Comparison of the Society of Thoracic Surgeons, Logistic EuroSCORE I and EuroSCORE II in Israeli Patients Undergoing Cardiac Surgery: Progress Towards a National STS-Linked Database

Background:
The need for reliable databases and risk models in cardiac surgery to monitor performance, predict and improve outcomes is well established. The purpose of this study was to compare three well-established cardiac surgery risk scores in order to determine the most accurate risk prediction model to be used in a Israeli National Adult Cardiac Surgery database linked to that of the American Society of Thoracic Surgerons (STS).

Methods:
The STS risk score, the Logistic EuroSCORE I and the EuroSCORE II were compared on their ability to accurately predict 30-day mortality. The study cohort consisted of 1,279 consecutive patients over a seven-year time period. Data were prospectively entered into our departmental STS-linked database, deindentifed, and then used to calculate STS, Logistic EuroSCORE I and the EuroSCORE II. Thirty-day mortality and the observed vs. expected 30-day mortality were calculated for each score.

Results:
The observed 30-day mortality for the entire cohort was 1.95%. The average STS, Logistic EuroSCORE I and EuroSCORE II predictive scores for 30-day mortality were 3.12%, 3.31% and 7.97% with O/E ratios of 0.62, 0.59 and 0.24, respectively. Model calibration was similar between the STS and EuroSCORE II, and both were significantly more accurate than Logistic EuroSCORE I which overestimated mortality.

Figure 1: Round black dots are the STS score, triangles are the EuroScore II, and the Logistic Euroscore are the gray squares.

Conclusions

This study demonstrated that both the STS and EuroSCORE II risk prediction models performed equally well, while the Logistic EuroSCORE I overestimated mortality. These data suggest that either the STS or the EuroScore II can be used as a reliable quality measure in Israeli patients undergoing cardiac surgery.