Globally, 148 million children under 5 are stunted, with risk factors varying by context. Our 'Impact of Growth Charts and Nutritional Supplements on Child Growth in Zambia' (ZamCharts) trial observed persistently high rates of stunting in all treatment groups after 18-months of intervention with monthly distributions of small-quantity lipid-based nutrient supplements (SQ-LNS) and/or installation of a wall-mounted growth chart in children's homes. We sought to identify determinants of stunting and height-for-age z-score in children 27-36 months who participated in the ZamCharts endline survey (n=1911). Multi-level, log-binomial models were used to estimate univariable and multivariable prevalence ratios for predictors of stunting. Multi-level models were also used to predict height-for-age z-score (HAZ) with and without baseline HAZ (assessed at ages 2-11 months). We also conducted a path analysis using covariance analysis of linear structural equations to assess underlying and modifiable risk factors for impaired linear growth. Significant predictors of stunting in the multivariable model included: low asset ownership; being male; using biomass as cooking fuel; lower maternal height; a mother with ≤ primary education; lower baseline HAZ. and not being randomized to SQ-LNS. Significant predictors of a lower mean HAZ in the full multivariable models included all the same risk factors that predicted stunting, but also living in an urban area, having more than one child <5 years in the household and diarrhea in the previous 2 weeks. The multivariable model explained 48% of variability in endline HAZ; the strongest predictor was baseline HAZ, which explained 29% of endline HAZ variability in the univariable model. Preventing stunting in Zambia will require investments in early life (pre- and post-natal) determinants of growth trajectory as well as improving complementary feeding practices and addressing risk factors for infectious diseases; SQ-LNS can also improve linear growth and reduce stunting.
Read full abstract