The development and validation of a tool to predict spontaneous preterm birth in high risk women. (18th April 2012)

• Record Type: Journal Article
• Authors: Smout, E; Foster, C; Abbott, D; Van der Westhuizen, M; Seed, P
• Is Part Of: Archives of disease in childhood. Volume 97(2012)Supplement 1
• Page Start: A6
• Page End: A7
• Publication Date: 2012-04-18
• Subjects: Infants -- Diseases -- Periodicals; Newborn infants -- Diseases -- Periodicals; Fetus -- Diseases -- Periodicals; 618.920105
• DOI: 10.1136/fetalneonatal-2012-301809.18
• Languages: English
• ISSNs: 1359-2998

Abstract

Abstract : Preterm birth (PTB) rates are rising, but models to predict spontaneous PTB are unreliable. FFN and cervical length are the best predictors of PTB but have never been combined in a prediction model with other clinical factors. Data from 414 women attending the Preterm Surveillance Clinic at St. Thomas' Hospital, London was analysed. These women were at high risk of PTB following previous PTB or cervical surgery. A parametric survival model for spontaneous PTB <37 weeks' was developed using both clinical and biometric data from the first 207 women. This model was used to calculate an individual percentage risk of delivering <37, <34 and <30 weeks' gestation, and within 7 and 14 days of testing. This model was then validated on a separate cohort of the remaining 207 women. The rate of spontaneous PTB was 15% (60/414) – 16% (31/207) in the original dataset and 15% (29/207) in the validation set. fFN result, cervical length and gestation of fFN test were the only significant predictors of PTB when combined. Using the model, the ROC AUC for predicting delivery <30, <34 and <37 weeks' gestation were similar for the original (0.96, 0.80, 0.75) and validation (0.90, 0.90, 0.80) data sets. The proportion of observed deliveries <34 weeks' gestation in the validation set (7.2%) was similar to that calculated by the original model (9.8%), p=0.22. Accurate PTB risk in high risk women can be calculated using a model combining FFN and cervical length. These variables supersede demographic information and past obstetric history.