Cardiovascular risk is the probability that a person will experience one or more cardiovascular events within a defined period. For an estimate to have clinical meaning, the predicted event, time horizon, population to which the model applies, and conditions under which predictors were measured must be specified. The 10-year risk of myocardial infarction or stroke, the risk of cardiovascular death, and the overall probability of cardiovascular disease including heart failure are not interchangeable measures. A percentage without this information may be mathematically correct yet misinterpreted.
Risk assessment helps guide prevention before an event and tailor treatment in people with already recognized disease. It is not a diagnosis of atherosclerosis, nor a test to explain chest pain, dyspnea, or neurological deficits. The distinction between primary prevention and secondary prevention identifies two different settings: the former estimates the probability of a first event, whereas the latter aims to prevent recurrence in someone who has already demonstrated clinical susceptibility. A subclinical atherosclerotic finding lies in an area that requires assessment of severity and prognostic significance and is not automatically equivalent to a previous myocardial infarction.
The epidemiological burden justifies a systematic approach. The Global Burden of Disease 2019 estimated approximately 523 million people living with cardiovascular disease and 18.6 million cardiovascular deaths worldwide that year. These figures cover cardiovascular diseases as a whole, not only atherosclerotic conditions, and refer to 2019 estimates. Individual probability varies widely by age, population, and social conditions. Even when age-standardized rates decline, population aging and growth can increase the absolute number of people affected. Prevention must therefore combine individual clinical decisions with interventions addressing population-level determinants of exposure.
Ten-year risk is useful for many treatment decisions but may underestimate the importance of exposures in younger people, whose immediate risk remains low despite a substantial cumulative burden. Conversely, advanced age may confer a high probability without every intensive intervention providing the same benefit, particularly in the presence of frailty and competing diseases. Modern assessment combines absolute risk, lifetime perspective, expected treatment benefit, and the practical ability to sustain treatment. The clinical goal is not to obtain a reassuring category but to reduce events and loss of independence through interventions proportionate to the person's circumstances.
A predictor is not necessarily a modifiable cause. For LDL, blood pressure, and smoking, genetic, epidemiological, and experimental evidence supports causal roles, and targeted interventions reduce events. For other markers, an association may reflect preexisting disease, shared factors, or a biological function that a simple measurement does not adequately describe. HDL-C is an example: it is used in prediction models, but pharmacologically raising its concentration has not demonstrated uniform benefit. This distinction prevents every value statistically associated with prognosis from being turned into a treatment target.
Atherosclerotic risk results from interactions among lipoprotein exposure, arterial wall susceptibility, inflammatory responses, and thrombotic propensity. The concentration and duration of exposure to apoB-containing particles determine the amount of material retained in the intima. Smoking and hypertension alter the endothelium and arterial wall; diabetes and insulin resistance modify lipid metabolism, inflammatory signaling, and vascular function. Observed risk therefore cannot be derived from simply counting the risk factors present. Two people with the same number of factors may have very different probabilities because intensity, duration, interactions, and previously accumulated damage differ. The monograph on atherosclerosis addresses arterial wall lesions; prognostic stratification instead integrates this process with the patient's overall clinical context.
Age and sex are powerful predictors but do not provide sufficient explanations on their own. Age incorporates part of cumulative exposure and biological changes; sex is associated with differing distributions of factors, age at onset, and reproductive history. In women, preeclampsia, hypertensive disorders of pregnancy, and premature menopause may provide additional information. A family history of premature disease suggests inherited and environmental contributions and should be clarified according to event type, degree of relatedness, and age. A vague report of “heart problems in the family” is of limited value if it includes nonatherosclerotic conditions or events occurring at a very advanced age.
Kidney function and albuminuria provide complementary information. Reduced eGFR indicates abnormalities of filtration, while albuminuria may document renal and vascular damage even when filtration is preserved. In diabetes, disease duration, glycemic control, and target-organ damage modify risk beyond the mere presence of the diagnosis. Visceral obesity and metabolic syndrome cluster mechanisms that also increase the risk of heart failure and kidney disease; an estimate limited to atherothrombosis does not necessarily capture all these outcomes. It is therefore essential to know which events the model includes rather than use one percentage as a universal measure of cardiovascular health.
Chronic inflammation, obstructive sleep apnea, physical inactivity, diet, pollution, and psychosocial conditions influence risk through different pathways. Some are included directly in prediction models; others may act as modifiers. An epidemiological association does not justify applying arbitrary multipliers. Social deprivation may also increase exposures and hinder access to care, adherence, and opportunities for physical activity or a healthy diet. Clinical assessment must recognize these constraints, because a theoretically effective prescription provides little benefit if it is not sustainable. The biology of risk and the circumstances that make it modifiable are part of the same preventive reasoning.
High risk is generally asymptomatic. History-taking begins by checking whether the person truly belongs in a primary prevention population: previous myocardial infarction, stroke, or transient ischemic attack of atherosclerotic origin, symptomatic peripheral artery disease, revascularization procedures, and significant anatomical findings alter the approach. Available records must be reviewed because a nonspecific label of “ischemia” or a prescription for aspirin does not, on its own, confirm a previous event. Current symptoms requiring independent diagnostic assessment must be sought in parallel. A low risk score is not a reason to delay investigation of suspicious chest pain or a recent neurological deficit.
Risk factor history must be specific and time-oriented. For smoking, current and previous use, duration, and cessation are distinguished; for blood pressure, repeated measurements, treatment, and home readings are considered; for lipids, pretreatment results are particularly useful. In diabetes, age at diagnosis, duration, HbA1c, kidney function, and complications are recorded. Actual physical activity, diet, sleep, weight trends, and alcohol intake are explored without attributing a preventive benefit to starting alcohol consumption. Family history focuses especially on early events and inherited dyslipidemias, which may make a decision based solely on a general-population model inappropriate.
Medication, dosage, actual use, and adverse reactions must be recorded, along with reasons for any discontinuation. A low LDL value achieved with a statin cannot be treated as the person's untreated level to conclude that therapy is no longer needed. Similar caution applies to blood pressure and glycemia during treatment. Frailty, falls, orthostatic hypotension, cognitive function, independence, comorbidities, and life expectancy affect net benefit, especially in older adults. Patient preferences inform the decision after understandable explanations of alternatives and absolute benefit; they do not replace the need for a correct estimate.
Physical examination includes standardized blood pressure measurement, heart rate and rhythm, weight, height, and adiposity distribution. A single blood pressure measurement during pain, anxiety, or acute illness can be misleading; home confirmation or 24-hour monitoring helps identify white-coat effects and masked hypertension. The examination looks for signs of target-organ damage, heart failure, abnormal pulses, and foot lesions when clinically relevant. Tendon xanthomas suggest familial hypercholesterolemia, but their absence does not exclude it. No individual examination finding alone quantifies overall risk or establishes that a person with a normal examination is safe.
Risk stratification is not a diagnosis based on disease criteria. Under European recommendations, the clinical population is identified first, then the appropriate model selected, and finally the result integrated with conditions that already determine risk and documented modifiers. People with clinical ASCVD are not reclassified using an algorithm for a first event. Familial hypercholesterolemia, markedly abnormal individual risk factors, diabetes with target-organ damage, and kidney disease also require specific approaches. This sequence prevents the most important practical error: applying a calculator to someone outside the population for which it was developed.
For apparently healthy adults aged 40–69 years without known cardiovascular disease or diabetes, SCORE2 estimates the 10-year risk of a first fatal or nonfatal cardiovascular event, including myocardial infarction and stroke. It uses age, sex, smoking, systolic blood pressure, and lipid profile, represented in the risk charts by non-HDL cholesterol. Non-HDL and LDL must not be interchanged in the input field. The model is calibrated for groups of countries with different levels of population risk; Italy belongs to the moderate-risk European group. This classification refers to national calibration, not to every Italian individual's risk being moderate.
SCORE2-OP is intended for older people, within its operational age range of 70–89 years, and accounts for the greater importance of competing mortality. The result remains a population-based estimate applied to an individual, not a certain prediction. For people beyond the validated age range, entering a younger age or automatically extrapolating the equation is inappropriate. For people under 40, 10-year risk often fails to represent the preventive concern adequately: extreme exposures, family history, and lifetime perspective become more relevant. The absence of an applicable score does not imply a lack of preventive indications.
It is necessary to distinguish the calculation from the decision threshold applied. The 2021 ESC prevention guidelines use age-dependent thresholds: under age 50, high risk is from 2.5% to less than 7.5%, and very high risk from 7.5%; at ages 50–69, high risk is from 5% to less than 10%, and very high risk from 10%; at age 70 and older, high risk is from 7.5% to less than 15%, and very high risk from 15%. Lower values define the low-to-moderate category in each group. These thresholds guide general prevention and must not be presented as diagnostic values or natural biological boundaries.
The 2025 ESC/EAS dyslipidemia update uses a different operational classification for lipid management in apparently healthy people to whom SCORE2 or SCORE2-OP applies: low risk below 2%, moderate from 2% to less than 10%, high from 10% to less than 20%, and very high from 20%. Clinical conditions that directly assign a risk category remain decisive. The two documents therefore do not allow the term “high risk” to be used without specifying the system: the same numerical probability may receive a different label because the decision framework has changed. Reports should state the model, version, percentage, and guideline framework used for the decision.
In type 2 diabetes, SCORE2-Diabetes adds age at diabetes diagnosis, HbA1c, and eGFR to the traditional predictors. The 2023 ESC guidelines recommend its use in the appropriate population, especially those aged 40–69 years without ASCVD or severe target-organ damage. Risk categories are low <5%, moderate from 5% to <10%, high from 10% to <20%, and very high ≥20%. With ASCVD or severe target-organ damage, classification does not depend on the calculated result. Type 1 diabetes, ages outside the tool's scope, and highly complex conditions require dedicated assessment. It is not appropriate to enter someone with diabetes into ordinary SCORE2 and then add an empirically selected adjustment factor.
US prediction models provide a different reference framework. The PREVENT equations, incorporated into the 2026 US dyslipidemia guidelines for relevant assessment, estimate events in their own populations and settings and include cardiometabolic and renal information. Separate estimates are available for atherosclerotic events, heart failure, and overall cardiovascular disease, across different time horizons. Total PREVENT risk should not be compared with SCORE2 risk as though they were the same measure, nor should US treatment thresholds be automatically imported into European practice. Differences between models may stem from included outcomes, calibration, population, and historical period, rather than an arithmetic error in either model.
Baseline testing should provide reliable data: total cholesterol, HDL-C, triglycerides, and LDL-C, blood glucose or HbA1c as needed, creatinine with eGFR, and albuminuria when indicated. Measuring apoB may help clarify particle burden in hypertriglyceridemia, diabetes, obesity, or apparent discordance with LDL-C; however, it must not be substituted for the predictor specified in an equation. Abnormal values should be interpreted with consideration of acute illness, medications, and secondary causes. A lipid profile does not always require fasting, but very high triglycerides or specific clinical questions may necessitate fasting measurement or a different LDL-C estimation method.
Lp(a) provides additional, largely inherited and relatively stable information. Measuring it at least once in adulthood may reveal risk not evident from ordinary LDL-C levels. Its relationship with events is continuous, while thresholds are practical tools: the 2025 ESC/EAS update considers levels above 50 mg/dL, giving 105 nmol/L as the molar reference, among risk modifiers; other documents use 125 nmol/L as an operational threshold for elevation. These values do not justify a universal conversion between units. When a result is expressed in nmol/L, reference values appropriate to that unit and the selected guideline must be used, while retaining the clinical context and laboratory method.
High-sensitivity C-reactive protein may contribute to assessing inflammatory risk, but an increase during infection, trauma, or an inflammatory exacerbation must not be interpreted as a stable cardiovascular characteristic. Troponin and natriuretic peptides have their own clinical indications and are not mandatory panel tests for classifying every healthy adult. Genetic markers and polygenic scores may improve understanding of susceptibility in selected settings; their widespread use requires evidence of utility, transferability across populations, and therapeutic implications. Adding many tests does not automatically improve prediction.
When a preventive decision remains uncertain near a threshold, imaging may document the burden of accumulated disease. Coronary artery calcium may improve discrimination and reclassification, but it is not a universal screening test, and a positive score of any magnitude does not automatically equate to secondary prevention. A CAC score of zero may lower estimated risk in a selected person; it does not negate an inherited dyslipidemia, the impact of substantial exposures, or a clinical suspicion of ischemia. Statin therapy may increase calcific density even while lowering risk, so serial CAC scores should not be used to decide whether to discontinue effective treatment.
Carotid or femoral plaques may modify assessment, particularly when the result would change a treatment choice; simple intima-media thickness does not have the same significance. The ankle–brachial index is useful when symptoms, findings, or context raise suspicion of peripheral artery disease; a clinically significant diagnosis may change the task from predicting a first event to managing already established disease. Exercise ECG, coronary CT angiography, and invasive angiography should not be ordered indiscriminately to refine a risk score in asymptomatic people. A test should have a reasonable prospect of changing management; otherwise it exposes patients to incidental findings, further procedures, and costs without demonstrated benefit.
Model quality includes discrimination, the ability to rank people with different risks, and calibration, the agreement between predicted and observed events. An algorithm may discriminate reasonably well yet systematically overestimate events in another population. The individual result also depends on input accuracy: nonstandardized blood pressure, incorrect smoking status, or lipid values already modified by medication can change classification. The result must be interpreted as an estimate conditional on available information, not as certainty or a guarantee that disease is absent. Models predicting a first event have not been validated to measure the risk reduction already achieved after treatment simply by re-entering treated values.
Moving from prediction to treatment requires estimating absolute benefit. A similar relative risk reduction prevents more events when baseline risk is high. For purely illustrative purposes, a treatment reducing a 12% risk by 25% over the same interval would bring it down to 9%, an absolute reduction of 3 percentage points; applied to a 4% risk, it would lower it to 3%, a reduction of 1 percentage point. This example does not represent the guaranteed effect of any particular drug. Benefit and number needed to treat depend on the population, duration, adherence, and outcome considered; they cannot be extrapolated as universal constants.
Lifestyle interventions are appropriate across the entire risk spectrum, but intensity and support must be proportionate to need. Smoking cessation, a healthy dietary pattern, tailored physical activity, less sedentary behavior, and adiposity treatment reduce multiple exposures. Effectiveness improves when goals are concrete and measurable: tobacco use, sustainable minutes of movement, food substitutions, and program continuity. Primordial prevention seeks to avert the development of risk factors beginning in youth, whereas primary prevention intervenes before a first event in people who may already have exposures or subclinical disease.
For atherogenic lipoproteins, treatment selection considers both risk category and untreated LDL-C. European targets become more stringent with increasing risk: at very high risk, <55 mg/dL and a reduction ≥50%; at high risk, <70 mg/dL and a reduction ≥50%; at moderate risk, <100 mg/dL; and at low risk, <116 mg/dL may be considered. The lipid target and the threshold for initiating medication are distinct concepts. The 2025 update maintains this principle and recommends choosing treatment intensity according to baseline values and the necessary reduction. Statins, ezetimibe, and other therapies with documented benefit are combined based on distance from target, tolerability, and risk, avoiding both therapeutic inertia and prescribing based solely on a numerical label.
Blood pressure management requires diagnostic confirmation and assessment of target-organ damage. For the many patients to whom it applies, the 2024 ESC systolic target of 120–129 mmHg depends on tolerability and correct measurement. In diabetes, cardiovascular and renal protection is integrated with glycemic control: not all HbA1c-lowering drugs have the same effects on myocardial infarction, heart failure, or kidney disease progression. Obesity may require drug or surgical treatment according to specific indications; the benefit of one agent in a selected population must not be extended to every person with excess weight. Every intervention should address a documented determinant and be supported by evidence relevant to the clinical profile present.
Aspirin is not an automatic response to a high risk score. In primary prevention, modest ischemic benefit may be offset by bleeding, particularly with older age, previous bleeding, or concomitant medications. In secondary prevention the balance changes, but the type and duration of the strategy depend on the event, vascular territory, and any indication for anticoagulation. Risk thresholds used for statins must not be transferred to antiplatelet prescribing. Similarly, selective anti-inflammatory therapy is not indicated simply because a biomarker is elevated in someone outside the clinical setting studied in trials.
In younger people, the lifetime perspective may justify early attention even with modest 10-year risk, especially with familial hypercholesterolemia or marked, persistent exposures. In older adults, time to benefit, independence, comorbidities, and competing mortality are considered; age alone neither mandates nor excludes treatment. Shared decision-making must make untreated risk, plausible benefit, and adverse effects understandable, without using so-called “cardiovascular age” as a diagnosis or alarmist device. This measure may aid communication but does not replace a validated model or define a new biological age for the organism.
Follow-up checks effectiveness, adherence, tolerability, and the development of conditions that change the clinical picture. Blood pressure, lipids, metabolic control, smoking, and activity are reassessed at intervals appropriate to the intervention; reviewing overall risk must not be reduced to frequently repeating the same calculation. If clinical disease develops, the preventive pathway changes. When treatment lowers measured values, attainment of the original target is assessed rather than retroactively cancelling the indication. Prognosis depends on continued exposure and treatment: a favorable biochemical response is one positive component, but persistence with therapy and control of other determinants remain essential.
Risk is a probability and does not itself cause complications in the way a structural disease does. The outcomes one seeks to avoid include myocardial infarction, stroke, and cardiovascular death; some models also include heart failure. Their origins are not uniform: atherothrombosis, hypertensive damage, metabolic and renal dysfunction, and other processes contribute to different degrees. Thus, reducing atherosclerotic risk does not imply an identical reduction in every form of heart failure or arrhythmia. Prognosis must be discussed in relation to the outcomes actually measured by the model and treatment studies.
A significant error is underestimation in young people with substantial exposures, women with unrecorded determinants, and conditions poorly represented by the model. Overestimation, conversely, may encourage treatments with little benefit relative to adverse effects, especially when uncalibrated equations are applied or frailty is ignored. Communication may magnify both errors: “low risk” does not mean an event is impossible, and “high risk” does not mean an event is certain. Symptoms must be addressed through clinical diagnosis even when the estimated preventive risk is modest.
Indiscriminate investigation for subclinical disease may generate false positives, incidental findings, and unnecessary procedures. Pharmacological prevention may cause adverse effects, while overly complex regimens may impair adherence to care overall. Appropriateness therefore requires considering net benefit and avoiding test repetition without a linked clinical decision simply because a number attracts attention. Effective strategies combine correct population selection, reliable measurements, treatment of modifiable causes, and assessment of clinical outcomes. No algorithm eliminates individual uncertainty, but proper use makes decisions more consistent and transparent.
Informational notice: the information contained on this page is provided solely for informational and educational purposes and does not replace the advice, diagnosis or treatment provided by a physician. If needed, always consult a qualified healthcare professional.
Artificial intelligence transparency: this page was created with the support of artificial intelligence tools, used to assist in the production and processing of its content.