[Estimation of the validity of pediatric severity rating scales in multidisciplinary children's intensive care units].

The aim of this study was to estimate the validity of pediatric severity rating scales at general pediatric intensive care units. This was a prospective, observational multicenter study that was performed from November 1, 2007, to January 31, 2008. The end points were the estimation of the severity of disease, by using the PRISM, PIM II, RELOD scores, and 28-day survival. The investigation covered 226 patients whose mean age was 4.1 +/- 0.3 years. Twelve (5.3%) patients died during the study. The Medcalc computer program was employed for statistical analysis. The area under the receiver operating curve (ROC) was 0.623 +/- 0.116 for PRISM (p = 0.289); 0641 +/- 0.116 for PELOD (p = 0.224); 0.833 +/- 0.096 for PIM II (p < 0.001). The calibration of scores according to the intervals of mortality risk (Hosmer-Lemeshov test) showed that the observed and predicted mortality rates were similar for PIM II scores (chi2 = 8.23; p = 0.084) and were different for PRISM scores (chi2 = 203.5; p < 0.001) and PELOD ones (chi2 = 26.16; p < 0.001). Thus, the PIM II scores showed the best validity. PRISM and PELOD can be used to estimate the severity of disease in individual subgroups of high mortality-risk patients.