Methods: Between June 2003 and January 2009, medical files of 456 pediatric patients who were monitored in a 12-bed postoperative cardiac surgery care unit following heart surgery were retrospectively analyzed. A total of 373 files were included in the study. Age, gender, diagnosis, the length of stay in the intensive care unit and hospital, survival rates, PIM 1, PIM 2 scores and m-SOFA scores on admission, at 24 and 48 hours and peak m-SOFA scores were recorded. Student’s t test was used to compare the normally distributed data, whereas Mann-Whitney-U test was used to compare non-parametric data. Calibration of the scores was performed using the Hosmer and Lemeshow Goodness of Fit test. Discrimination power of the scores was analyzed using the receiver operating characteristic (ROC) curves.

Results: Fifty patients (13.4%) died perioperatively. Peak and m-SOFA scores on admission were significantly higher in nonsurvivors (9.8±2 and 9.2±2, respectively) than survivors (5±2.5 and 4.6±2.5, respectively; p<0.01). Calibration with Hosmer- Lemeshow Goodness of Fit test was chi-square df (8)=30.4, p=0.0002 for PIM 1 and chi-square df (9)=13.5, p=0.13 for PIM 2. Discrimination power and calibration strength of PIM 2 score was good (ROC 0.82), whereas PIM 1 had a better value (ROC 0.87) of discrimination power with a poor calibration strength. The ROC values of peak and m-SOFA scores on admission were observed to have a good discrimination power (0.93 and 0.92, respectively).

Conclusion: Our study results demonstrate that peak and m-SOFA scores on admission are improved for the prediction of mortality in pediatric cardiac surgery, compared to PIM 1 and PIM 2 scores.