Generation of PLASE score for patent ductus arteriosus using the PLASE study database.

Abstract

No echocardiographic model, to the best of our knowledge, has been established to predict the future need for patent ductus arteriosus (PDA) surgery. This study aimed to develop a novel predictive score (PLASE score) for anticipating the need for PDA surgery using the PLASE study database. The included infants with gestational age (GA) < 30 weeks were allocated to derivation and validation groups (2:1). Logistic regression models were constructed to predict the future need for PDA surgery utilizing three clinical and three echocardiographic indices measured at 3 days of age as candidate variables. ROC-AUCs and 95% confidence intervals (CIs) were obtained by 3-fold cross-validation and the percentile method, respectively. The model with the largest ROC-AUC was tested in the validation data. Derivation and validation data included 463 and 229 patients, respectively, with 55 and 22 surgical cases, respectively. The ROC-AUC was maximized in the model using GA and all three echocardiographic indices (0.846 [95% CI, 0.805-0.886]). In the validation data, the ROC-AUC for the same model was 0.827 (0.744-0.911). We created a surgical prediction model using simple indices at 3 days of age, and the validation data demonstrated good predictive ability. No early predictive model has been established for the future need of patent ductus arteriosus (PDA) surgery in preterm infants. A new prediction model was created with the Patent ductus arteriosus and Left Atrial Size Evaluation study in preterm infants (PLASE) database (N = 692), incorporating gestational age and three simple echocardiographic indices measured at 3 days of age. The model demonstrates high discrimination and calibration. This model provides risk stratification for preterm PDA and may contribute to early preterm management.