Comparison of Pediatric Risk of Mortality III, Pediatric Index of Mortality 2, and Pediatric Index of Mortality 3 in Predicting Mortality in a Pediatric Intensive Care Unit

Abstract

Aims To compare and validate the Pediatric Risk of Mortality (PRISM) III, Pediatric Index of Mortality (PIM) 2, and PIM 3 scores in a tertiary care pediatric intensive care unit (PICU) (Indian setting). Materials and Methods All consecutively admitted patients in the PICU of a public hospital (excluding those with unstable vital signs or cardiopulmonary resuscitation within 2 hours of admission, cardiopulmonary resuscitation before admission, and discharge or death in less than 24 hours after admission) were included. PRISM III, PIM 2, and PIM 3 scores were calculated. Mortality discrimination for the three scores was calculated using the receiver operating characteristic (ROC) curve, and calibration was performed using the Hosmer–Lemeshow goodness-of-fit test. Results A total of 350 patients were included (male:female = 1.3:1) over the study duration of 18 months (median age: 12 months [interquartile range: 4–60 months]). Nearly half were infants (47.4%). Patients with central nervous system disease were the highest (22.8%) followed by cardiovascular system (20.6%). Mortality rate was 39.4% (138 deaths). The area under the ROC curve for the PRISM III score was 0.667, and goodness-of-fit test showed no significant difference between the observed and expected mortalities in any of these categories (p > 0.5), showing good calibration. Areas under the ROC curve for the PIM 2 and PIM 3 scores were 0.728 and 0.726, respectively. For both the scores, the goodness-of-fit test showed good calibration. Conclusions Although all the three scores demonstrate good calibration, the PIM 2 and PIM 3 scores have an advantage regarding the better discrimination ability, ease of data collection, simplicity of computation, and inherent capacity of not being affected by treatment in PICU.