MEASURING GESTATIONAL AGE WITH OVULATION TO BETTER UNDERSTAND PREGNANCY LENGTH, EMBRYONIC GROWTH, AND THEIR PREDICTORS

Gestational age assignment across pregnancy and at delivery is used to time prenatal care, provide necessary interventions, and measure outcomes, including preterm birth and growth restriction. The common methods for assigning gestational age, last menstrual period (LMP), and ultrasound are error-prone. Ovulation is a third method with higher accuracy. This dissertation quantifies the measurement error and changes in exposure-outcome associations that arise when using LMP and ultrasound-based gestational age measures.We used fertility monitor data from the Effects of Aspirin in Gestation and Reproduction (EAGeR) trial for all analyses. We defined preterm birth by the clinical cut point of 37 weeks, and we determined size for gestational age by the INTERGROWTH-21 standard.In aim 1, we found that pregnancies were longest, preterm birth least common (prevalence = 0.07, 95% CI: 0.04, 0.10), and small for gestational age most common when LMP was used to estimate gestational age at delivery. The opposite was true for ultrasound which estimated the prevalence of preterm birth to be 0.10 (95% CI: 0.07,0.13). Ovulation, our gold standard measure, identified the prevalence of preterm birth to be 0.08 (95% CI: 0.06, 0.12). LMP showed better specificity but worse sensitivity than ultrasound for detecting preterm birth.In aim 2, we developed growth curves for LMP and ovulation-based gestational age at early study ultrasound and selected by adjusted R2. Compared with ovulation, we found that crown-rump length was smaller when LMP was used to measure gestational age. As such, the growth trajectory was steeper in the curve defined by ovulation than in the curve defined by LMP. In aim 3, we measured the prevalence difference in the associations of three predictors: aspirin, maternal body mass index, and maternal age, with two outcomes: preterm birth and small for gestational age. Among older participants, we observed slight differences in the prevalence of preterm birth for gestational age assessed by LMP and ultrasound compared to ovulation. These findings suggest that the technique used to measure gestational age may impact analyses, and this error's magnitude and direction can be predicted.