Maternal smoking during pregnancy and birthweight: a propensity score matching approach

Abstract

There is accumulated evidence of the existence of a deleterious effect of smoking on birth outcomes. Whether there is a causal link or a mere statistical association is not clear. Understanding the effect of smoking on pregnancy is a critical issue because of the public policy implications for dissuading maternal smoking. This study was designed to distinguish causal links from statistical association in the relationship between fetal exposure to maternal smoking and birth outcomes. Although the task involves several aspects of estimation we restrict our focus to the issue of self-selection. We explore this issue by using the propensity score method and compare that with parametric estimators. First we estimate the treatment effect of smoking during pregnancy on different birth outcomes. Then, we extend the method to the case of the multi-treatment “intensity of smoking”. The deleterious effect of smoking is found robust to the different estimation methods used.