Clinical background
SCORE2 was developed to replace its predecessor SCORE, which estimated only the risk of fatal cardiovascular disease (CVD) and therefore underestimated the true clinical burden of non-fatal events. SCORE2 instead estimates the 10-year risk of a composite of both fatal and non-fatal CVD, which better reflects the decision the clinician actually faces: whether primary prevention, above all lipid-lowering treatment, is warranted in an otherwise healthy person.
The tool is embedded in the 2021 ESC guidelines on CVD prevention, where it forms the basis for risk stratification of people without established CVD or diabetes. Its central advantage over earlier models is regional recalibration: the same risk factor profile gives a different absolute risk in different parts of Europe, so that a patient who in a low-risk region falls below the treatment threshold may lie above it in a high-risk region.
Calculating Systematic Coronary Risk Evaluation 2 (SCORE2)
SCORE2 rests on sex-specific Cox models adjusted for competing risks, in which the linear predictor (LP) is a function of age, smoking status, systolic blood pressure, total cholesterol and HDL cholesterol, with interaction terms between age and the other variables. The estimated risk is calculated as:
where is the sex-specific baseline survival at 10 years and LP the sex-specific linear predictor. The result is then recalibrated to one of four ESC risk regions via a complementary log-log transformation, which means that the same LP gives a different absolute risk depending on the region chosen.
The derivation cohort comprised 677,684 individuals from 45 cohorts in 13 European countries, with 30,121 CVD events during follow-up [1]. The models were developed separately for men and women with separate coefficients for all the predictors. Regional recalibration was based on estimates of CVD incidence and risk factor distribution in data from 10,776,466 individuals, divided into four risk regions by country-specific CVD mortality: low risk, moderate risk, high risk and very high risk [1].
Interpretation in practice
The estimated 10-year risk should be interpreted against the ESC risk thresholds for the age group 50 to 69 years, in which 5% and 10% are the boundaries for high and very high risk [2,4]. For people aged 40 to 49 years, a single threshold of ≥5% applies for classification as high risk, according to the ESC guidelines. The calculator requires the user to select the cardiovascular risk region explicitly, and the choice markedly affects the result: a 50-year-old male smoker with a systolic blood pressure of 140 mmHg, a total cholesterol of 5.5 mmol/L and an HDL of 1.3 mmol/L has an estimated 10-year risk of 5.9% in the low-risk region versus 14.0% in the very high risk region [1].
| Risk level | 10-year risk | Clinical action |
|---|---|---|
| Low to moderate | < 5% | Lifestyle advice, reassessment when risk factors change |
| High | 5% to < 10% | Consider lipid-lowering treatment, intensified lifestyle advice |
| Very high | ≥ 10% | Lipid-lowering treatment indicated, active modification of risk factors |
The choice of region is the single step that most affects the result. Using the wrong region — for example applying the low-risk region's recalibration to a patient from a high-risk region — can lead to substantial underestimation of risk and thereby to undertreatment.
Validation and performance
External validation was performed in 25 cohorts from 15 European countries with 1,133,181 individuals and 43,492 CVD events [1]. The c-index ranged from 0.67 (95% CI 0.65 to 0.68) to 0.81 (95% CI 0.76 to 0.86) depending on the cohort, reflecting differences in population risk and follow-up.
In an external validation in the Clinical Practice Research Datalink (CPRD, United Kingdom, n = 518,015, 12,675 CVD events), SCORE2 achieved a c-index of 0.737 (95% CI 0.732 to 0.741) [3]. The same study showed that adding risk modifiers such as the coronary artery calcium score, high-sensitivity troponin T and a family history gave a modest improvement in discrimination to 0.740 (95% CI 0.636 to 0.745) without worsening calibration [3]. This indicates that SCORE2 as a standalone instrument captures most of the predictive information in the conventional risk factors, but that its discrimination is moderate.
In a Korean population cohort (NHIS, n = 324,384), SCORE2 overestimated risk in all age groups and in both sexes, with a c-index of 0.604 for men and 0.625 for women [2]. Calibration was thus inadequate in an East Asian population, underlining that the model was developed and recalibrated for European populations.
Limitations
SCORE2 is not validated for patients with known CVD, diabetes mellitus, familial hypercholesterolaemia or chronic kidney disease, and should not be used in these groups. For people over 69 years there is a separate model, SCORE2-OP, with its own coefficients and recalibration. The tool does not capture the effect of certain risk modifiers used in clinical practice, among them the coronary artery calcium score, family history, CRP, lipoprotein(a) and socioeconomic status, which under the ESC guidelines may justify adjusting the estimated risk upwards or downwards [3]. The flexible method described by Hageman et al. for adding such modifiers nonetheless yields only a marginal increase in discrimination [3].
A further limitation is that risk is not static. A patient who today lies below the treatment threshold may cross it simply by growing older or by a change in risk factors, so reassessment should take place regularly. Finally, the model overestimates risk in populations outside Europe, as shown in East Asian cohorts [2], and should not be used without region-specific recalibration in such populations.
Application in the SCAPIS cohort
In the Swedish SCAPIS study (Swedish Cardiopulmonary Bioimage Study), the 10-year risk was estimated with SCORE2 in 26,570 apparently healthy individuals aged 50 to 64 years [4]. Of these, 32% were judged to have a high and 4% a very high 10-year risk of CVD, which meant that 35% of the total population became eligible for lipid-lowering treatment under the ESC guidelines. Among those eligible for treatment who underwent coronary CT angiography, 38% had no evidence of coronary atherosclerosis [4]. Using the SCORE2 risk charts instead of the equation increased the proportion eligible for treatment to 52%, of whom 44% had no atherosclerosis [4]. The results highlight a clinical tension: a substantial proportion of an apparently healthy Swedish middle-aged population becomes eligible for treatment under current guidelines, and some of them have no demonstrable subclinical atherosclerosis.
References
- SCORE2 working group and ESC CVD Risk Collaboration. SCORE2 risk prediction algorithms: new models to estimate 10-year risk of cardiovascular disease in Europe. Eur Heart J 2021;42(25):2439–54. PMID: 34120177
- Choi J, Sung S, Park SK et al. SCORE and SCORE2 in East Asian Population: A Performance Comparison. JACC Asia 2024;4(4):265–74. PMID: 38660103
- Hageman SHJ, Petitjean C, Pennells L et al. Improving 10-year cardiovascular risk prediction in apparently healthy people: flexible addition of risk modifiers on top of SCORE2. Eur J Prev Cardiol 2023;30(15):1705–14. PMID: 37264679
- Yari A, Ueda P, Lundman P et al. Eligibility for lipid-lowering therapy when applying systemic coronary risk estimation 2 according to guidelines on apparently healthy middle-aged individuals. Eur J Prev Cardiol 2024;31(15):1890–7. PMID: 38842486
