Fetal Medicine Foundation competing-risks model for small-for-gestational-age neonate in midgestation: external validation in nulliparous patients.

Abstract

The aim of this study was to examine the performance of the Fetal Medicine Foundation (FMF) competing-risks model for the prediction of a small-for-gestational-age (SGA) neonate in a large, diverse, nulliparous cohort. This was a retrospective single-center cohort study of nulliparous singleton pregnancies undergoing an ultrasound examination between 19 + 0 and 24 + 0 weeks' gestation. We calculated the risk of SGA at various birth-weight thresholds using the FMF competing-risks model, which utilizes maternal factors with or without ultrasound-derived estimated fetal weight (EFW). The outcomes predicted by the model were compared with the actual outcomes of the included pregnancies using the area under the receiver-operating-characteristics curve (AUC) and detection rates at fixed false-positive rates. These findings were compared with those of the original FMF model-development study. In our cohort of 8974 nulliparous patients, the FMF competing-risks model had a higher predictive performance for SGA < 3^rd percentile compared with SGA < 10^th percentile. The model also performed better in the prediction of a SGA neonate requiring preterm delivery compared with a SGA neonate delivered at term. Utilization of both maternal factors and midgestation EFW improved the predictive performance of the model compared with using maternal factors alone. The AUC of the model including maternal factors and midgestation EFW was 0.740 for preterm SGA < 10^th percentile, 0.766 for preterm SGA < 3^rd percentile, 0.719 for term SGA < 10^th percentile and 0.734 for term SGA < 3^rd percentile. These values were significantly lower than those seen in the FMF model-development cohort, which were 0.781 (P = 0.01), 0.809 (P = 0.04), 0.766 (P < 0.01) and 0.790 (P < 0.01), respectively, but similar to the AUCs seen in a subgroup of nulliparous patients in the FMF study. The FMF competing-risks model for SGA performs well in a large independent cohort of nulliparous patients but appears to have better predictive performance in a population that includes parous patients. © 2025 International Society of Ultrasound in Obstetrics and Gynecology.