Objective: To compare the performance of the Pediatric Index of Mortality (PIM), PIM2, the Pediatric Risk of Mortality (PRISM), and PRISM III in Australia and New Zealand.
Design: A two-phase prospective observational study. Phase 1 assessed the performance of PIM, PRISM, and PRISM III between 1997 and 1999. Phase 2 assessed PIM2 in 2000 and 2001.
Setting: Ten intensive care units in Australia and New Zealand.
Patients: Included in the study were 26,966 patients aged <16 yrs; 1,147 patients died in the intensive care unit.
Measurements and main results: Discrimination between death and survival was assessed by calculating the area under the receiver operating characteristic plot for each model. The areas (95% confidence interval) for PIM, PIM2, PRISM, and PRISM III were 0.89 (0.88-0.90), 0.90 (0.88-0.91), 0.90 (0.89-0.91), and 0.93 (0.92-0.94). The calibration of the models was assessed by comparing the number of observed to predicted deaths in different diagnostic and risk groups. Prediction was best using PIM2 with no difference between observed and expected mortality (standardized mortality ratio [95% confidence interval] 0.97 [0.86-1.05]). PIM, PRISM III, and PRISM all overpredicted death, predicting 116%, 130%, and 189% of observed deaths, respectively. The performance of individual units was compared during phase 1, using PIM, PRISM, and PRISM III. There was agreement between the models in the identification of outlying units; two units performed better than expected and one unit worse than expected for each model.
Conclusions: Of the models tested, PIM2 was the most accurate and had the best fit in different diagnostic and risk groups; therefore, it is the most suitable mortality prediction model to use for monitoring the quality of pediatric intensive care in Australia and New Zealand. More information about the performance of the models in other regions is required before these results can be generalized.