Statistical Analysis Of Risk Factors On Time To Weaning Of Breastfed Newborns

The issue of infants' health has always been a concern,and the research from WHO has shown that breast milk is the best food for newborns aged 0-2.The Global Strategy for Infant and Young Children Feeding recommends that the breastfeeding last for 6 months at least for every infant.However,facts are far from the target both in developed and developing countries.Thus it is of great significance and value to study the factors that significantly affect time of weaning as well as the magnitudes of their effects.Time to weaning data is failure time data with censoring and our purpose is to study its distribution as well as the effects of its risk factors.Our main objectives are finding out which risk factors may cause earlier weaning and giving advice to reduce the risk of early weaning.The data set that will be analyzed in this project comes from National Longitudinal Survey of Youth in U.S.This data set consists of individual records from 927 newborns whose mother chose to breast feed their children,which includes the duration of breastfeeding as response variable and race of mother,age of mother at child's birth,education,poverty status,smoking status during pregnancy,alcohol-drinking during pregnancy,year of child's birth and lack of prenatal care status as explanatory variables.We've made good use of survival analysis methods to analyze the data.In particular,we will use Kaplan-Meier method to estimate survivor function of breastfeeding.Also we will establish a stratified Cox regression model to estimate the effect of each explanatory variable.Considering that the factor of education does not meet the proportional hazard assumption,we propose to stratify the sample according to three academic levels: higher education,high school and below high school,which is our new idea for analyzing this data set.For each layer,a Cox regression model is established;and in order to obtain a regression model that describes the distribution of time to weaning in the most suitable way,AIC criterion will be introduced as a variable selection criterion,which can help interpret the data better,and we are interested in which factors are selected and how to explain the effects of them.In this project,some new analysis ideas are implemented to obtain new conclusions on the data.According to statistical inference results,12-month maternity leave in the United States serves as a risk factor on weaning;for the mothers with higher education,smoking during pregnancy and late year of child's birth will increase the risk of weaning,and mothers of white race has the longest breastfeeding duration while Asians and Hispanics have the shortest;for the mothers only with high school education,smoking and late year of child's birth will increase the risk;while for the mothers with education below high school,poverty is the most important factor that has effect on time to weaning,whose effect varies with time,with poor women tending to have shorter breastfeeding durations before 24 weeks while longer ones after 24 weeks.In summary,based on the analysis results,we provide some suggestions and explanations on the effects of various risk factors according to knowledge of medicine,social sciences as well as maternal and child health.