ABSTRACT
OBJECTIVE: To develop a model for the prediction of adverse perinatal outcome in growth-restricted fetuses requiring delivery before 28 weeks in order to provide individualized patient counseling. METHODS: This was a retrospective multicenter cohort study of singleton pregnancies with antenatal suspicion of fetal growth restriction requiring delivery before 28 weeks' gestation between January 2010 and January 2020 in six tertiary public hospitals in the Barcelona area, Spain. Separate predictive models for mortality only and mortality or severe neurological morbidity were created using logistic regression from variables available antenatally. For each model, predictive performance was evaluated using receiver-operating-characteristics (ROC)-curve analysis. Predictive models were validated externally in an additional cohort of growth-restricted fetuses from another public tertiary hospital with the same inclusion and exclusion criteria. RESULTS: A total of 110 cases were included. The neonatal mortality rate was 37.3% and, among the survivors, the rate of severe neurological morbidity was 21.7%. The following factors were retained in the multivariate analysis as significant predictors of mortality: magnesium sulfate neuroprotection, gestational age at birth, estimated fetal weight, male sex and Doppler stage. This model had a significantly higher area under the ROC curve (AUC) compared with a model including only gestational age at birth (0.810 (95% CI, 0.730-0.889) vs 0.695 (95% CI, 0.594-0.795); P = 0.016). At a 20% false-positive rate, the model showed a sensitivity, negative predictive value and positive predictive value of 66%, 80% and 66%, respectively. For the prediction of the composite adverse outcome (mortality or severe neurological morbidity), the model included: gestational age at birth, male sex and Doppler stage. This model had a significantly higher AUC compared with a model including only gestational age at birth (0.810 (95% CI, 0.731-0.892) vs 0.689 (95% CI, 0.588-0.799); P = 0.017). At a 20% false-positive rate, the model showed a sensitivity, negative predictive value and positive predictive value of 55%, 63% and 74%, respectively. External validation of both models yielded similar AUCs that did not differ significantly from those obtained in the original sample. CONCLUSIONS: Estimated fetal weight, fetal sex and Doppler stage can be combined with gestational age to improve the prediction of death or severe neurological sequelae in growth-restricted fetuses requiring delivery before 28 weeks. This approach may be useful for parental counseling and decision-making. © 2023 International Society of Ultrasound in Obstetrics and Gynecology.
