Decoding Mortality: How Well Do Pediatric Scoring Systems Work?
"A deep dive into the Paediatric Index of Mortality (PIM) scores and their effectiveness in assessing risk in intensive care units."
In pediatric intensive care units (PICUs), accurately assessing a child's risk of mortality is crucial for guiding treatment decisions and evaluating the quality of care. Various mortality risk assessment scores have been developed to aid in this process, with the Paediatric Index of Mortality (PIM) and its updated version, PIM 2, being among the most widely used.
These scores, initially created and validated in specific populations and timeframes, need to be carefully evaluated before being applied in different environments. A recent study examined the utility of PIM and PIM 2 in the Red Cross War Memorial Children's Hospital (RCWMCH) in South Africa, the only children's hospital in sub-Saharan Africa providing exclusive tertiary and quaternary services for children.
This article delves into the findings of the study, comparing the performance of PIM and PIM 2 in the RCWMCH PICU with their performance in the units where they were initially derived. By exploring the scores' discrimination and calibration, this analysis sheds light on their effectiveness as mortality risk assessment tools in a unique healthcare setting.
Common Pediatric Mortality Scores
The Pediatric Risk of Mortality (PRISM) and Pediatric Index of Mortality (PIM) are described as two commonly used mortality-risk scoring systems for children. One study set out to assess how available PIM scores discriminate and calibrate mortality risk among children admitted to a pediatric intensive care unit in Bogotá. The provided source material does not report a mortality rate or comparative performance results for that Bogotá study.
How Accurate Are PIM and PIM 2 Scores?
The study retrospectively audited case records and prospectively collected patient data from all admissions to the RCWMCH PICU during 2000 (for PIM) and 2006 (for PIM 2), excluding premature infants, children who died within 2 hours of admission, and those transferred to other PICUs. Researchers then analyzed this data to assess how well the PIM and PIM 2 scores predicted mortality in this specific population.
- Discrimination: Both PIM and PIM 2 demonstrated good discrimination, with AUC values of 0.849 and 0.841, respectively. This indicates that both scores are reasonably effective at distinguishing between patients who will survive and those who will not.
- Calibration: PIM showed poor calibration (x²=19.74; p=0.02), while PIM 2 demonstrated acceptable calibration (x²=10.06; p=0.35). This suggests that PIM 2's predicted mortality rates were more closely aligned with the actual observed mortality rates in the RCWMCH PICU than PIM's predictions.
- Standardized Mortality Risk Ratios (SMRs): The SMRs for age and diagnostic subgroups for both scores fell within wide confidence intervals. This makes it difficult to draw definitive conclusions about the scores' performance within specific patient subgroups.
Evidence Across Settings and Outcomes
A systematic review examined risk-prediction models for deterioration or mortality among children aged 0–18 in low- and middle-income countries, covering emergency departments, pediatric wards, and PICUs. Another systematic review reports that children faced increased risk of death for up to 10 years after intensive-care admission, alongside greater physical and psychosocial morbidity and poorer quality of life than healthy controls. In a prospective observational cohort at the University of Gondar hospital in northwest Ethiopia, higher PIM2 scores were among the independent predictors of PICU mortality, along with weekend admission, mechanical ventilation, and critical-illness diagnoses.
Calibration Varies by Risk Level
A multicenter validation reported that PIM2 and PIM3 calibration varied across predicted-risk ranges. Both scores overestimated mortality in the 0%–3% risk range; PIM3 tended to underestimate mortality among patients in the 9%–46% range, while PIM2 overestimated mortality in the 70%–100% range. These findings show that the models’ estimates did not align uniformly with observed mortality across risk strata in that validation.
Implications for Pediatric Intensive Care
The study highlights the importance of evaluating mortality risk assessment models in different healthcare settings before relying on them for clinical decision-making. While PIM and PIM 2 have been validated in various populations, their performance can vary depending on the specific characteristics of the patient population and the healthcare environment.
Context-Sensitive Validation
A 2021 meta-analysis source lists an assessment of PIM and PRISM III for predicting mortality in a pediatric intensive care unit in Hong Kong. A separate article published October 16, 2025, reports that PIM3 may provide a snapshot of mortality risk from clinical indicators, while discrepancies highlight differing clinical realities across healthcare settings. Its researchers called for a more nuanced understanding of predictor variables in pediatric care.
Comparing Models and Their Evaluation
A 2022 study aimed to compare the prognostic performance of PIM2, PIM3, PRISM, and PRISM4 to determine which score was most applicable to its pediatric study cohort. A separate report describes assessing PIM3 performance through calibration and discrimination, using the standardized mortality ratio and Pearson’s χ² goodness-of-fit test for calibration. The supplied source material does not provide a result establishing one score as best across settings.
Scores and Clinical Decision-Making
A March 2024 single-center study reports a correlation between higher PIM3 scores and increased mortality risk in critically ill pediatric patients. Its conclusion says the scores guided clinical decision-making. The supplied snippet does not specify how decisions changed or report patient outcomes attributable to using the score.
The finding that PIM 2 demonstrated better calibration than PIM in the RCWMCH PICU suggests that it may be a more appropriate tool for assessing mortality risk in this specific setting. However, the wide confidence intervals for SMRs in age and diagnostic subgroups indicate the need for further research to evaluate the scores' performance in specific patient populations.
Ultimately, mortality risk assessment models like PIM and PIM 2 can be valuable tools for benchmarking quality of care and identifying areas for improvement in pediatric intensive care units. By carefully evaluating and calibrating these models, healthcare professionals can ensure that they are providing the best possible care for critically ill children.