Starting with 94,189 pregnancies that ended between January 1, 1986, and December 31, 1993, we sequentially excluded multiple births (n = 2,859) and women with missing data on smoking during pregnancy {n — 4,146), which left us with 87,184 pregnancies among 61,667 women for analysis. Gestational age was estimated on the basis of an algorithm incorporating both the clinical estimate of gestational age at delivery and the last menstrual period. When the difference in gestational age between the clinical estimate and the date of the last menstrual period was greater than 3 weeks, the charts were reviewed to determine the source of the disparity. If the discrepancy could not be resolved, the clinical estimate was retained. When the disparity was less than 3 weeks, the estimate based on the last menstrual period was retained. If only the clinical estimate or the estimate based on the last menstrual period was available, that particular estimate was used in isolation. If neither of these estimates was available, gestational age was coded as missing.
Smoking during pregnancy Information on cigarette smoking was collected both at the first prenatal visit and at the time of admission to the hospital for delivery. Women were asked whether they had ever smoked during the current pregnancy and, if they were smokers, the number of cigarettes smoked per day. Since the Atlee database contains only one smoking variable, changes in smoking behavior during the course of pregnancy were not recorded. If the average number of cigarettes smoked per day differed between the first prenatal visit and the time of delivery, the highest number was recorded in the database (19).