Abbreviations and acronyms:
ASCVD: atherosclerotic cardiovascular disease
BMI: body mass index
CVD: cardiovascular disease
CAC: coronary artery calcium
CONOR: COhort of NORway
ESC: European Society of Cardiology
non-HDL chol: non-high density lipoprotein cholesterol
O.P.: old persons
SCORE: Systematic COronary Risk Evaluation
WHO: World Health Organization
- In apparently healthy people aged 40-69 years it is recommended to estimate the 10-year total atherosclerotic cardiovascular disease (ASCVD) risk with the SCORE 2 model and in those aged ≥70 years with the SCORE 2 O.P. model. These models are calibrated to four subgroups of countries varying in low, moderate, high and very high age-standardised cardiovascular mortality rates. Charts with the results of these models are provided in references 2, 6 and 11.
- Age-specific 10-year ASCVD risk cut-offs together with considerations of risk modifiers and qualifiers, comorbidities and patient preferences, should be used to guide decisions and actions of preventive strategies in apparently healthy people.
All current guidelines on the primary prevention of ASCVD in clinical practice, including those of the European Society of Cardiology (ESC), recognise the multifactorial origin of the underlying pathology and recommend the use of standardised risk estimation systems to stratify the apparently healthy population into subgroups according to total ASCVD risk. The reason for that is to classify the population into different risk categories in order to adjust the intensity of risk factor management to the severity of the total ASCVD risk: the higher the risk, the more intense the prevention strategy should be. These adjustments are needed because the resources for preventive medicine are limited in most health care systems; therefore, one should use them as efficiently as possible. This has to do with cost-benefit and benefit-risk considerations. Risk stratification can also improve the equity in the distribution of effective prevention strategies.
But ASCVD risk estimation has also limitations; health professionals and the public tend to look at these estimates on an individual basis while the strategies that are proposed for its ultimate use are based on a group approach. The stratification of the population into subgroups at low, moderate, high or very high risk is reliable, but within each of these categories, identifying which person will develop the disease is less certain. ASCVD risk estimation of a given person remains hazardous, very approximate, and therefore, at present, elusive. Total ASCVD risk estimation models are not crystal balls for prophesying .
Total ASCVD risk estimation
How to estimate total ASCVD risk? First of all, there are patients in whom total ASCVD risk is anyhow high or very high and they need strict management of their risk factors; according to the 2021 ESC prevention guidelines , these are patients with documented ASCVD, those with type 2 or type 1 diabetes mellitus, patients with genetic/rarer lipid or blood pressure disorders and patients with moderate and severe chronic kidney disease. Their total ASCVD risk is high or very high.
In apparently healthy persons, representing the large majority of the adult and elderly population, the total ASCVD risk is built up by several risk factors that interact with each other, sometimes synergistically, and that may, in combination, result in high levels of total ASCVD risk. But the clinical estimation of these combined effects is unreliable [3,4]. Moreover, clinicians treat patients, not isolated risk factors. To overcome these problems, risk estimation systems have been developed. Which one should be preferred?
Strictly speaking, risk estimates are applicable only to the population from which they are derived. But in daily practice, one should extend this using the model that is based on observations in a recent cohort of people that is recognisably comparable to the patient population with whom one works.
The SCORE model
SCORE  is the risk estimation system endorsed by the ESC and integrated in the ESC prevention guidelines since 2003. The SCORE system is based on a dataset from 12 large European cohorts including 205,OOO participants free of cardiovascular disease at baseline, aged 40 to 65 years, on 3 million person-years of observation and on over 7,000 fatal cardiovascular (CV) events as outcome. The model estimates the risk of a first fatal CVD event within the coming 10 years. The SCORE model has been externally validated; it can be recalibrated at the national level using more recent prevalence data on the risk factors involved. An electronic version is available on the ESC web as HeartScore (www.heartscore.org). SCORE produced two versions, one for low- and one for high-risk countries; four different risk categories were suggested: low, moderate, high and very high risk.
The SCORE model has great advantages but also limitations. The model is based on risk factor levels and CVD mortality rates in cohorts that were examined many years ago. When applied to populations in which the incidence of ASCVD has declined the model may overestimate total ASCVD risk, and the reverse is true when applied to populations in which the prevalence of the disease is on the increase. Given the huge variation in the dynamics of ASCVD incidence between European countries, the production in SCORE of only two versions, one for low- and one for high-risk countries may also be considered as a limitation.
The SCORE 2 model
To overcome these and other restrictions, a new model (SCORE 2) has been developed, validated and published , and has become part of the 2021 ESC prevention guidelines .
Sex-specific competing risk-adjusted risk models were derived from 45 cohort studies, recruited between 1990 and 2009 in 13 European countries and including more than 670,000 apparently healthy men and women aged 40-69 years at baseline, in whom more than 30,000 ASCVD events occurred. Results were recalibrated to four distinct European regions using age-, sex-, and region-specific risk factor values and CVD incidence rates derived using data from approximately 10.8 million individuals. Four versions have been developed for low, moderate, high and very high risk regions based on age- and sex-standardised World Health Organisation (WHO) CVD mortality rates. In each of these regions, risk stratification is proposed in three risk categories: low to moderate, high and very high ASCVD risk. The model estimates the 10-year risk of developing a fatal or a non-fatal cardiovascular event including CV mortality, non-fatal myocardial infarction and non-fatal stroke.
External validation took place in 25 prospective cohorts in 15 European countries including approximately 1.1 million individuals and 43,000 CVD events.
In SCORE 2 non-high density lipoprotein cholesterol (non-HDL cholesterol) is used compared with total cholesterol in SCORE. Non-HDL cholesterol reflects better the majority of atherogenic lipoprotein fractions and takes also account of effects of HDL-cholesterol on CVD risk. In comparison with the SCORE model the results of SCORE 2 suggest that this model estimates the total burden of ASCVD better particularly among younger individuals and shows better risk discrimination (6).
Among the strengths of SCORE 2 one can list the following:
- it provides risk estimates for both fatal and non-fatal CVD events, which is a better reflection of the actual CVD burden;
- it has been systematically recalibrated using contemporary CVD rates;
- SCORE 2 accounts for the impact of competing risks by non-CVD deaths; this prevents overestimation of CVD risk and overestimation of the benefit of treatments in populations where the risk of competing non-CVD deaths is high, particularly in older subjects;
- the recalibration of SCORE 2 to four distinct European regions defined by varying CVD risk levels improves on the two-level regional stratification of the SCORE model;
- the derivation, calibration and validation of SCORE 2 are based on powerful, extensive and contemporary datasets, enhancing the accuracy, generalisability and validity of the approach.
In terms of limitations of SCORE 2 one could discuss the representativeness of the study cohorts, but these are the best of what is available. In many cohorts results are missing regarding other CV risk factors that may be important in CV risk estimation, such as family history, socio-economic status, physical activity or dietary habits. SCORE 2 is also limited to the adult population aged less than 70 years.
The SCORE 2 O.P. model
The age limit of SCORE 2 was well realised particularly because age is the dominant driver of ASCVD risk; around two thirds of all CVD deaths in Europe occur in the population aged 75 years and over. CVD is also responsible for considerable morbidity and reduction in quality of life. Death is our fate but prevention of disability-adjusted life-years in the elderly is still a major challenge to preventive cardiology. The consequences of non-fatal CV events such as stroke or heart failure on the quality of life should be prevented by all means also in the elderly. In observational studies it has been documented that CV risk factors such as arterial hypertension, dyslipidaemia and diabetes mellitus continue to predict ASCVD in the elderly. There is also evidence from randomised clinical trials demonstrating the benefits in terms of ASCVD prevention by treating elevated blood pressure, diabetes and dyslipidaemia in older persons [7-9]. Therefore, it is of interest to estimate the CVD risk even in the elderly population. That was already realised by the authors of the SCORE model; a SCORE O.P. model was developed on a small dataset  but it was not integrated in previous prevention guidelines.
The SCORE 2 working group decided to develop an updated SCORE 2 O.P. system .
SCORE 2 O.P. is derived from a dataset based on the Norwegian CONOR cohort (Cohort of Norway) from which competing risk-adjusted, sex-specific coefficients were derived from observations in 28,500 participants. This also included age-interactions for all risk factors to account for differences in the relationship between risk factors and CV outcomes across different ages.
The model was recalibrated to four distinct European regions using contemporary region-specific CVD event rates and risk factor levels.
The model was externally validated in approximately 340,000 individuals from different CV risk regions. The model estimates the sex-specific 10-year risk of developing a fatal or non-fatal CV event in age categories between 70 and 89 years as a function of smoking status, systolic blood pressure and non-HDL cholesterol level.
However, total ASCVD risk is also affected by other qualifying or modifying risk factors than those included in the SCORE 2 and SCORE 2 O.P. models such as socio-economic status, ethnicity, physical inactivity, a family history of premature ASCVD, body composition (BMI, waist circumference), frailty, environmental exposure (air, soil, noise) and inflammatory conditions (rheumatoid arthritis). These factors should be taken into account particularly when the person’s estimated risk according to SCORE 2 or SCORE 2 O.P. is close to decision thresholds, such as in the category at intermediate total ASCVD risk. These risk modifiers may then change the estimated total ASCVD risk in both directions. But the inclusion of risk modifiers to existing ASCVD risk estimation systems, separately or as a multimarker, improves risk estimation only moderately  and much less than what could be expected from the relative risks reported for these factors in observational studies .
Demonstrating the predictive value of a biomarker, independent of conventional risk factors, is by itself insufficient proof that this marker carries incremental value in existing risk estimation models. In the 2021 ESC prevention guidelines the routine collection of other potential modifiers, such as genetic risk scores, circulating or urinary biomarkers is not recommended .
Risk estimation can also be improved by considering results from investigations looking for subclinical ASCVD, in particular by using imaging techniques.
Coronary artery calcium (CAC) score is a surrogate marker of subclinical ASCVD in apparently healthy individuals, and predicts ASCVD independently. Risk estimates provided by CAC scoring may improve the accuracy of existing ASCVD risk estimation systems. But in people at high or at very low total ASCVD risk according to existing models, CAC scoring has a limited ability to reclassify ASCVD risk. Plaque detection by carotid ultrasound is an alternative when CAC scoring is unavailable or not feasible. This may help decision-making regarding drug therapies from both the health professional’s and the patient’s viewpoint in individuals with intermediate total ASCVD risk estimates.
ASCVD risk stratification
Finally, there is another important change in the 2021 ESC prevention guidelines, compared with the previous version, that will affect risk management strategies; it relates to risk stratification within a given region.
Total ASCVD risk is a continuum; all cut-offs to define risk categories are arbitrary, not evidence-based but mainly based on practical considerations in relation to the health care system, the health insurance plan and economic determinants. The choice of the cut-offs reflects the ability of the health system to care for the persons at risk. They depend on the risk/benefit ratios and on the available resources. From a strict theoretical viewpoint, a total ASCVD risk estimate can be labelled as ‘high’ when it reaches a level above which the chance of developing ASCVD is increased and a reduction of the total ASCVD risk is more effective at this level than harmful and cost-efficient.
In the 2016 ESC prevention guidelines, “low to moderate CVD risk” was defined as a SCORE level of <5% - less than 5% chance of dying from CVD within 10 years - with a recommendation for lifestyle changes to maintain this low to moderate risk status. High risk was defined as when the SCORE estimate was between 5 and 10%; this would qualify the patient for intensive lifestyle advice and in some circumstances drug treatment for certain candidates. A SCORE estimate of 10% or more was considered as very high risk, and in these cases drug treatment for risk factor control is more frequently required.
In the 2021 ESC prevention guidelines the translation of CVD risk into risk stratification has substantially changed. The cut-offs for different categories of ASCVD risk are numerically different for various age groups to avoid undertreatment in the young and overtreatment in older persons.
This is presented in Table 1. As age is a major driver of CVD risk but lifelong risk factor treatment benefit is higher in younger people, the risk thresholds for intensifying preventive actions are lower for younger patients.
Table 1. ASCVD risk categories based on SCORE 2 and SCORE 2 O.P. (10-year risk of a fatal or a non-fatal CVD event) in apparently healthy people according to age*.
|<50 years||50-69 years||≥70 years|
Low to moderate ASCVD risk
High ASCVD risk
|2.5 to <7.5%||5 to <10%||7.5 to <15%|
|Very high ASCVD risk||≥7.5%||≥10 %||≥15%|
ASCVD: atherosclerotic cardiovascular disease; CVD: cardiovascular disease
Reproduced from  (Visseren FLJ, et al, (2021 ESC Guidelines on cardiovascular disease prevention in clinical practice. Eur Heart J. 2021;42:3227-337) permission of Oxford University Press on behalf of the European Society of Cardiology (www.escardio.org/Guidelines).
The division of the population into three distinct age groups (<50, 50-69, and >70 years) results in a discontinuous increase in risk thresholds for low to moderate, high, and very high risk. In reality, age is continuous, and a sensible application of the thresholds in clinical practice requires some flexibility in handling these risk thresholds as patients move towards the next age group or have recently passed the age cut-off.
Using these cut-offs in risk stratification one should always remember its purpose: to adapt the intensity of intervention in accordance with the level of total ASCVD risk. But there is no high threshold for total ASCVD risk that implies mandatory treatment. Across the entire range of ASCVD risk the decision to initiate interventions remains a matter of individual considerations and shared decision-making. Risk categories do not automatically translate into recommendations for starting lipid-lowering or antihypertensive drug therapies; shared decision-making will also consider other risk modifiers, comorbidities and patient preferences.