Addressing the poverty-nutrition gap for SDG 1 and SDG 2: a Bayesian geo-additive analysis of spatial heterogeneity and nonlinear maternal risks in childhood stunting in India (NFHS-4 and NFHS-5).

Authors

V Vinay Shankar, G Rithanya Grishma, R Niveditha, K Sathya Narayana Sharma, Saralees Nadarajah, S Yashuvanthra Dhevi, J E Yuvaraj Kumar, G S Heerthihaa

Published in

Frontiers in public health. Volume 14. Pages 1854301. Epub Aug 03, 2026.

Abstract

Childhood stunting remains a critical indicator of chronic undernutrition and a major barrier to achieving Sustainable Development Goals 1 and 2. Despite gradual improvements, substantial geographic disparities in stunting risk persist across India. This study characterised the spatial and temporal heterogeneity of childhood stunting using harmonized district-level data from NFHS-4 (2015-16) and NFHS-5 (2019-21).
A Bayesian spatial geo-additive logistic regression model was applied to 482,960 children aged 0-59 months from 575 harmonized districts. The framework incorporated district-level BYM2 spatial random effects, district-specific survey-round interaction effects, and RW2 smoothers for child age and maternal BMI, estimated using Integrated Nested Laplace Approximation (INLA).
The spatial interaction model demonstrated substantially improved fit over the non-spatial model (ΔWAIC = 4,105.2), with a high estimated spatial fraction (φ = 0.983; 95% CrI: 0.938-0.999), indicating that most residual district-level variation was spatially structured. Model predictions showed strong agreement with weighted survey prevalence estimates (Pearson r = 0.975; R ² = 0.950). After accounting for district-specific temporal interactions, there was no evidence of a uniform national change in stunting risk between survey rounds (AOR = 0.986; 95% CrI: 0.960-1.013), with substantial heterogeneity in district-level temporal trajectories. Stunting risk peaked between 18 and 24 months of age and was elevated among children of severely undernourished mothers. Based on 95% posterior credible intervals of district-specific survey-round interaction effects, 59 districts demonstrated credible improvement, 35 demonstrated credible worsening, and 481 showed no credible evidence of change. A joint dOR-prevalence framework identified 213 districts (37.0%) as highest-priority areas, combining elevated adjusted district-level risk with high absolute stunting burden.
These findings demonstrate that childhood stunting in India remains strongly patterned by geography, with temporal changes varying substantially across districts. Geographically targeted and maternal-focused interventions will be essential for achieving equitable reductions in stunting and accelerating progress toward national nutrition goals.

PMID:
42609395
Bibliographic data and abstract were imported from PubMed on 18 Aug 2026.

Read full publication at:
Please sign in to see all details.

Stats

1-terrible, 9-excellent. How would you rate this publication? Sign in in to submit your rating.

Post a comment

You need to be signed in to post comments. You can sign in here.

Comments

There are no comments yet.