'J-shaped' relationship between drinking during pregnancy and birth weight: reanalysis of prospective epidemiological data.

Worldwide epidemiological studies on the effects of drinking during pregnancy on birth weight and prematurity were surveyed. Far more studies have reported no significant effects on birth weight than have reported significant decreases. Statistical analyses of the means from the prospective studies in this area indicated that both maternal smoking and alcohol consumption during gestation are associated with a significant decrease in birth weight. The effect of smoking is three times greater than the effect of alcohol. When the data were stratified by smoking status, maternal alcohol consumption did not have a significant effect on birth weight for non-smokers, but among smokers there was a significant linear trend with a threshold for decreased birth weight at about an average of two drinks per day. There was also a significant pattern of increased birth weight associated with low levels of alcohol consumption, suggesting an inverted 'J-shaped' function between drinking during pregnancy and birth weight.