Patterns of change in human birth weight distributions under the influence of a model nonspecific toxic exposure: Direct maternal cigarette smoking

Fetal growth restriction is one of the most common findings in animal toxicology studies of reproductive/developmental effects for environmental chemicals. Projection of the possible implications of such effects for human health requires a mathematical model of how human birth weight distributions can be expected to change under the influence of a non-specific toxic exposure. The present analysis draws on recent U.S. national aggregate data from birth certificates on the birth weights of over ten million babies born to mothers who do and do not report cigarette smoking during pregnancy. These data are analyzed empirically both by straightforward interpolation of percentiles and by fitting to a mixture distribution of lognormals. The percentile interpolations of the aggregate data indicate differential effects between the higher and lower ends of the birth weight distribution. Specifically, the lighter end, which is at greater baseline risk of adverse effects (e.g., infant mortality, neurodevelopmental delays), has a smaller absolute change (in grams) due to smoking, but greater relative change (in log grams). The mixture distribution analysis indicates that the data are best summarized in the form of three component lognormals with modes at approximately 3400, 2640, and 1330 g, comprising about 84.4%, 13.2%, and 2.5% for nonsmoking mothers. For smoking mothers, the same analysis can be summarized similarly, with the three modes shifted downward by about 158, 53, and 10 g respectively, and the proportions of the population in the two lighter modes were increased to 18.7% and 3.2%, respectively. The robustness of these findings is examined in analyses of subsets of these data, classified by year, race, and gestational age, and hypotheses for the possible quantitative relationship with smoking are generated. Possible public health implications will be discussed for outcomes strongly associated with birth weights and for other exposures that may cause fetal growth retardation.