Can generic scores (Pediatric Risk of Mortality and Pediatric Index of Mortality) replace specific scores in predicting the outcome of presumed meningococcal septic shock in children?

Objective: To compare, in children with septic shock and purpura, the accuracy in predicting death of two specific scores (the MenOPP bedside clinical [MOC] score of Gedde Dahl and the score of Groupe Francophone de Reanimation Pediatrique [GFRP]), the G-reactive protein (CRP) level, and the two pediatric generic scores (the Pediatric Risk of Mortality [PRISM] and Pediatric Index of Mortality [PIM] scores). Design: Prospective, population-based study with analysis of previous comparative studies. Setting: A 14-bed pediatric intensive care unit in a university hospital. Patients: All children admitted consecutively to the pediatric intensive care unit with septic shock and purpura (n = 58, with 16 deaths [27.6%]) from January 1993 to May 2000. Interventions: None. Measurements and Main Results: The MOC and GFRP scores and the GRP level were prospectively determined at admission. The PRISM score was prospectively calculated within 24 hrs of admission or at the time of death, and the PIM score was calculated retrospectively between 1993 and 1997 and then prospectively from admission data. The nonparametric estimate of the area under the receiver operating characteristic curves (AUC) was calculated from the raw data using the Wilcoxon-Mann-Whitney two-sample statistic, and the standard error of the AUCs was calculated with DeLong's method. All the scores had an AUC >0.80, the PRISM probability of death having the best one (0.96 +/- 0.02). The PRISM value, which is easier to calculate, had an AUC of 0.95 +/- 0.02. The PRISM score performed significantly better than the PIM score (AUG, 0.83 +/- 0.06; p < .01) and the CRP level (AUC, 0.80 +/- 0.06; p <.01); however, there was no significant difference between the MOC (AUC, 0.91 <plus/minus> 0.04) and GFRP scores (AUC, 0.87 +/- 0.05). Analyzing literature and calculating AUCs from original data of previous studies, we observed that the superiority of the PRISM scare had never been demonstrated in meningococcal diseases. Conclusions: The PRISM score performed better than the PIM score, and was not surpassed by specific scores. Thus, we propose its use for outcome prediction in children with septic shock and purpura. However, if the PRISM score is to be used as inclusion criterion for clinical trials, it should be evaluated within a few hours after admission.