Ultrasound estimation of birthweight in twin pregnancy: comparison of biometry algorithms in the STORK multiple pregnancy cohort.

Abstract

OBJECTIVES: The aims of this study were first, to ascertain the accuracy of formulae for ultrasonographic birth-weight estimation in twin compared with singleton pregnancies and second, to assess the accuracy of sonographic examination in the prediction of birth-weight discordance in twinpregnancies. METHODS: This was a retrospective cohort study including both singleton and twin pregnancies. Routine biometry was recorded and estimated fetalweight (EFW) calculated using 33 different formulae. Only pregnancies that delivered within 48 h of the ultrasound scan were included (4280 singleton and 586 twin fetuses). Differences between the EFW and actual birth weight (ABW) were assessed by percentage error, accuracy in predictions within ± 10 % and ± 15% of error and use of the Bland-Altman method. The accuracy of prediction of the different cut-offs of birth-weight discordance intwin pregnancies was also assessed using the area under the receiver-operating characteristics curve (AUC). RESULTS: The overall mean absolute percentage error was ≤ 10 % for 25 formulae in singleton pregnancies compared with three formulae in twinpregnancies. The overall predictions within ± 10% and ± 15% of the ABW were 62.2% and 81.5% in singleton and 49.7% and 68.5% in twinpregnancies, respectively. When the formulae were categorized according to the biometric parameters included, those based on a combination of head, abdomen and femur measurements showed the lowest mean absolute percentage error, in both singleton and twin pregnancies. The predictive accuracy for 25% birth-weight discordance using the Hadlock 2 formula, as assessed by the AUC, was 0.87. CONCLUSIONS: Ultrasound estimation of birth weight is less accurate in twin than in singleton pregnancies. Formulae that include a combination of head, abdomen and femur measurements perform best in both singleton and twin pregnancies.