Abstract:
|
Objectives: Evidence in the literature has shown that the first 1000 days are the most important in determining later children growth patterns. Furthermore, the determinants of growth were shown to span remote factors such as the national GDP, regional environmental factors such as sanitation, and direct factors such as nutrition, infections and family situations. The objectives of this analysis were to determine the most influential determinants of growth above two years of life. Methods: The data came from the Healthy Birth, Growth and Development knowledge integration (HBGDki) low to middle income longitudinal studies. The analysis combined quantile regression or GAMLSS with boosting algorithms to allow exploring high number of covariates. Model discrimination was based on goodness of-fit plots, predictive checks, and the generalized Akaike criterion. Results: The quantile regression analysis has detected factors significant on the 5% quantile but not on the median such as breastfeeding. GAMLSS gave equivalent results on the median. Conclusion: Additional data is being added to characterize the potential differential effects of key factors on the lower growth quantiles.
|
ASA Meetings Department
732 North Washington Street, Alexandria, VA 22314
(703) 684-1221 • meetings@amstat.org
Copyright © American Statistical Association.