Relative contributions of the correlates of stunting in explaining the mean length-for-age z-score difference between 24-month-old stunted and non-stunted children living in a slum of Dhaka, Bangladesh: results from a decomposition analysis

Abstract

Objective: Using MAL-ED (Etiology, Risk Factors, and Interactions of Enteric Infections and Malnutrition and the Consequences for Child Health) Bangladesh birth cohort data, we sought to measure the relative contributions of the most predictive correlates of stunting to mean length-for-age z (LAZ) score difference between stunted and non-stunted children at 24 months of age.

Setting: Dhaka, Bangladesh PARTICIPANTS: 211 slum-dwelling children enrolled within 17 days of their birth.

Variables and method: The explanatory variables were identified from the following groups: maternal characteristics, birth characteristics, macronutrient intake, socioeconomic status, morbidity and serum micronutrient level. At step 1, predictive correlates of stunting were identified longitudinally (from 9 to 24 months of age) using generalized estimating equations (GEE) model. Then, the relative contributions of the most predictive correlates of stunting to mean LAZ score difference between stunted and non-stunted children at 24 months of age was measured using Blinder-Oaxaca decomposition analysis RESULTS: The GEE multivariable model identified maternal height, birth weight, people per room, gender, having separate room for kitchen and energy intake as the most predictive correlates of stunting. At 24 months, mean LAZ score difference between stunted and non-stunted children was 1.48. The variable by variable decomposition of the LAZ gap identified maternal height (coefficient: -3.04; 95% CI: 0.35 to -6.44), birth weight (coefficient: -0.21; 95% CI: 0.88 to -1.30), people per room (coefficient: 0.31; 95% CI: 0.92 to -0.30) and energy intake (coefficient: -0.12; 95% CI: 0.22 to -0.46) as the top most factors responsible for the mean LAZ score difference between stunted and non-stunted children at 24 months of age.

Conclusions: The relative contributions of maternal height and birth weight to LAZ gap signifies that improvement in nutritional status of a women during her adolescence and pregnancy would have an impact on birth weight of her offspring, and ultimately, on linear growth of the child.

Keywords: decomposition analysis; length-for-age z-score; maternal height; relative contribution; stunting.