Pediatric Index of Cardiac Surgical Intensive Care Mortality Risk Score for Pediatric Cardiac Critical Care.

Objective: Comparison of clinical outcomes is imperative in the evaluation of healthcare quality. Risk adjustment for children undergoing cardiac surgery poses unique challenges, due to its distinct nature. We developed a risk-adjustment tool specifically focused on critical care mortality for the pediatric cardiac surgical population: the Pediatric Index of Cardiac Surgical Intensive care Mortality score.

Design: Retrospective analysis of prospectively collected pediatric critical care data.

Setting: Pediatric critical care units in the United States.

Patients: Pediatric cardiac intensive care surgical patients.

Interventions: Prospectively collected data from consecutive patients admitted to ICUs were obtained from The Virtual PICU System (VPS, LLC, Los Angeles, CA), a national pediatric critical care database. Thirty-two candidate physiologic, demographic, and diagnostic variables were analyzed for inclusion in the development of the Pediatric Index of Cardiac Surgical Intensive care Mortality model. Multivariate logistic regression with stepwise selection was used to develop the model.

Measurements And Main Results: A total of 16,574 cardiac surgical patients from the 55 PICUs across the United States were included in the analysis. Thirteen variables remained in the final model, including the validated Society of Thoracic Surgeons-European Association of Cardio-Thoracic Surgery Congenital Heart Surgery Mortality (STAT) score and admission time with respect to cardiac surgery, which identifies whether the patient underwent the index surgical procedure before or after admission to the ICU. Pediatric Index of Cardiac Surgical Intensive Care Mortality (PICSIM) performance was compared with the performance of Pediatric Risk of Mortality-3 and Pediatric Index of Mortality-2 risk of mortality scores, as well as the STAT score and STAT categories by calculating the area under the curve of the receiver operating characteristic from a validation dataset: PICSIM (area under the curve = 0.87) performed better than Pediatric Index of Mortality-2 (area under the curve = 0.81), Pediatric Risk of Mortality-3 (area under the curve = 0.82), STAT score (area under the curve = 0.77), STAT category (area under the curve = 0.75), and Risk Adjustment for Congenital Heart Surgery-1 (area under the curve = 0.74).

Conclusions: This newly developed mortality score, PICSIM, consisting of 13 risk variables encompassing physiology, cardiovascular condition, and time of admission to the ICU showed better discrimination than Pediatric Index of Mortality-2, Pediatric Risk of Mortality-3, and STAT score and category for mortality in a multisite cohort of pediatric cardiac surgical patients. The introduction of the variable "admission time with respect to cardiac surgery" allowed prediction of mortality when patients are admitted to the ICU either before or after the index surgical procedure.