Accuracy of second and third trimester ultrasound in the determination of chorionicity in twin pregnancies: a cross-sectional observational study.

Background: Determination of chorionicity in twin pregnancies is essential for optimizing management. In South Africa, many women first access antenatal care at late gestations. The accuracy of ultrasound in the determination of chorionicity up to 14 weeks' gestation is well documented. However, there is little data on determination of chorionicity in late pregnancy. We thus aimed to determine the accuracy of five ultrasound parameters performed after 14 weeks' gestation in determining chorionicity.

Method: This was a cross-sectional diagnostic validity study where postpartum placental histology was accepted as the standard test for chorionicity. All twin pregnancies > 14 weeks' gestation in whom chorionicity had not been determined and without any complications were included. Data collected included demographic data, ultrasound-determined number of placentae, fetal sexes, membrane 'take-off' (lambda or T-Sign), thickness, and number of layers in the dividing membrane. Placentae and membranes were examined by the Anatomical Pathology department. Each of the 5 ultrasound parameters for chorionicity was analyzed separately in relation to the corresponding histology result.

Results: A total of 85 patients were included, for whom histology was available in 58. Of these, 12 were monochorionic diamniotic and 44 dichorionic diamniotic. The diagnostic accuracies of discordant gender, number of placentas, lambda sign, membrane thickness and number of layers in the membranes were 44.1, 49.1, 76.7, 76.1 and 82.9, respectively. We developed a step-wise algorhithm for the determination of chorionicity.

Conclusion: Determination of chorionicity late in twin pregnancies remains inaccurate. However, a step-wise algorithm for the application of 5 parameters is helpful.