Pediatric risk and index of mortality in an intensive care unit

Aim: The purpose of the study is to determine the discriminative ability and calibration of Pediatric Risk of Mortality (PRISM I) I and Pediatric Index of Mortality (PIM) II in predicting the mortality in children admitted to a medical-surgical pediatric intensive care unit (PICU) in Turkey. Material and Method: A total of 277 children were evaluated from September 1, 2007, to, August 31, 2008, prospectively. Of these 277 patients, 39 patients (14.7%) died at the end of the PICU stay. Discrimination between death and survival was assessed by calculating the area under the ROC curve for each model. The standardized mortality ratios (SMRs) were also determined. Calibration was assessed using Hosmer Lemeshow's test. Results: In our analysis, PRISM I (area under ROC curve: 0.884; SMR: 1; Hosmer Lemeshow chi-square p value: 0.09) and PIM II (area under ROC curve: 0.912; SMR: 1; Hosmer Lemeshow chi-square p value: 0.30) showed an adequate discrimination between death and survival as well as good calibration. Conclusions: In conclusion, both PRISM I and PIM II are reliable models for evaluating the prognosis of children in PICU. However, PIM II seems to be more useful because of its higher discrimination, better calibration and easier to use. (Turk Arch Ped 2010; 45: 18-24)