EuroSCORE 2 Calculator: Assessing Risk for Cardiac Surgery

As the field of cardiovascular surgery advanced rapidly towards the end of the 20th century, new procedures were being developed and mortality rates were falling due to continuous improvements in the quality of care in Europe. This progress was evident both during and after surgical procedures. Surgeons began to recognize the need for a tool that could accurately calculate the actual mortality risk associated with specific cardiovascular interventions. This tool would aid in better assessing which patients were the most suitable candidates for surgery and who would benefit the most, or the least. ^[8]

As a result of the combined efforts of multiple research groups, several scoring systems intended to calculate the risk of cardiovascular intervention were developed. The most promising among these were the EuroSCORE and the Society of Thoracic Surgeons Risk Algorithm (STS Score). ^{[1], [8]}

The first EuroSCORE (European System for Cardiac Operative Risk Evaluation) was introduced in 1999, representing a significant effort by a large body of researchers. They gathered data from 132 healthcare centers across the European Union and compared all the clinical parameters most associated with mortality risk. The researchers aimed to identify the clinical, laboratory, and other factors that most significantly influenced operative cardiac mortality risk. ^{[1], [8]}

The original EuroSCORE was based on data from 132 centers and 20,014 patients (19,030 analyzed after applying exclusion criteria) from eight European countries, with the majority of patients from Germany, France, the UK, Italy, and Spain. The parameters identified as posing the most risk included renal, respiratory, and vascular disease, cardiac status and LV reserve, other cardiovascular diagnostics, specific previous interventions, the urgency of the subsequent procedure, and others. The researchers found that age was strongly correlated with mortality, particularly after 60 years. ^[1]

Previous thoracic aortic surgery and active endocarditis were also strongly correlated with mortality, followed by neurological dysfunction, pulmonary hypertension, extracardiac arteriopathy, and others. A function was created to accurately calculate the mortality of patients undergoing or planning to undergo a cardiovascular intervention, based on the correlation power of each of these factors with mortality.

Despite the high accuracy of the score at the time (AUC of over 0.75 for both the additive and logistic mortality functions of the original EuroSCORE), surgeons soon realized that this score, along with all the other new inventions, procedures, policy changes, continuously upgraded medical devices, and overall improving patient management, all contributed to a steadily decreasing risk of mortality and morbidity for patients requiring cardiothoracic surgery. ^{[1], [8]}

While this score served for several years in calculating mortality risk for such procedures, it became clear that if a new, updated score was not developed quickly, the calculations made by this score would become obsolete. In addition, researchers recognized that the original score had room for improvement before it could become a perfect mortality risk calculator, and they were eager to create a new, updated version. ^{[1], [2]}

In response to this need, the new EuroSCORE II was developed, largely maintaining the same parameters as the original score. This new score was based on a slightly larger study population of 24,385 (of which 22,381 remained for further analysis), from 43 participating countries and an even larger number of care units (154) than before. ^{[1], [2]}

The team responsible for the creation of the original score later observed that the tool had gained popularity outside of the European Union, and they saw the need to include patients outside of these borders to ensure that foreign patients were adequately represented by the score's results and that the data would be validated for use in these cases as well. For the creation of this score, data was gathered from all six relevant continents.

The performance of the new score compared to applying the original scores in today's conditions showed that the EuroSCORE II had an AUC of 0.8095, compared to the original score, which had an AUC of 0.7896 and 0.7894 for the logistic and additive functions, respectively.

A follow-up study attempting to validate the new score on a dataset of 18,706 cases drew the following conclusions: ^[2]

1. The logistic score of the new tool ranged from 0.8 to 87 (with a mean of 6.2).
2. The EuroSCORE resulted in a clinically insignificant improvement from a C-statistic value of 0.77 to 0.79.
3. Some tests (the Hosmer-Lemeshow test) showed that the EuroSCORE had poor predictive power, but it was still better calibrated than the original and a modified original EuroSCORE (SCTS-modified).
4. For isolated CABG, the STS risk model had a better C-statistic than the EuroSCORE II: 0.81 vs 0.79.
5. The EuroSCORE II had a poor C-statistic result for isolated AVR (aortic valve replacement) at 0.7, which the authors found "troubling."
6. The EuroSCORE model also performed poorly for cases of combined AVR and CABG.

Another study conducted in 2013 on a cohort of 3,800 patients (4,342 cardiac procedures considered for the study) revealed the following information: ^[3] The following procedures were performed on the study cohort (over approximately five years):

• 1,231 myocardial revascularizations
• 1,727 isolated valve surgeries
• 301 combined coronary + valvular interventions
• 416 aortic surgeries
• 123 other major cardiac surgery procedures
• The average mortality rate was 5.66%; the mean mortality rates calculated with the EuroSCORE and EuroSCORE II were 9% and 4.46%, respectively.
• The new score had lower mortality rates for all pathology subgroups compared to the original score (the most significant difference was for thoracic aortic pathology, at 6.58% vs. 16.04%).
• The mortality results closest to the observed reality were closest for coronary interventions (0.19% difference in mortality between the two), followed by: (global difference of 1.2% - for all surgeries -), aortic surgery: 1.59% difference (observed mortality 8.17%, II-score: 6.58%), valve replacement: 1.72% difference; mixed procedure: 2.14%, and other surgeries (mortality rates noted above).
• The EuroSCORE II also had lower mortality rates compared to the ground truth, likely because the creators of the score wanted to make the score more future-proof. Both EuroSCOREs had an AUC of over 0.7, and for the newer score, the AUC was statistically better (0.851 vs 0.818 for all the procedures examined). The lowest AUC for the II-Score was for mixed surgeries at 0.769 (for the rest it had an AUC over 0.8), and for the original score, it was for mixed procedures (0.786) followed by valve replacements (0.779), other procedures having an AUC over 0.8.

In another 2017 study that observed mortality rates of approximately 3%, the median EuroSCORE value was 1.3%, with an AUC of 0.85, suggesting a good correlation with actual mortality rates. ^[4]

A study on 11,788 patients (data used retrospectively from patients who had cardiac surgery between 2001 and 2016) compared the performance of the original EuroSCORE, the EuroSCORE II, and the STS Score and found the following results: ^[6]

1. Mean predicted mortality was 7.8%, 3.3%, and 2.7%, respectively.
2. The AUC for the three scores was (in the same order): 0.819, 0.844, and 0.846.
3. For operative mortality, the EuroSCORE II (4% observed) had better absolute prediction and discriminative ability (5.8% mortality, AUC of 0.754) compared to the original score (12.5%, AUC of 0.688).
4. For i) CABG + valve surgeries, the STS score had the best results; for other surgeries (ii) CABG alone, iii) aortic valve surgery, iv) mitral valve surgery), the results between the scores were closer to one another.
   AUC for the EuroSCORE, EuroSCORE II, and the STS Score for the above surgeries:
   • i) 0.822, 0.848, 0.844
   • ii) 0.699, 0.761, 0.771
   • iii) 0.872, 0.898, 0.890
   • iv) 0.777, 0.819, 0.844

• The in-hospital mortality rate was 2.2%.
• The AUC of the EuroSCORE I was 0.82 (both additive and logistic) and 0.825 for the EuroSCORE II.
• The EuroSCORE II had fair calibration up until values (mortality) reached 30%; beyond 30%, it tended to overestimate mortality. ^[7]

Overall, these studies emphasize the significance of scores like EuroSCORE and EuroSCORE II in assessing mortality risks associated with cardiovascular interventions. The continuous refinement of these scores ensures accurate predictions in an evolving medical landscape.