Back to browse results
Fitting Ordinal Regression Analysis to Anthropometric Data
Authors: Pradhan A
Source: Journal of Nepal Health Research Council, 9(18): 61-6; DOI: 10.33314/jnhrc.v0i0.257
Topic(s): Birth weight
Child health
Children under five
Wealth Index
Country: Asia
Published: NOV 2011
Abstract: Background: National level surveys on nutritional status since 1975 to 2006 in Nepal do not indicate the satisfying level of nutrition. Nepal demographic and health survey 2006 uncover that the percent prevalence for underweight and wasted children of under five years of age are 39% and 13% and 49 % of the under five children are stunted. Understanding the factors that affect the nutrition of children is essential. Some studies in other countries show wealth index, size at birth and education as significant contributors. This analysis analyze the factors associated with nutritional status among children of under five years of age. Methods: This study was cross sectional which used secondary data of the Demographic and Health Survey, 2006 conducted in Nepal. STATA 9, SPSS 13 and SPSS 17 are used for analysis. In this analysis, the outcome variables namely stunting, underweight and wasting are in ordered form. Hence ordinal regression is considered as suitable method. Results: Ordinal regression well suit the data to model nutritional status through different predictors in case of underweight and wasting however stunting model fails to satisfy the assumption behind ordinal regression. Hence for stunting, model with constraints imposed to certain variables is formed. Conclusions: Underweight is seen significantly less in households with high wealth index, among children of big and average size at birth and among educated women. Education, wealth index and size at birth are found important factors affecting wasting among children. Wealth index and education of mothers are significantly affecting for stunting among children. Keywords: ordinal regression, stunting, underweight, wasting.