Menstrual cycle length and the calculation of gestational age

The method of estimating gestational age that is most commonly employed by epidemiologists calculates age as the time between the first day of the last menstrual period and the date of birth. The method assumes an invanant 28-day menstrual cycle with ovulation occurring at midcycle. Data from 5,688 mothers and their infants who were enrolled in the Boston sector of the National Collaborative Perinatal Project from 1959 through 1965 were used to study whether this is a valid assumption by examination of the association between gestational age and cycle length. In addition, two methods for adjusting gestational age for cycle length were explored: 1) the Ogino-Knaus adjustment which assumes that the postovulatory phase of the menstrual cycle is a constant 14 days long, and 2) an empirically derived approach. Cycle length was grouped into five categories. With the last menstrual period method, the proportion of postterm (more than 42 weeks) and postdate (more than 40 weeks) births increased with longer cycle lengths (from 0 to 14.6% for postterm and from 21.0 to 53.7% for postdate). With the Ogino-Knaus correction, the trend was weaker and in the opposite direction (10.5 to 2.4% for postterm and 57.9 to 22.0% for postdate). With the empirically derived method, these trends were virtually eliminated (5.3 to 9.8% for postterm and 36.7 to 36.5% for postdate). The results suggest that the previously reported tendency for the last menstrual period method to yield a high proportion of postdate and postterm deliveries may, in part, be due to misclassification of gestational age in women with long menstrual cycles. The Ogino-Knaus method appears to overcorrect for cycle length. The empirically derived method needs further validation; however, it does suggest potential directions to be explored.