Limitations that apply to new risks in GBD 2016 or risks with significant estimation updates are presented here. For low birthweight and short gestation, we have included the effect of low birthweight and short gestation only on neonatal outcomes; we have not found the evidence to meet our inclusion criteria for the link between low birthweight and short gestation and NCDs in adult age groups. Our analysis of RRs has used a very large US-linked birth cohort dataset and much more limited data from middle-SDI and low-SDI populations. Given the large number of observations from the USA, our results are heavily influenced by the pattern of RRs across birthweight and gestational age in that population. The microdata used to develop the ensemble distributions for birthweight and gestational age are largely from middle-SDI and high-SDI locations. The estimation of alcohol use relies heavily on sales data, which are limited and whose quality we cannot easily assess. Also, the estimation of unrecorded consumption of alcohol is based on limited data and has significant uncertainty; nevertheless, we feel it is important to include
