Predictive accuracy of Southwest Thames Obstetric Research Collaborative (STORK) chorionicity-specific twin growth charts for stillbirth: a validation study

Objective: Twin pregnancy is associated with a 2-3-fold increased risk of stillbirth compared with singleton pregnancy. Despite the fact that the growth pattern in twins has been shown to be different from that in singletons, it is controversial whether twin-specific growth charts should be used routinely. A major goal of prenatal ultrasound is to identify fetuses with growth restriction at risk of stillbirth. The main aim of this study was to compare the performance of chorionicity-specific twin charts with singleton charts, both customized and non-customized, in the antenatal prediction of small-for-gestational-age (SGA) stillborn and liveborn fetuses.

Methods: This was a multicenter cohort study analyzing data from the Southwest Thames Obstetric Research Collaborative (STORK) multiple pregnancy cohort (2000-2009) and a second cohort of twin pregnancies at St George's University Hospital (SGH) (2011-2016). The STORK cohort was used to compare the performance of the twin charts and non-customized singleton charts of Poon et al. and the SGH cohort was used to validate the twin-specific charts and compare their performance against customized (Gestation Related Optimal Weight (GROW)) and non-customized (Poon) singleton charts. The primary outcome was the prediction of SGA cases that were stillborn and those that were liveborn in twin pregnancies. Estimated fetal weight (EFW) available from the last examination (24 weeks' gestation and onwards) before delivery or demise was used to classify the fetuses as SGA (EFW < 10^th centile or < 3^rd centile) or appropriate for gestational age. The proportions of predicted SGA stillbirths and SGA live births were calculated using the three different charts.

Results: The STORK cohort consisted of 1850 dichorionic (DC) and 300 monochorionic (MC) twin pregnancies. The SGH cohort consisted of 579 DC and 180 MC twin pregnancies. The stillbirth rates in the STORK and SGH cohorts were 1.1% and 1.3%, respectively. In those liveborn in the STORK cohort, using a 10^th -centile cut-off to define SGA, the non-customized singleton chart classified a significantly greater proportion as SGA than did the twin chart, regardless of chorionicity (P < 0.001). However, there was no significant difference between the twin and the non-customized singleton charts with regard to the proportion of stillbirth cases that were classified as SGA (P = 0.479). In the SGH cohort, the non-customized singleton chart classified 8.5% of all liveborn fetuses as SGA (EFW < 10^th centile) compared with 12.8% using the customized singleton chart and 7.1% using the twin chart (P < 0.001 and P = 0.005, respectively). However, there was no significant difference among the three charts in the proportion of stillbirths classified as SGA, regardless of chorionicity (P = 0.999). Similar results were obtained when the third centile cut-off was used to define SGA.

Conclusions: Compared with the STORK chorionicity-specific twin charts, the customized and non-customized singleton charts classified prenatally as SGA more liveborn fetuses. However, the three charts classified as SGA a similar proportion of stillborn cases. Our preliminary results suggest that these twin charts could safely reduce unnecessary medical intervention in twin pregnancies. Further research on the topic is needed before clinical recommendations can be made. Copyright © 2018 ISUOG. Published by John Wiley & Sons Ltd.