Using SCORE2 with a risk chart or online calculator: impact on model performance, treatment eligibility, and...

Current European Cardiovascular Disease (CVD) prevention guidelines recommend 10-year risk assessment using the SCORE2 model to identify individuals eligible for preventive treatment. Risk can be estimated using conventional risk charts or online calculators, though these methods may differ in precision and treatment classification.

Individuals without established CVD or diabetes mellitus were included from CPRD (UK, Europe's low risk region, n = 977 616) and HAPIEE (Czech Republic and Poland, high risk region and Lithuania, very high risk region, n = 11 739). During median 8.4 years [interquartile range (IQR) 5.0-10.4], 22 898 CVD events occurred. SCORE2 risk was estimated via two methods: an online calculator (unrounded SCORE2 algorithm) and risk charts from the 2021 ESC Prevention Guidelines. Predicted risks were higher with the risk charts than with the online calculator. In the low risk region, the median 10-year risk was 4.0% (IQR 2.0-6.0) with the risk charts vs. 3.7% (IQR 2.3-5.8) with the calculator. In the high/very high-risk region, risk was 9.0% (IQR 5.0-15.0) and 8.4% (IQR 4.5-13.9), respectively. Chart-based risk assessment resulted in higher treatment eligibility (6.3% vs. 4.0% in the low risk region; 51% vs. 43% in high/very high risk region). Discrimination was higher with the online calculator: difference in C-statistic +0.010 [95% confidence interval (CI) 0.008-0.012] in low risk region and +0.008 (95% CI 0.005-0.010) in high/very high risk region. Calibration was adequate for both approaches. Assuming a 50% relative risk reduction for preventive treatment, this corresponded to 53 vs. 46 events prevented per 1000 treated in the low-risk region and 80 vs. 74 in the high/very-high-risk region (calculator vs. risk charts).

Risk assessment using SCORE2 risk charts yields too high predicted risks and too broad treatment eligibility. By avoiding rounding of risk factors, the online calculator shows better discrimination.