An investigation for sonographic prediction of twin weight discordance

Abstract：Objective To evaluate the effectiveness of sonographic prediction of birth weight discordance in twin pregnancies, and to ivestigate the best predictive sonographic variables. Methods Using intertwin birth weight difference ≥ 25% as weight discordance, this retrospective study recruited 399 weight concordant twins and 46 weight discordant twins with complete clinical data and 1658 sonographic examinations. Correlation tests were applied to evaluate the relation of single sonographic variables and intertwin birth weight difference. The ROC was used to compare the sensitivity, specificity, and predictive values of single variables and intertwin difference of estimated fetal weight (△EFW) calculated by Hadlock equation to predict weight discordance. Results (1) After 20 gestational age, the areas under the ROC curves of all sonographic variables had statistically significance (P < 0.05) , and intertwin △ EFW had the biggest areas under the ROC curves in the whole gestation. (2) As a single variable, AC-r had better predictability for twin weight discordance than HC-r or FL-r. (3) After decreasing cut-off point of △EFW from ≥25% to ≥20% , the sensitivities were increased with 10% ～25% in the most gestational ages, and the specificities were always maintained more than 90%. Conclusion Sonographic prediction of birth weight discordance is feasible and effective in twins after 20 gestational age. AEFW is the best variable to predict twin weight discordance, if △EFW ≤20% is used as the cut-off point. When u-sing single predictive sonographic variable, AC-r with different cut-off points corresponding to different gestational weeks is suggested.