# Comparative Evaluation of PRISM III, PIM 2, and PELOD Scores as Predictors of Outcome in Pediatric Intensive Care Unit Patients: An Experience From a Single Center in South Gujarat, India

## TL;DR

This study compares three scoring systems for predicting mortality in pediatric ICU patients in India, finding PRISM III to be the most accurate.

## Contribution

The study evaluates PRISM III, PIM 2, and PELOD scores in a single-center Indian PICU setting for mortality prediction.

## Key findings

- PRISM III showed the highest predictive accuracy with an AUC of 0.984.
- PELOD had the lowest predictive accuracy with an AUC of 0.872.
- All three scores showed good calibration, but PRISM III outperformed the others.

## Abstract

Aim: To assess and compare Pediatric Risk of Mortality III (PRISM III), Pediatric Index of Mortality 2 (PIM 2), and Pediatric Logistic Organ Dysfunction (PELOD) scores for predicting mortality in a pediatric intensive care unit (PICU) setting.

Design and setting: A cross-sectional, hospital-based, analytical observational study conducted in an eight-bed PICU.

Methods: PICU patients between the ages of 1 month and 17 years fulfilling the inclusion criteria were included in this study. PIM 2 score was calculated within the first hour of admission; PRISM III and PELOD scores were calculated at 24 hours. Outcomes of survival or non-survival were recorded. Scoring system performance was assessed using indices of discrimination and calibration. Predictive accuracy for mortality was evaluated with receiver operating characteristic (ROC) curve analysis, and calibration between predicted and observed mortality was examined using the Hosmer-Lemeshow goodness-of-fit test.

Results: Data from 138 patients were analyzed (M: F=0.8:1) with a mean age of 62.5 months and a median of 24 months. Respiratory ailments were the highest (46.4%), followed by central nervous system diseases (18.1%), and overall mortality was 20.3%. Area under the ROC curve was 0.984 for PRISM III, 0.836 for PIM 2, and 0.872 for PELOD. PRISM III was a better predictor of mortality risk, followed by PIM 2 and PELOD. The goodness-of-fit test showed good calibration for all three scores, but PRISM III had the better power compared to PIM 2 and PELOD.

Conclusion: In our PICU setting, the PRISM III score was a better predictor of mortality risk, and the performance of PELOD was the least.