First-trimester growth discordance and adverse pregnancy outcome in dichorionic twins

Objective: To evaluate the association between first-trimester growth discordance and adverse pregnancy outcome in dichorionic twin pregnancies.

Methods: This was a retrospective cohort study of consecutive women with dichorionic twin pregnancies undergoing an ultrasound scan at our institution between 7 and 14 weeks' gestation. Study groups were defined by the presence or absence of ≥ 11% crown-rump length (CRL) discordance. Pregnancies were excluded if one twin was dead on initial ultrasound or if a termination was performed. The primary outcome was loss of one or both fetuses before 20 weeks. Secondary outcomes included fetal anomaly, fetal demise after 20 weeks (stillbirth), small-for-gestational-age (SGA) at birth, admission to the neonatal intensive care unit (NICU) and preterm delivery before 34 weeks.

Results: Of 805 dichorionic twin pregnancies undergoing first-trimester ultrasound, 610 met the inclusion criteria. Eighty-six had ≥ 11% CRL discordance and, of these, nine (10.5%) had a fetal loss at < 20 weeks (risk ratio (RR) 7.8 (95% CI, 3.0-20.5)). In the surviving pregnancies, an increased risk of fetal anomalies was seen (27.3 vs 17.4%, RR 1.6 (95% CI, 1.1-2.4)). In surviving pregnancies unaffected by anomalies, no increased risk of stillbirth, SGA, NICU admission or delivery before 34 weeks was noted in the discordant group. A post-hoc power analysis demonstrated 80% power to detect a five-fold increase in the risk of stillbirth and 90% power to detect a two-fold increase in other outcomes.

Conclusion: Dichorionic pregnancies in which a CRL discordance of at least 11% is noted are at increased risk for fetal anomalies and fetal loss prior to 20 weeks' gestation. However, patients can be reassured that, in the absence of structural anomalies, CRL discordance does not appear to be associated with other adverse outcomes in continuing pregnancies, although the power to detect small increases in the risk of stillbirth may have been limited by the sample size.