal response variable and non-linear relationship of interval-scale anthropometric measure with nominal-scale predictors. Multiple classification analysis relaxes the linearity assumption and additionally prioritizes the predictors. Main objective of the study is to show how does multiple classification analysis perform like linear and logistic regression analyses for exploring and ranking the determinants of child malnutrition. METHODS: Anthropometric data of under-5 children are extracted from the 2011 Bangladesh Demographic and Health Survey. The analysis is carried out considering several socio-economic, demographic and environmental explanatory variables. The Height-for-age Z-score is used as the anthropometric measure from which malnutrition status (stunting: below −2.0 Z-score) is identified. RESULTS: The fitted multiple classification analysis models show similar results as linear and logistic models. Children age, birth weight and birth interval; mother’s education and nutrition status; household economic status and family size; residential place and regional settings are observed as the significant predictors of both Height-for-age Z-score and stunting. Child, household, and mother level variables have been ranked as the first three significant groups of predictors by multiple classification analysis. CONCLUSIONS: Detecting and ranking the determinants of child malnutrition through Multiple classification analysis might help the policy makers in priority-based decision-making. TRIAL REGISTRATION: “Retrospectively registered” ELECTRONIC SUPPLEMENTARY MATERIAL: The online version of this article (10.1186/s40795-017-0194-7) contains supplementary material, which is available to authorized users.