Development and validation of nomograms for predicting preterm delivery.

Abstract

The objective of the study was to develop a statistical model for predicting risk of preterm delivery after in utero transfer for threatened preterm delivery in tertiary care centers. This study was an observational study including a total of 906 patients transferred for threatened preterm delivery at Paule-de-Viguier and Croix-Rousse University Hospitals. Clinical and sonographic data from 1 series were used to construct logistic regression models for predicting preterm delivery and were validated on an independent series. An Internet-based tool was developed to facilitate the use of the nomograms. Based on multivariate analyses, 2 nomograms were built: 1 to predict delivery within 48 hours after transfer and 1 to predict delivery before 32 weeks. Discrimination and calibration of the predictive models were good when applied to the validation set (concordance index 0.73 and 0.72, respectively). We developed and validated nomograms to predict the individual probability of preterm birth after transfer for threatened preterm delivery.