Numerous studies have shed light on the effects of maternal smoking during pregnancy on the offsprings’ health outcomes. In particular, maternal smoking during pregnancy was found to be associated with lower offspring birthweight. Previous studies normally address confounding by adjusting for maternal age, socioeconomic status, breastfeeding and offspring gender. However, concerns about unmeasured confounding biases due to the mother’s lifestyle, the households’ socioeconomic status, and mother’s genotypes still remain.

The negative control method is a recently popularized approach that utilizes measured covariates as proxies of unmeasured confounders, in order to detect, reduce and correct for confounding bias. This method may help eliminate the confounding bias when studying the causal effect of maternal smoking on child birth weight. For example, paternal alcohol consumption during the mother’s pregnancy can serve as a potential negative control variable since it’s a proxy of the household’s socioeconomic status but shouldn’t directly affect the offspring’s birth weight. Therefore, any observed association between paternal alcohol consumption and the child’s birth weight is likely a result of confounding bias. Other negative control variables may include negative control exposures (NCEs) such as mother’s smoking behavior after giving birth, and negative control outcomes (NCOs) such as the child’s injuries since age 9.

In this project, we will utilize multiple negative control variables to obtain a more accurate estimate of the causal effect of maternal smoking during pregnancy on infants’ birthweight with the ALSPAC data.