|Does a geographical context explain regional variation in child malnutrition in India?
|Yadav A, Ladusingh L, and Gayawan E
|Journal of Public Health, 23(5):277-287; DOI: 10.1007/s10389-015-0677-4
|Aim: Eradication of child malnutrition is important to ensure the future of children and for its association with morbidity, mortality, and impaired childhood development. This paper investigates the influence of geographical context on child malnutrition by mapping the residual net effect of malnutrition while controlling for bio-demographic and socioeconomic risk factors simultaneously.
Subject and method: We used data from the publically available National Family Health Survey to investigate the important risk factors of childhood diarrhoea, fever and acute respiratory infection using Bayesian geo-additive semi-parametric models.
Results: The results indicate that geoaddative models are needed to adequately assess nonlinear covariates effects and geographic effects within a joint model. With a traditional regression model these effects are difficult to model and to detect. Further, there were considerable geographical variations in child malnutrition across states in India. It is spatially structured and rates remain very high in the central region in comparison to other regions of India. Malnutrition was significantly high for children suffering from diarrhoea, fever and acute respiratory infection. Further, sex, birth order, consumption of Vitamin A, breastfeeding, caste, religion and wealth quintile were found to have significant effects on malnutrition.
Conclusion: The findings suggest that both spatial effect and morbidity, including individual and household factors have significant impact on malnutrition among children. Understanding these relationships can facilitate design of intervention programs. Clearly tackling the problem of malnutrition in India will require multisectoral interventions, targeted at all levels of social organization.
Keywords: Child malnutrition; Child morbidity; Geoaddative regression; Markov Chain Monte Carlo