Prediction Model Assessing Risk of Delivery after Diagnosis of Abnormal Umbilical Artery Doppler.

OBJECTIVE: Umbilical artery Doppler (UAD) velocimetry abnormalities are associated with increased neonatal morbidity and mortality. Currently, there are no risk stratification methods to assist in antepartum management such as timing of antenatal corticosteroids (ACS). Therefore, we sought to develop a model to predict risk of delivery within 7 days following diagnosis of abnormal UAD velocimetry in patients with fetal growth restriction (FGR). STUDY DESIGN: Retrospective single referral center study of liveborn singleton pregnancies complicated by FGR and ≥1 abnormal UAD velocimetry value (≥95th percentile for gestational age [GA]). We considered 17 variables and used backward stepwise logistic regression to create a multivariable model for the prediction of delivery within 7 days. We assessed model fit with calibration, discrimination, likelihood ratios, and area under the curve. Internal validation of the model was assessed by using the bootstrap method. RESULTS: Between 2008 and 2015, a total of 176 patients were eligible and included for model development. Median (range) GA at initial eligibility was 32.1 weeks (28.1-36.1 weeks) and from initial eligibility until delivery was 21 days (0-104 days). Fifty-two patients (30%) were delivered in the 7 days following inclusion. GA at first abnormal UAD, severity of first abnormal UAD, oligohydramnios, preeclampsia, and pre-pregnancy BMI were included in the model. The model had an area under the ROC curve of 0.94 (95% confidence interval [CI]: 0.90-0.98), sensitivity of 85%, and specificity of 91%. If the model alone were used for ACS timing, 85% of the cohort who delivered in the following week would have received ACS, and ACS would not have been given to 91% who delivered later. Internal validation yielded similar results with a mean area under the curve (95% CI) of 0.94 (0.88-0.98). CONCLUSION: If validated externally, our model can be used to predict risk of delivery in patients with FGR and abnormal UAD velocimetry, potentially improving timing of ACS. KEY POINTS: · Risk of delivery in seven days can be predicted.. · Risk of delivery can inform corticosteroid timing.. · External validation can further develop a clinical aid..