Dating Discrepancies on Research Ultrasonography and Risk of Pregnancy Loss in a Prospective Cohort

To estimate the risk associated with discrepancies between last menstrual period (LMP)-based and ultrasound-based gestational dating and pregnancy loss in a prospective cohort of individuals of normal fertility who underwent standardized early-pregnancy ultrasonography.

Participants in a community-based, prospective pregnancy cohort were recruited preconceptionally or in early pregnancy. Participants underwent standardized research ultrasonography targeted for the sixth week of gestation. We calculated the magnitude of lag between ultrasound-based age and LMP-based age at the research ultrasonogram. Cox proportional-hazards models were used to estimate the association between this difference and pregnancy loss. To assess for effect modification, analyses were stratified by week of research ultrasonogram, developmental features observed on the ultrasonogram, and menstrual regularity.

Among 4,935 participants, the median difference between LMP-based and ultrasound-based gestational age on the research ultrasonogram was 1 day (interquartile range -1 to 5 days), and 9.3% of pregnancies ended in loss. Risk of pregnancy loss increased exponentially with each additional day ultrasound-based dating lagged LMP-based dating ( P <.001). This association persisted when stratified by week of ultrasonography and was more pronounced among pregnancies with a measurable crown-rump length. Ultrasound-based gestational age lagging LMP-based gestational age by more than 3 days was associated with a fivefold increased risk of pregnancy loss (adjusted hazard ratio [HR] 5.34, 95% CI, 4.37-6.52), and a lag of more than 5 days was associated with a greater than sixfold increased risk (adjusted HR 6.99, 95% CI, 5.78-8.44). These findings persisted when analyses were restricted to individuals with regular cycles and certain LMP dates.

Increasing lag between ultrasound-based dating and LMP-based dating among asymptomatic patients was strongly associated with pregnancy loss risk. This clinically quantifiable measure can inform concern for pregnancy loss before symptom onset among individuals with a certain LMP.