Performance of pediatric index of mortality-2, pediatric index of mortality-3 and pediatric risk of mortality IV in an Egyptian pediatric intensive care unit.

Mohamed, Montaser Mohamed; Abdelatif, Rania Gamal; Bakheet, Mohamed Abdel Aal Mohamed; Mahmoud, Ramadan Aboelhassan Ahmed

doi:10.21608/smj.2023.206040.1383

Performance of pediatric index of mortality-2, pediatric index of mortality-3 and pediatric risk of mortality IV in an Egyptian pediatric intensive care unit.

Document Type : Original Article

Authors

¹ pediatric,medicine,sohag university,egypt

² Pediatric department, Faculty of Medicine, Sohag University, Egypt

³ pediatric Department, Faculty of Medicine, Sohag university, Sohag, Egypt.

⁴ pediatric department, Faculty of medicine, sohag university, Sohag, Egypt

10.21608/smj.2023.206040.1383

Abstract

Introduction: Severity of illness (SOI) scores have been developed to predict the outcome and to provide a better quality of care with available resources.

Objective of the study: was to evaluate the predictive ability of Pediatric index of mortality-2 (PIM-2), pediatric index of mortality-3(PIM-3) and Pediatric risk of mortality IV (PRISM IV) in a resource-limited PICU.

Materials and methods: We conducted a prospective cohort study in PICU in Sohag University Hospital in the period from March 2018 to June 2020. We recorded the baseline patient characteristics, admission diagnoses, variables of PIM-2, PIM-3 and PRISM IV models and outcomes of children admitted to the PICU. We utilized Area Under Receiver Operating Characteristics (AU-ROC) curves and Goodness-of-fit (GOF) test to evaluate the discrimination and calibration of the three models.

Results: Of 451 patients enrolled, 171 (37.9%) died. Sepsis was the major admission diagnosis. The discrimination was acceptable for PRISM IV, while it was poor for both PIM-2 and PIM-3 as indicated by the AU-ROC which was (0.74; 95%CI: 0.62- 0.86 for PRISM IV) vs (0.69; 95% CI: 0.58- 0.81 for PIM-2) and (0.69; 95% CI: 0.57-0.81for PIM-3) (p < 0.0001). The calibration was poor for all scores as the p-value of GOF test for was < 0.0001 for all scores.

Conclusion: The discrimination of PRISM IV was acceptable and the best among the three models. All scores had poor calibration and under-predict mortality in our setting. We suggest utilizing them as quality indicators rather than in mortality prediction.

Keywords

Main Subjects