Objective: The purpose of this study was to evaluate the accuracy of algorithms for the assignment of gestational age with the use of the last menstrual period and early ultrasound information.
Study design: Gestational age estimates that are based on last menstrual period, ultrasound scanning, or their combination were compared among women who attended prenatal care clinics in central North Carolina (n = 3655) by an evaluation of digit preference in the last menstrual period dates and a comparison of mean gestational age, preterm and postterm categories with the use of kappa statistics, difference between actual and expected delivery date, and birth weight among subgroups with discrepant assignments.
Results: Last menstrual period reports show digit preference, assign gestation 2.8 days longer on average than ultrasound scanning, yield substantially more postterm births (12.1% vs 3.4%), and predict delivery among term births less accurately. Misclassification of births as postterm was more common in younger women, those of nonoptimal prepregnancy body weight, cigarette smokers, and women who reported last menstrual period using preferred dates of the month.
Conclusion: Last menstrual period estimates of the duration of gestation are subject to both random error and a systematic tendency to overstate the duration of gestation, most likely because of delayed ovulation.