Comparison of PRISM IV and PIM III prognostic scores as mortality indicators among paediatric intensive care unit patients

Abstract

Introduction: Identifying prognosis of patients admitted in the paediatric intensive care unit (PICU) is of paramount importance to better allocate medical resources, reduce patient-doctor conflict and for overall better patient care. Many scoring systems have been formulated to accomplish this task including paediatric risk of mortality (PRISM) IV and paediatric index of mortality (PIM) III.Objectives: To compare PRISM IV and PIM III as prognostic scoring tools in the PICU of a tertiary healthcare centreMethod: This prospective observational study was conducted in the PICU, Dhiraj hospital, SBKS MI&RC, Gujarat, India, from March 2021 to September 2022. It included all patients from age 1 month to 18 years who were admitted in the PICU for >24 hours and gave consent for this study. They were thoroughly examined and investigations were done within the first hour of admission as per standard guidelines and their PRISM IV and PIM III scores were calculated and data were analysed.Results: A total of 74 patients was enrolled in this study. Whilst 43 (58.1%) patients were discharged, 18 (24.3%)] were discharged against medical advice(DAMA) moribund, 4 were DAMA (non-moribund) and 9 (12.2%) died. Receiver Operating Characteristic (ROC) curves were made for both PRISM IV and PIM III scores. Discriminatory powers of PIM III (AUC 0.725; 95% CI: 0.609 to 0.823) and PRISM IV (AUC 0.8; 95% CI: 0.691 to 0.884) were acceptable. Whilst PRISM IV was the better predictor of mortality at cut off point of >4 with area under curve of 0.8 for correctly predicting mortality, this difference was not statistically significant (p value=0.354). Multivariate logistic regression analysis showed that the duration of stay(days) and PRISM IV value >4 were significant independent risk factors of mortality after adjusting for confounding factors.Conclusions: Both PRISM IV and PIM III were good prognostic scoring tools in the PICU of the tertiary healthcare centre. Multivariate logistic regression analysis showed that the duration of stay(days) and PRISM IV value >4 were significant independent risk factors of mortality after adjusting for confounding factors.