Development of a pediatric multiple organ dysfunction score: Use of two strategies

Background. An organ dysfunction (OD) scoring system for critically ill chi ldren is not yet available, and the method for developing such a system is not well defined. The aim of this study was to compare two developmental me thods for assessing OD in critically ill children. Methods. Consecutive adm issions between January and May 1997 in three French and Canadian pediatric intensive care units (PICUs) were studied prospectively. Physiologic data were selected using a Delphi method; the most abnormal values during PICU s tay were recorded. The outcome measure was the vital status at PICU dischar ge. Six organ systems were studied: hepatic, cardiovascular, renal, hematol ogic, respiratory, and neurologic. For each of the six organ systems, the P Ediatric Multiple OD (PEMOD) system included one variable and the PEdiatric Logistic OD (PELOD) system included several variables. Severity levels and relative weights of ODs were determined according to the mortality rate (P EMOD) or by logistic regression (PELOD). Results. There were 594 admissions , including 51 deaths (9%). Severity levels and relative weights of ODs wer e: four levels graded from 1 to 4 for the PEMOD system and three levels wit h scores of 1, 10, and 20 for PELOD system. For both systems, calibrations were good (p = 0.23 and p = 0.44 respectively). The PELOD system was more d iscriminant than the PEMOD system (areas under the ROC curves 0.98 and 0.92 , respectively, p < 10(-5)). Moreover, with the PEMOD system, four ODs did not contribute significantly to the prediction of PICU outcome. Conclusions . The PELOD system was more discriminant and had the advantage of taking in to account both the relative severities among ODs and the degree of severit y of each OD.