External Validation of a Risk Score Assessment for Pediatric Ventricular Assist Device Mortality

The ACTION score was internally validated to predict mortality after pediatric ventricular assist device (VAD) placement. This study aims to externally validate the ACTION score and compare its performance to the INTERMACS profile for prediction of mortality in pediatric VAD candidates. The publicly available Pedimacs dataset (531 patients) was used for external validation. The Pedimacs dataset as well as a contemporary ACTION dataset (1,304 patients) were used to compare the performance of the ACTION score and INTERMACS profile. A statistically significant linear trend in the proportion of death across increasing ACTION scores (p < 0.0001) was observed in the Pedimacs dataset, with a receiver operating characteristic (ROC) curve demonstrating an area under the curve (AUC) of 0.71. Both the ACTION score and the INTERMACS profile demonstrated a significant linear trend in the proportion of death across scores in the ACTION dataset. Receiver operating characteristic analysis demonstrated an AUC of 0.71 and 0.58 for the ACTION score and INTERMACS profile, respectively. The ACTION score is externally validated with the Pedimacs dataset and is a fair predictor of mortality. The ACTION score performed better than the INTERMACS profile in predicting mortality using both historical Pedimacs data and contemporary ACTION data.