Abstract

Background: This paper deals with three key indicators of anthropometric failure viz. stunting, wasting, and underweight, their significant correlates, and spatial dependence across 640 districts of India. The paper uses data from three different rounds of NFHS conducted from 1998 to 2015. The spatial analysis uses district-level information collected first time in NFHS-4 (2015-16).

Methods: Analytical methods used in this paper include multivariate logistic regression to examine the adjusted effects of various predictors and predicted probabilities to explain changes in the likelihood of anthropometric failures over time. Besides, bivariate LISA maps and spatial error models provide spatial dependence and clustering in anthropometric failures among children.

Results: Results highlight that women’s education, maternal nutrition, birth order, birth weight, and the wealth quintiles of households were essential markers of the anthropometric failures among the children. The likelihood of anthropometric failures declined considerably during 1998-2005 in comparison to 2005-15, especially among the children from the wealthiest quintiles. Spatial clustering in the prevalence of anthropometric failures portrayed that Moran’s I values were significant for the utilization of ICDS services and mothers having low BMI. The univariate Moran’s I statistics were 0.62 and 0.72 for stunting and underweight, respectively. When spatial weights were considered, the autoregression model noticeably became stronger in predicting the prevalence of stunting and underweight.

Conclusions: Evidence on significant correlates and spatial dependence of anthropometric failure shed the importance of strengthening multisectoral convergence in various nutrition-specific and nutrition-sensitive interventions in combating the anthropometric failures in the context of nature and patterns emerging in various hotspots and cold spots.

Keywords: Anthropometric failures; Nutrition; Stunting; Underweight; Wasting; Moran’s I; Spatial autocorrelation; Autoregression