The importance of total cardiovascular risk assessment in clinical practice

Abstract

Background: The Third Joint European Guidelines on Cardiovascular Disease (CVD) Prevention highlight the importance of total cardiovascular risk prediction in planning preventive strategies. Aim: To review the development of risk prediction systems in Europe and the practicalities of utilizing risk prediction in everyday clinical practice. Methods: Randomised clinical trials, cohort studies and current guidelines for the management of cardiovascular risk were reviewed. Relevant articles published between 1960 and 2004 were identified on PubMed and the Internet. Results: Several risk factors have been confirmed as major independent predictors of CVD, with their modification reducing cardiovascular risk. Total risk assessment is essential as multiple risk factors confer greater risk than the sum of their components. Framingham Heart Study data have been used as the basis for several risk prediction systems, which have been incorporated into numerous guidelines. However, these data are from a relatively small American cohort and tend to overestimate risk in some European populations. This and other limitations prompted the development of the SCORE (Systematic Coronary Risk Evaluation) risk charts using a much larger dataset from 12 European cohort studies. An electronic version of the charts, HeartScore, is now available to provide practitioners with an interactive risk prediction and management system that can be tailored to individual countries.

Conclusion: Evidence-based risk assessment systems can facilitate logical, individually tailored risk management and contribute to better patient care. Secular changes in coronary heart disease (CHD) incidence and mortality as well as in risk factors such as obesity and diabetes mean that such systems should be capable of modification and evolution.

• Cardiovascular risk
• clinical practice
• Framingham
• guidelines
• SCORE

Introduction

The publication of the 2003 European Guidelines on Cardiovascular Disease Prevention signified a landmark in public healthcare [1]. It represents a consensus between eight European professional societies: the European Society of Cardiology (ESC), the European Atherosclerosis Society, the European Society of Hypertension, the European Society of General Practice/Family Medicine, the International Society of Behavioural Medicine, the European Heart Network, the European Association for the Study of Diabetes, and the International Diabetes Federation – Europe.

The recommendations stress the importance of total cardiovascular risk prediction in planning preventive strategies. In this paper, we review the importance of total risk prediction, the development of risk prediction systems (in particular at a European level) and the practical implications involved in introducing this concept into everyday clinical practice.

Importance of total risk prediction

As early as the 1960s, the Framingham Heart Study had shown that several risk factors were major independent predictors of cardiovascular disease (CVD), namely: advancing age, cigarette smoking, hypertension (particularly systolic), elevated total cholesterol or low-density lipoprotein cholesterol (LDL-C), low-level high-density lipoprotein cholesterol (HDL-C) and diabetes [2]. Two recent independent assessments have confirmed that such risk factors are present in 80–90% of patients, regardless of population assessed [3], [4].

By the 1990s, it became clear that risk factors tend to interact, with their combined effect greater than the sum of the individual risk components (Table I, Figure 1) [5], [6]. From Table I, it can be appreciated that a 60-y-old person with a cholesterol level of 5.0 mmol/l may be at five times higher risk than a subject of the same age with a cholesterol level of 8.0 mmol/l, if the latter is normotensive, a non-smoker and female.

Figure 1.  Cumulative cardiovascular risk [6]. Risk factors: HDL-C < 0.85–0.91 mmol/l (<33–35 mg/dl); total cholesterol 5.7–6.2 mmol/l (220–239 mg/dl); SBP 150–160 mmHg; cigarette smoking (+); diabetes (+); ECG, ventricular hypertrophy. (Reprinted from Medical Clinics of North America, 79, Kannel WB & Wilson PW, An update on coronary risk factors, 951–971, Copyright (1995), with permission from Elsevier.)

Table I.  Ten-year risk of death from cardiovascular disease from the SCORE chart for high-risk European populations [5].

	Subject 1	Subject 2
Gender	Female	Male
Age (y)	60	60
Smoking status	Non-smoker	Smoker
Cholesterol level (mmol/l)	8.0	5.0
Systolic blood pressure (mmHg)	120	140
10-y risk of death (%)	2	10

In addition, the 1990s witnessed a major change in the management of CVD, with the acceptance of the evidence base that lifestyle modification and pharmacotherapy could substantially reduce risk factors and the risk of recurrent events [7–16]. Thirty years ago, observational data suggested that cessation of smoking was associated with a halving of mortality after an acute coronary syndrome [17]. Treatment of hypertension was shown to unequivocally reduce mortality [18]. Five landmark clinical trials [10], [11], [14–16] demonstrated the ability of statins to lower circulating levels of LDL-C, and showed that this was associated with substantial reductions in the risk of primary and secondary major coronary events.

More recently, the importance of type 2 diabetes and the metabolic syndrome in risk predictions has become accepted. The term “metabolic syndrome”, which may be associated with insulin resistance, refers to a clustering of cardiovascular risk factors that includes hyperglycaemia, dyslipidaemia (predominantly increased triglycerides and decreased HDL-C levels), hypertension and a procoagulant state. The presence of this clustering of risk factors predicts both an increased likelihood of developing overt diabetes and an increased risk of vascular disease of at least twofold in men and up to fourfold in women [19].

These studies indicate both the importance of assessing total risk and the fact that risk factor reduction is highly effective in reducing mortality. Moreover, without the ability to estimate total risk quickly, easily and reasonably reliably, fundamental clinical management errors result in either under- or overuse of medication.

Development of risk prediction systems

The risk factors highlighted by the Framingham Heart Study subsequently served as the basis for several risk prediction systems, using both tables and charts [20–23]. Despite such advances, globally CHD remains the leading cause of mortality [24]. Two surveys, EUROASPIRE I and II, conducted in 15 European countries, clearly showed that CHD prevention in daily practice is inadequate. The surveys found that up to 85% of physicians reported utilizing risk assessment/guidelines; however, one-third of patients were obese, more than half had high blood pressure or high serum cholesterol, and one in five coronary patients were active smokers [25], [26].

These data illustrate the considerable potential to further reduce CHD morbidity and mortality and, together with new data from studies such as the Heart Protection Study (HPS) and Anglo-Scandinavian Cardiac Outcomes Trial (ASCOT) [27], [28], provided the impetus to develop the Third Joint European Societies’ Task Force guidelines for CVD prevention [1]. These guidelines recommend charts for total risk estimation, based on the results of the Systemic Coronary Risk Evaluation (SCORE) project [5]. An overview of both the Framingham and SCORE risk charts is provided, the differences between them examined, and the potential implications for clinical practice highlighted.

The Framingham study

The Framingham Heart Study was undertaken by the United States Public Health Service in order to investigate possible factors, both biological and environmental, that might explain the epidemic of cardiovascular disease observed in the United States during the 1930s [29], [30]. Altogether, 5209 healthy residents, with over 2800 women, between the ages of 30 and 60 were recruited into this landmark study in 1948 from the town of Framingham, Massachusetts, in the USA. In particular, the Framingham Heart Study established a strong association between LDL-C and CHD, and also the protective effect of HDL-C. The Framingham Heart Study also established the role of blood pressure in the development of CVD, and dispelled the myths that blood pressure may be less harmful in women and that the elderly may tolerate higher blood pressures. Smoking was also found to be associated with increased risk of myocardial infarction (MI) in the Framingham population, and the risk increased with the number of cigarettes smoked. The fact that filters in cigarettes gave no protection for CHD was also another finding. The Framingham Heart Study is an ongoing study recruiting the children (Framingham Offspring Study) and grandchildren (Third Generation Study) of the original cohort.

The Framingham risk charts

One of the most well-known risk prediction systems is based on the Framingham Heart Study. The Third Adult Treatment Panel from the National Cholesterol Education Program [31] from the USA suggests risk estimation using an updated Framingham risk score based on risk factor categories (Figure 2) [32]. The updated Framingham risk equation uses “hard” CHD as its endpoint, which includes MI and CHD death. The previous Framingham risk equation, which was also used by the older European guidelines, the Second Joint Task Force for CVD Prevention in Clinical Practice, uses “total” CHD, which includes stable angina, unstable angina as well as MI and CHD death. It is generally felt that the “conversion” from the Framingham “hard” CHD risk score to the Framingham “total” CHD risk score is roughly two-thirds to three-quarters depending on which risk category an individual is in.

Figure 2.  The Framingham risk charts [31]. To assess CHD risk in people without clinically manifest CHD, with multiple (>2) risk factors, use the Framingham risk scoring system to assign point values for age, total cholesterol, HDL-C, systolic BP and smoking status. The total number of points derived from scores in the five categories corresponds to a specific 10-y CHD risk.

The limitations of Framingham

The development of the Framingham risk charts represented a major and ground-breaking advance in integrated risk prediction. However, they were based on a relatively small number of individuals from a small town. Most were Caucasian and from a similar socio-economic class. In certain populations in both Europe [33] and the USA [34], where the average CHD risk was similar to the Framingham cohort, reasonable risk estimates are possible with the Framingham risk charts. However, studies have shown that the underlying prevalence of major risk factors differs significantly from Framingham in populations with a greater diversity of age [35] or ethnic composition [34], [36–39]. Furthermore, when the American data have been applied to many European populations, including Italy [40], Denmark [41] and Germany [42], the Framingham risk charts tended to overestimate the risk of an event. While the major risk factors account for most of the difference in risk between populations, other variables such as diet, psychosocial factors and genetic makeup may be important in determining absolute risk within populations.

Paper-based risk charts such as those based on the Framingham and SCORE projects are limited in the number of variables that can be displayed. Thus, it is difficult to accommodate the effect of obesity, exercise and family history. The role of newer risk factors, such as plasma levels of lipoprotein(a), homocysteine [43–45], coagulation factors (e.g. fibrinogen and plasminogen activator inhibitor) [46], [47], circulating adhesion molecules (e.g. VCAM-1, ICAM-1) [48], [49] as well as markers of inflammation (especially C-reactive protein, CRP) [50] and oxidative stress, in determining absolute risk is still being established. Electronic, interactive risk prediction systems such as HeartScore (described below) allow more flexibility in displaying the effects of multiple risk factors.

A further limitation of Framingham is that the definition of some non-fatal endpoints used in the study differs from that used in other cohort studies and clinical trials. It is, therefore, difficult to validate Framingham or to relate its findings to the results of other studies. The limitations of Framingham are summarized in Box 1. This is not to gainsay the visionary and pioneering role of the Framingham investigators in establishing the multifactorial aetiology of atherosclerotic vascular disease and in laying the foundations upon which all risk prediction systems are based.

Box 1. The limitations of Framingham

• Mainly North American, Caucasian participants of certain socio-economic class

• Therefore, applicability to different ethnic and socio-economic groups is uncertain

• Does not incorporate all risk factors

• Some endpoint definitions differ from those used in other studies, and the choice of endpoints has changed over time

The SCORE project

SCORE (Systemic Coronary Risk Evaluation) was initiated in 1994, at the suggestion of the ESC, to address some of the limitations of existing risk prediction systems. The SCORE investigators utilized 12 mainly population-based cohort studies (see Box 2 for participating countries and patient numbers) [5].

Box 2. SCORE: participating countries and patient numbers [5]

Table

Belgium 10 641 Denmark 9945 Finland 37 296 France 7337 Germany 3968 Italy 53 439

Norway 48 425 Russia 3325 Spain 4701 Sweden 7435 UK 7292 UK – Scotland 12 285

The pooled dataset involved 205 178 asymptomatic individuals (88 080 women and 117 098 men) with no evidence of pre-existing CVD, representing 2.7 million years of follow-up. The study was designed to evaluate 10-y risk, with total CVD mortality (CHD, peripheral arterial disease and ischaemic stroke) as the primary endpoint, and to capture regional variations in risk. A total of 7934 cardiovascular deaths, 5652 from CHD, were recorded.

The SCORE risk charts

Based on the study findings, new risk charts were developed (Figures 3 and 4, tables are only available on-line) [5], [20]. The SCORE risk charts include the following risk factors: systolic blood pressure, gender, smoking status, age (40–65 y), and either total cholesterol or total cholesterol/HDL-C ratio. Diabetes was not incorporated as there was no uniform collection of data. Risk predictions for diabetic patients should be judged to be at least twice as high in men and four times higher in women compared with the charts. As the endpoint was CVD death, high risk was redefined in the SCORE project as a 10-y risk of ≥ 5%. There is, however, no single level of risk defining the threshold for intervention, since risk increases in a progressive and graded way; the higher the risk, the greater the yield from risk factor modification.

Figure 3.  The new risk charts based on SCORE data—in low-risk regions—based on (A) total cholesterol (URL for downloadable SCORE charts for cholesterol: http://www.escardio.org/initiatives/prevention/SCORE+Risk+Charts.htm) or (B) total cholesterol/HDL-C ratio [5]. (Reprinted from European Heart Journal, 24, Conroy RM, Pyorala K, Fitzgerald AP, Sans S, Menotti A, De Backer G, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project, 987–1003, Copyright (2003), with permission from the European Society of Cardiology.) Qualifiers (20) Note that total CVD risk may be higher than indicated in the chart: • As the person approaches the next age category • In asymptomatic subjects with pre-clinical evidence of atherosclerosis (e.g. CT scan, ultrasonography) • In subjects with a strong family history of premature CVD • In subjects with low HDL-C levels, raised triglyceride levels, impaired glucose tolerance, or with raised levels of CRP, fibrinogen, homocysteine, apolipoprotein B or lipoprotein(a) • In obese and sedentary subjects

Figure 4.  The new risk charts based on SCORE data—high-risk regions—based on (A) total cholesterol, or (B) total cholesterol/HDL-C ratio [5]. (Reprinted from European Heart Journal, 24, Conroy RM, Pyorala K, Fitzgerald AP, Sans S, Menotti A, De Backer G, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project, 987–1003, Copyright (2003), with permission from the European Society of Cardiology.) Qualifiers (20) Note that total CVD risk may be higher than indicated in the chart: • As the person approaches the next age category • In asymptomatic subjects with pre-clinical evidence of atherosclerosis (e.g. CT scan, ultrasonography) • In subjects with a strong family history of premature CVD • In subjects with low HDL-C levels, raised triglyceride levels, impaired glucose tolerance, or with raised levels of CRP, fibrinogen, homocysteine, apolipoprotein B or lipoprotein(a) • In obese and sedentary subjects

SCORE: redefining risk assessment

The SCORE risk charts represent an advance on those based on Framingham as they utilize a large dataset, across a wide range of countries. This made it possible to develop charts for both low-risk regions (e.g. Belgium, France, Greece, Italy, Luxembourg, Portugal, Spain and Switzerland; Figure 3, tables are only available on-line) and high-risk regions (all other European countries and Russia; Figure 4, tables are only available on-line). It is also possible to recalibrate the charts for individual countries if up-to-date mortality data and risk factor distributions are available. This process also adjusts for secular changes in mortality that may have occurred since the original cohorts were studied; the existing charts may over-predict risk in countries with a falling CVD mortality, and under-predict if mortality is rising. Country-specific charts have been produced for Sweden, Germany, Spain, the Czech Republic and Greece, and others are in preparation.

The SCORE risk charts are based on total atherosclerotic CVD mortality, rather than CHD in isolation, which provides a more realistic estimate of the burden of atherosclerotic risk. The charts do not presently predict morbidity or total events, although a high risk of death automatically indicates a higher risk of non-fatal or total events.

Inspection of the risk charts would suggest that the HDL-C/total cholesterol ratio does not offer any better prediction than total cholesterol alone. This point requires further examination, as low HDL-C may be particularly relevant in subjects with multiple risk factors. It is likely that the inclusion of total cholesterol and HDL-C may improve risk prediction.

The key advantages and limitations of the SCORE risk charts are summarized in Boxes 3 and 4, and the situations when CVD risk may be higher than indicated in the charts are presented in Figures 3 and 4, [5].

Box 3. Key advantages of the new SCORE risk charts

• Based on a large dataset, across a range of countries

• Provide risk equations for high- and low-risk regions---enables regional risk to be estimated. National risk charts are in development

• Primary endpoint is total CVD mortality---provides patients and healthcare providers with a robust and reproducible estimate of risk

• Display 10-y risk as values as well as categories---emphasizes that there is no single level of risk defining the threshold for intervention

• Allow visualization of the lifetime impact of risk factors and relative risk to facilitate counselling of younger persons at high relative but low absolute risk

Box 4. Key limitations of the new SCORE risk charts

• Any paper chart is limited in the variables which can be included

• Non-fatal and total events are not included

• No separate charts for diabetes or metabolic syndrome

• Many new risk factors are not included

Recognition of the advantages of the SCORE risk charts is their adoption as a risk prediction system by the Third Joint European Task Force in their guidelines for the prevention of CVD in clinical practice [1]. In addition, an electronic, fully interactive version of SCORE, called HeartScore, has been produced (http://www.escardio.org/knowledge/decision_tools/heartscore/). HeartScore includes two European versions, based on the four SCORE risk charts (high and low risk, total cholesterol and total cholesterol/HDL-C ratio) and the European CVD guidelines. It provides easily understood graphic representations of risk and the expected effect of intervention, which can be used both for making management decisions and for counselling patients.

The future of SCORE: meeting the practical challenges

The HeartScore program is flexible and can be updated as new cohort studies become available, and can incorporate new languages, new risk factors and new endpoints as knowledge evolves, providing physicians with an efficient and accurate risk assessment tool. Future SCORE projects will include newer cohorts to allow the incorporation of newer risk markers, in particular components associated with insulin resistance and diabetes. Recent re-examination of the SCORE dataset has indicated that the impact of self-reported diabetes on risk can be evaluated and that the impact may be greater than that suggested by Framingham data.

The evaluation of CVD morbidity and hence total events is a particular issue, and one that has been requested by many physicians. Factors limiting the study of morbidity include changing definitions for angina, unstable angina, MI, acute coronary syndrome, heart failure and shock. Improvements in diagnostic tests have contributed to the changes in definitions. For example, the evolution of creatine kinase and troponin testing has confirmed that there is a spectrum between angina and MI with gradually increasing tissue necrosis. The natural history of CHD has changed as a result of such improvements in diagnostic techniques, and also due to societal lifestyle changes in nutrition and smoking, and the development of effective therapies. A further factor limiting the study of morbidity is ascertainment bias. The accuracy and determination to record endpoints varies from study to study. More recent drug trials probably make more rigorous attempts to ascertain all endpoints.

Arising out of these issues is the question of what is the “multiplier” to convert mortality to morbidity. Depending upon decisions regarding the above variables, this might be anywhere between 1.5 and 10. At present, when superimposing the current SCORE charts on the older European Framingham-based charts, the conversion rate from coronary events (fatal and non-fatal) to atherosclerotic CVD mortality is about four to one. Thus, the high risk figure of 20% (for a CHD event, fatal or non fatal) in the 1998 charts converts to a high risk figure of 5% (for CVD death) in the new mortality charts.

Future developments in risk prediction using the SCORE system include the examination of the effect of self-reported diabetes, the impact of a history of prior MI, a more detailed examination of HDL-C and risk, the study of body mass index and family history and risk, the further development and translation of local risk prediction systems, and steps to integrate newer risk markers. HeartScore has much potential to develop into a more sophisticated risk evaluation, management and audit tool.

The Fourth Joint European Task Force on the Prevention of Cardiovascular Disease in Clinical Practice has recently begun its work under this author's chairmanship. Dialogue with SCORE and HeartScore will continue in a coordinated effort to make CVD risk evaluation and management more accessible to busy health professionals.

Conclusion

The management of CVD has advanced tremendously in the last half-century, with the recognition of risk factors, the development of effective therapies—in particular, statins—and the demonstration that treatment of risk factors can substantially reduce morbidity and mortality. To aid the process, evidence-based risk assessment systems, such as the Framingham risk charts, have been developed and incorporated into management guidelines. In Western Europe, the USA, Australia and Canada, these advancements have been associated with a decline in the rate of CVD mortality.

However, despite such advances, worldwide, CVD remains the primary cause of death, and there remain some challenges to effective management. EUROASPIRE I and II clearly demonstrated that not all risk factors are being treated and, with increasing rates of obesity, type 2 diabetes and the metabolic syndrome—conditions characterized by the presence of multiple CVD risk factors—the assessment of total risk becomes ever more important.

The development of the evidence-based SCORE risk charts, which have been adopted by the Third Joint European Societies’ Task Force guidelines on CVD prevention in clinical practice, should aid practitioners in assessing total risk and making risk predictions that are tailored to their individual countries. By assessing total CVD mortality risk, practitioners can prioritize high-risk patients and more accurately tailor therapeutic regimens for individuals. In addition, HeartScore allows quick and easy risk estimation and the potential benefits of intervention, with advice that can be tailored to each patient. This should further aid daily practice by also providing appropriate and up-to-date management advice.