Gestational age estimates from singleton births conceived using assisted reproductive technology

Information on gestational age for public health research and surveillance in the US is usually obtained from vital records and is primarily based on the first day of the woman's last menstrual period (LMP). However, using LMP as a marker of conception is subject to a variety of errors and results in misclassification of gestational age. Pregnancies conceived through assisted reproductive technology (ART) are unique in that the estimates of gestational age are not based on the LMP, but on the date when fertilisation actually occurred, and thus most gestational age errors are likely to be due to errors introduced in recording and data entry. The purpose of this paper was to examine the birthweight distribution by gestational age for ART singleton livebirths reported to a national ART surveillance system. Gestational age was categorised as 20-27, 28-31, 32-36 and 37-44 weeks; birthweight distributions were plotted for each category. The distributions of very-low-birthweight (VLBW; < 1500 g), moderately low-birthweight (1500-2499 g) and normal-birthweight infants for each gestational week were examined. At both 20-27 and 28-31 weeks, there was an extended right tail to the distribution and a small second mode. At 32-36 weeks, there were long tails in either direction and at 37-44 weeks, an extended tail to the left. There was a high proportion of VLBW infants at low gestational ages and a decreasing proportion of VLBW infants with increasing gestational age. However, there was also a fairly constant proportion of normal-birthweight infants at every gestational age below 34 weeks, which suggested misclassification of gestational age. Approximately 12% of ART births classified as 28-31 weeks' gestation had a birthweight in the second mode of the birthweight distribution compared with approximately 29% in national vital statistics data. Even when the birthweight and dates of conception and birth are known, questions remain regarding the residual amount of misclassification and the true nature of the birthweight distributions.