Objective: To evaluate the performance of the newest version of the Pediatric Index of Mortality 3 score and compare it with the Pediatric Index of Mortality 2 in a multicenter national cohort of children admitted to PICU.
Design: Retrospective, prospective cohort study.
Setting: Seventeen Italian PICUs.
Patients: All children 0 to 15 years old admitted in PICU from January 2010 to October 2014.
Interventions: None.
Measurement and main results: Eleven thousand one hundred nine children were enrolled in the study. The mean Pediatric Index of Mortality 2 and 3 values of 4.9 and 3.9, respectively, differed significantly (p < 0.05). Overall mortality rate was 3.9%, and the standardized mortality ratio was 0.80 for Pediatric Index of Mortality 2 and 0.98 for Pediatric Index of Mortality 3 (p < 0.05). The area under the curve of the receiver operating characteristic curves was similar for Pediatric Index of Mortality 2 and Pediatric Index of Mortality 3. The Hosmer-Lemeshow test was not significant for Pediatric Index of Mortality 3 (p = 0.21) but was highly significant for Pediatric Index of Mortality 2 (p < 0.001), which overestimated death mainly in high-risk categories.
Conclusions: Mortality indices require validation in each country where it is used. The new Pediatric Index of Mortality 3 score performed well in an Italian population. Both calibration and discrimination were appropriate, and the score more accurately predicted the mortality risk than Pediatric Index of Mortality 2.