Application of EuroSCORE II and STS score for risk assessment in Indian patients—are they useful?

To the Editor,

I read with interest the article by Shales S, comparing EuroSCORE II and Society of Thoracic Surgeons (STS) score for risk prediction in Indian patients undergoing coronary artery bypass grafting (CABG) [1]. I congratulate the authors for reporting a well-conducted study. The authors conclude that both EuroSCORE and STS score predict the mortality outcome fairly accurately in Indian patients and can be used as a risk assessment tool.

Are the authors justified in reaching this conclusion based on their results?

Risk models are evaluated by their ability for calibration and discrimination. A well-calibrated model gives the ratio of observed to expected outcome (goodness of fit) close to 1 and p value > 0.05. The Hosmer–Lemeshow goodness of fit for this study showed p value of 0.07 for the STS score and p value of 0.01 for EuroSCORE. This means that EuroSCORE is not well calibrated in the study population. Discrimination means how well the model differentiates a population that had an event from that did not. Area under the curve (AUC) by plotting a receiver operating characteristic (ROC) curve is an effective way to summarize the diagnostic accuracy of the test (i.e., ability to diagnose patients with and without the disease) or discriminatory power (differentiates a population that had an event from that did not) of a model. A value of 0 indicates a perfectly inaccurate test or model and a value of 1 reflects a perfectly accurate test or model. In general, an AUC of 0.5 suggests no discrimination (random chance), 0.7 to 0.8 is considered acceptable, 0.8 to 0.9 is considered excellent, and more than 0.9 is considered outstanding or the closer the ROC curve gets to the upper left-hand corner (0, 1), the better the test is at discriminating [2]. AUC curves also allow the investigators’ to compare the performance of two or more diagnostic tests [3] or in this case the different scoring systems. The AUC curves for STS score were 0.72 and for EuroSCORE were 0.71 in this study; meaning that both scoring systems had acceptable discriminatory ability. However when analyzed for low-risk groups and intermediate-risk groups, the AUC was well below 0.7 for both scoring systems (0.62 & 0.65 for EuroSCORE; 0.67 & 0.52 for STS score), meaning poor discriminatory ability. In any center, the vast majority of patients (> 90%) of CABGs fall in low or intermediate-risk categories and the percentage of patients falling in the high-risk CABG category is very low (3% with EuroSCORE and 0.7% with STS score in this study). Hence, the conclusions from this paper were not relevant for the majority of patients who underwent CABG. In contrast, these scoring systems consistently give AUC values above 0.8 in European and North American study populations. It is well-known fact that model-derived risks validated in one location usually have lower performance when applied in another location and that a “region-specific” scoring system outperforms “international models.” Hence, further validation of these scoring systems in the Indian population requires a large multicenter study. For further progress of the specialty in India, a central registry and population-specific risk assessment tools are essential [4, 5].

Funding

None.

Declarations

Ethics approval

Not required.

Informed consent

Not required.

Competing interests

The authors declare no competing interests.