India has substantially reduced the burden of under-five child malnutrition over the last two decades. Despite this, it is still gigantic and differs remarkably across districts, while the demographic and socio-economic groups are most affected by it. This paper aimed to decrypt the place-specific spatial dependence and heterogeneity in associations between district-level nutritional status (stunting, wasting and underweight) and its considered correlates using a geocoded database for all 640 Indian districts from the latest fourth wave of the National Family Health Survey, 2015–16. Univariate Moran’s I and LISA statistics were used to confirm the spatial clustering and dependence in under-five nutritional status. The Ordinary Least Square (OLS), Geographically Weighted Regression (GWR), Spatial (lag/error) models were employed to examine the effects of correlates on the district-level nutritional status. The mean (Moran’s I) district-level stunting, wasting and underweight were 38% (0.634), 21% (0.488) and 36% (0.721), respectively. The GWR results disclosed that the spatial heterogeneity in relationships between district-level nutritional status and its driving forces were strongly location-based, altering their direction, magnitude and strength across districts. Overall, the localized model performed better, and best fit the data than the OLS and spatial (lag/error) models. This nationwide study confirmed that the spatial dependencies and heterogeneities in the district-level nutritional status indicators were strongly explained by a multitude of factors and thus can help policymakers in formulating effective nutrition-specific programmatic interventions to speed up the coverage of under-five malnutrition status in most priority districts and geographical hot spots across India.

在过去的二十年里，印度大大减轻了五岁以下儿童营养不良的负担。尽管如此，它仍然是巨大的，并且在不同地区之间存在显着差异，而人口和社会经济群体受它的影响最大。本文旨在使用来自最新第四波国家家庭的所有 640 个印度地区的地理编码数据库，解密地区级营养状况（发育迟缓、消瘦和体重不足）之间关联的特定地点空间依赖性和异质性及其考虑的相关性健康调查，2015-16 年。单变量 Moran's I和 LISA 统计数据用于确认 5 岁以下营养状况的空间聚类和依赖性。使用普通最小二乘 (OLS)、地理加权回归 (GWR)、空间（滞后/误差）模型来检查相关性对地区营养状况的影响。平均值（Moran's I) 地区发育迟缓、消瘦和体重不足的比例分别为 38% (0.634)、21% (0.488) 和 36% (0.721)。GWR 结果显示，地区营养状况与其驱动力之间关系的空间异质性强烈地基于位置，改变了它们在地区间的方向、幅度和强度。总体而言，本地化模型比 OLS 和空间（滞后/误差）模型表现更好，并且最适合数据。