A Statistical Study on Risk Factors for Low Birth Weight Using Logistic Regression & Roc Curve | Original Article

ABSTRACT:

The Focus of the study is to identify mother’s behavioral characteristics that has significant effect on child’s birth weight. Low birth weight is the leading cause of infant and child mortality and contributes to several poor health outcomes. Proper knowledge of risk factors of low birth weight is important for identifying those mothers at risk and thereby for planning and taking appropriate actions. The study proposes to evaluate maternal periodontal parameters to predict preterm delivery and low birth weight delivery to correctly classify between low birth weight cases and non-cases. The discriminatory performance of binary logistic regression model is measured using two approaches. The first approach is the use of fitted binary logistic regression model to correctly predict the subjects that are cases and non-cases. The alternative approach is based on receiver operating characteristic (ROC) curve for the fitted binary logistic regression model and then determining the area under the curve (AUC) as a measure of discriminatory performance. The data is abstracted from Baystate Medical Centre Springfield, Massachusetts. This data set contains information on 189 births to women seen in the obstetrics clinic and 59 of these births were low birth weight. The goal of the study is to determine whether these variables were risk factors in the clinic population being studied by Baystate Medical Centre. The logistic regression model is built in R-language. The present analysis identifies that mother’s smoking status (p=0.024), presence of hypertension (p=0.032), presence of Uterine cancer (p=0.097) and history of premature labor (p=0.004) are statistically significant on her neonate birth weight. However, to reduce the infant mortality due to low birth weight this study suggests that a mother should be non-smoker, free of hypertension, free of Uterine cancer and without any premature labor.