Comparison of prognostic scores at a pediatric ICU

Objective

The main aim of the pediatric intensive care unit (PICU) is promoting qualified care with the objective of achieving the best results and better prognosis for critically ill children. One means of comparing the quality and efficacy of care provided at a given unit is made by risk adjustment systems. The principal scores that have been developed for the pediatric population are the Pediatric Risk of Mortality (PRISM) and Pediatric Index of Mortality (PIM) scores. Our aim in this study was to compare the commonly used pediatric mortality risk scoring systems and to assess the feasibility of using these scoring systems in developing countries

Methods

The cohort study conducted prospectively in an eight-bed tertiary medical PICU in University Children's Hospital from December 2002 to July 2004. The scoring systems compared were PRISM III-12, PRISM III-24 and PIM-2. Observed and expected mortality were compared by the Lemeshow-Hosmer goodness-of-fit χ² test. Mortality was also standardized for case mix using the standardized mortality ratio (SMR). Mortality discrimination was quantified by calculation of the area under the receiver-operating characteristic curve.

Results

During the study period, 334 patients enrolled to the study. Eighty-four (25.7%) of the 334 patients studied died. Estimated mortality by PRISM III-12 was 38.71 with a standardized mortality rate of 2.17, by PRISM III-24 was 46.99 with a standardized mortality rate of 1.78, and by PIM-2 was 32.4 with a standardized mortality rate of 2.45. The Hosmer-Lemeshow test gave a chi-square of 31.1 (P < 0.001) for PRISM III-12, 21.2 (P < 0.001) for PRISM-III 24, and 34.6 (P < 0.001) for PIM-2. The area under the ROC curve was 0.80 in PRISM III-12, 0.85 in PRISM III-24, and 0.76 for the PIM model. Only 9.3% (n = 31) of the patients was admitted electively to PICU. The presence of mechanical ventilation (38% vs 7%, P < 0.01) and the existence of chronic organ disease (50% vs 24%, P < 0.001) were significantly associated with mortality.

Conclusion

The PIM test was less well calibrated overall. PRISM III-24 offers better capacity for discriminating between survivors and nonsurvivors in our country. We observed an underestimation of mortality in every scoring system. The underestimation of mortality may be associated with the existence of high proportion of chronic organ disease in our PICU.