Social
Caste-based social inequalities in child undernutrition in India: exploring the role of environmental and health services interventions using mediation analyses Arijit Nandi* Arijit Nandi Sam Harper
Child undernutrition has improved in India over the past 30 years, but social inequalities persist. Children from historically disadvantaged groups, including those designated by the Indian Government as belonging to a scheduled caste (SC), scheduled tribe (ST), or other backward caste (OBC), have higher rates of stunting, wasting, and underweight than other children. Mediation analysis can help to identify interventions for eliminating these inequalities.
Using a sample of 490,986 under 5-year-old children born between 2010-2020 and included in the Indian National Family Health Survey, we quantified caste-based inequalities in stunting, wasting, and underweight, defined using measured height and weight. We then used marginal structural models with inverse probability weights (IPWs) to examine if two distinct potential interventions, increasing coverage of delivery in a health sector facility with assistance from health personnel and improved household drinking water and sanitation facilities, mitigated these inequalities.
82.6% of children belonged to a SC, ST, or OBC, and 37.0% of children were stunted, 19.5% were wasted, and 32.8% were underweight. The prevalence of stunting, wasting, and underweight were 8.6 (95%CI: 8.2, 9.0), 2.7 (95%CI: 2.3, 3.0), and 8.1 (95%CI: 7.7, 8.5) percentage-points higher, respectively, among children belonging to a SC, ST, or OBC compared to those who did not, representing the total effects. Mediation models using IPWs to estimate controlled direct effects (CDE) on the prevalence difference scale showed that setting coverage of the interventions to 100% eliminated nearly one-half of the inequality for wasting (CDE=1.3, 95%CI=-0.6, 3.2 for health facility delivery; CDE=1.3, 95%CI=-0.5, 3.2 for improved water and sanitation), but had a minimal effect on disparities in stunting and underweight. Results were compared to analyses using the “product method” and randomized intervention analogues to natural direct and indirect effects.