Paediatric Index of Mortality 3 in Sweden: Validation and Recalibration to Improve Mortality Prediction in Paediatric Intensive Care

AIM: In Sweden, critically ill children are admitted to one of four dedicated paediatric intensive care units (PICUs) or general intensive care units (ICUs), reflecting a partly centralised system. The Paediatric Index of Mortality 3 (PIM3) is a widely used risk stratification tool for predicting mortality in critically ill children. This study aimed to validate PIM3 in a national Swedish cohort, recalibrate the model where indicated and compare outcomes between PICUs and ICUs. METHODS: This retrospective cohort study used prospectively collected data from the Swedish Intensive Care Registry of admissions reported between 1 January 2016 and 30 September 2022. Discrimination and calibration were assessed, and recalibration was performed using logistic regression. RESULTS: 19 183 admissions were included, 10 714 males and 8469 females. Prior to recalibration, PIM3 overestimated mortality in both settings, with standardised mortality rates (SMRs) of 0.72 (95% Confidence Interval (CI) 0.64-0.80) in PICUs and 0.52 (95% CI 0.42-0.63) in ICUs. Following recalibration, SMR was 0.92 (95% CI 0.70-1.20) and 0.96 (95% CI 0.59-1.49) in PICUs and ICUs, respectively. CONCLUSION: PIM3 demonstrated excellent discrimination but systematically overestimated mortality in the Swedish context. Recalibration corrected this bias and attenuated outcome differences between PICU and ICU settings.