Reinterpreting the effects of maternal smoking on infant birthweight and perinatal mortality: a multivariate approach to birthweight standardization.

Infants of women who smoke during pregnancy have lower birthweights and have been observed to have higher rates of perinatal mortality than infants of non-smokers. It is not clear whether this increased risk of mortality is due to an excess of small births among smokers or to an independent effect of smoking. Although infants of smokers have overall higher mortality rates than non-smokers, low birthweight (< 2500 g) infants of smokers have lower mortality rates than low birthweight infants of non-smokers. However, comparison of birthweight-specific mortality between two groups is problematic when there are differences in the birthweight distributions. Methods that have been developed to standardize for these differences by comparing mortality rates relative to their own mean do not allow for simultaneous control of confounding variables. Using data from over 13,000 births of women who participated in a prepaid health care plan we present a method to standardize for birthweight while adjusting for variables that may confound the relationship between maternal smoking and perinatal mortality. After controlling for race, maternal age, education, parity, and number of cigarettes smoked, we found that 85% of the increased mortality due to smoking was attributable to an excess of small births in the birthweight distribution of offspring of smoking mothers, while 15% was due to higher birthweight-specific mortality at almost all standardized birthweights. Contrary to previous reports, we found that low birthweight infants of smoking mothers are at higher risk of perinatal mortality if a population-specific standard for birthweight is used.