Validation of the European SCORE2 algorithm in one European and one Latin-American cohort studies with contrasting populations

SCORE2 estimates the 10-year risk of developing non-fatal and fatal cardiovascular disease (CVD), and it was designed for mostly Europeans. However, its transferability to non-European populations has not been extensively studied. The aim of this study was to evaluate the performance of the SCORE2 in ethnically and socioeconomically diverse populations, including those outside of Europe.

We leveraged cohort studies from Mexico (N = 95,660, mean age 52.4 years, 32.6% men, mean follow-up 9.7 years), and Portugal (N = 1,216, 52.9 years, 36.9% men, 8.8 years follow-up). SCORE2 has differentially classified frequency of CVD events across levels of risk severity, ranging from 0.7–1.0% CVD events among those with low-moderate risk-profiles, vs. 6.6–9.0% among those with high risk-profiles, respectively. The area under the receiver operating curve (AROC) was 0.762 in Mexico and 0.718 in Portugal. Sensitivities (95% CI) were 85.7 (84.2–87.2) and 95.0 (86.1–99.0), and specificities were 42.8 (42.5–43.1) and 24.3 (21.9–26.9) for Mexico and Portugal, respectively. In all cases, the positive predictive values were below 7%, whereas the negative predictive values were all above 98%. Harrell’s C were 0.717 and 0.692, and Brier’s scores were 0.022 and 0.052 for Mexico and Portugal, respectively. In Mexico, SCORE2 showed good calibration, with a slight overestimation in the higher-risk categories. In Portugal, overestimation in higher-risk categories was more substantial.

SCORE2 estimates modest to acceptable performance for predicting the 10-year CVD in Mexicans but performed less favourably in the Portuguese cohort, indicating a need for recalibration.