SIGN 97: Risk estimation and the prevention of cardiovascular disease

Guideline Index Page | SIGN Methodology

2 Cardiovascular risk

2.1 Risk factors

The INTERHEART study assessed the importance of risk factors for coronary artery disease worldwide.19 Nine measured and potentially modifiable risk factors, accounted for more than 90% of the proportion of the risk for acute myocardial infarction. Smoking, history of hypertension or diabetes, waist hip ratio, dietary pattern, physical activity, alcohol consumption, blood apolipoproteins and psychosocial factors were identified as the key risk factors. The effect of these risk factors was consistent in men and women across different geographic regions and by ethnic group. The British Regional Heart Study also found that smoking, blood pressure and cholesterol accounted for 90% of attributable risk of CHD.20 Evidence level 2++

Worldwide, the two most important modifiable cardiovascular risk factors are smoking and abnormal lipids. Hypertension, diabetes, psychosocial factors and abdominal obesity are the next most important but their relative effects vary in different regions of the world.

2.2 The concept of risk and why it matters

Most cardiovascular deaths will occur in individuals at moderate risk as they constitute the largest group. High risk individuals will have the most to gain from risk factor modification and historically are given the highest priority in clinical practice.21

When estimating risk, total CVD mortality, rather than CHD endpoints should be used to encompass stroke prevention as well as CHD prevention. Stroke deaths are underestimated using traditional CHD endpoints.4 Current risk prediction systems do not predict accurately the different risk profiles that exist in different ethnic groupings and cultures. A risk score derived from Caucasian cohorts may substantially overpredict the risk in a Chinese population.22 CVD risk prediction based on absolute risk is now advocated for treatment decisions for aspirin, statins, antihypertensives and in people with atrial fibrillation, for warfarin.

2.2.1 Predicting risk

Intervention studies have shown that while relative risk reduction may remain broadly constant, absolute risk reduction varies considerably because it is a function of the initial level of baseline risk. Consider the example in Table 1 of a man with a baseline risk of a cardiovascular event of 10% over ten years who takes effective preventative treatment and lifestyle measures. His relative risk falls by a third, while his absolute risk is reduced to 6.7%, an absolute risk reduction of 3.3%. If another man with a higher baseline risk of 30% takes the same effective treatments his relative risk also falls by about a third to 20%. However, his absolute risk reduction is 10%. Relative risk reductions in CHD events in the statin trials appear similar regardless of baseline risk and baseline cholesterol (except where baseline cholesterol is<5 mmol/l when relative risk reduction is less.3,23 This would support the concept that the best way to target patients is to calculate absolute risk.

Table 1: Example illustrating absolute and relative risk reductions

Baseline ten year CVD risk Relative risk reduction Post-treatment ten year CVD risk Absolute risk reduction
10% 33% 6.7% 3.3%
30% 33% 20% 10%

Overprediction of CVD risk means that people with little to gain potentially become patients and are exposed to the questionable benefits and risks of lifelong treatment. Underprediction means that people with much to gain may not be offered preventative treatment. The best way to target patients for risk reducing interventions is to calculate absolute risk.

2.3 Risk Scoring Systems

2.3.1 Framingham based scoring systems

A large number of risk scoring systems for CHD and CVD have been devised for use in clinical practice, the majority of which are based on the American Framingham study.24,25 The Framingham equations are the most widely accepted method for projecting cardiovascular disease/coronary disease risks, and are used in the British, European and New Zealand guidelines.

These risk scoring systems are reliable in ranking individual CHD and CVD risks within populations, based on conventional risk factors, but have been shown to give a variable performance when predicting actual events within populations.15 Framingham risk equations are based on event rates which occurred in a predominately white, United States population during the 1970s. CHD rates have been declining in the US and many other countries, resulting in a tendency for the event rates predicted by Framingham?based scores to be higher than actual event rates in populations.

Framingham-based scoring systems tend to overestimate risk in low and medium risk groups and underestimate risk for certain subgroups including British Asians; people with Type 1 diabetes; people with Type 2 diabetes with nephropathy; those with familial hypercholesterolaemia; those with a strong family history of premature CHD; those with left ventricular hypertrophy on electrocardiography; and those with chronic renal disease.26 Framingham significantly underpredicted CHD risk in a Scottish male general population cohort (Renfrew and Paisley) which is explained partly by the high CHD mortality rates in this population. A Framingham- based risk score also underestimated the true level of CVD and CHD risk in men with lower socioeconomic status whether this was assessed using social class or a postcode-based deprivation score.16

These results were tested in an analysis commissioned by SIGN based on the Scottish Heart Health Extended Cohort (SHHEC), involving 6,419 men and 6,618 women aged 30-74 years from 25 local government districts in Scotland, for whom baseline data were collected between 1984 and 1995.17 While the Framingham score overestimated the actual observed CHD risk in the cohort as a whole, it seriously underestimated the large gradient in risk by socioeconomic status, particularly in women. Application of the score as a basis for preventive treatment would result in relative undertreatment of the most socially deprived, compared with the least deprived, potentially exacerbating social disparities in disease rates.

While risk scores are superior to clinical assessment alone, they can be misleading when used to guide treatment decisions among people at different levels of social deprivation or of different ethnic backgrounds. Without correction, such scores may foster the relative undertreatment of the socially deprived.18

In order to reduce the deprivation-related difference in the numbers eligible for preventive treatment, risk scoring systems need to adjust for deprivation, as the ASSIGN (ASsessing cardiovascular risk using SIGN guidelines to ASSIGN preventive treatment) score has been developed to do (see section 2.3.4).

2.3.2 Using scoring systems in practice

Basing treatment decisions on predetermined levels of a risk score replaces potentially arbitrary decisions with transparency, consistency and potential for audit. It may maximise efficient use of limited resources and implies fairness in ensuring equitable distribution. Determining by score those whose condition warrants treatment eliminates many possible sources of bias.

No clear information is available on how GPs are using risk scoring systems in Scotland. This makes it difficult to predict the effect of introducing a new system. The proposed introduction of ASSIGN offers the prospect of an improved understanding of how general practice manages risk and allows an opportunity to evaluate resource usage and the effectiveness of preventative interventions.

A central aim of this guideline is to ensure that the scoring system recommended to identify high risk, is as accurate as possible and that the treatments suggested are appropriate and in line with scientific evidence. Risk scoring systems are important tools but are limited by changes in disease and population patterns.

2. 3. 3 The joint British Cardiac Hypertension and Hyperlipidaemia Societies

The British Cardiac, Hypertension, and Hyperlipidaemia Societies (JBS) have jointly provided modified charts of graded risk which are valid for use in primary care as their diagnostic accuracy is unaffected by approximations in age and blood pressure.27 The JBS guidelines were updated in December 2005 with the publication of JBS2.28 The revised guidelines include all atherosclerotic CVD (acute coronary syndromes, exertional angina, cerebrovascular disease and peripheral arterial disease), rather than CHD alone. In JBS2 the definition of high risk has been lowered from a 30% or greater ten year risk of CHD (equivalent to over 40% ten year CVD risk) to a =20% CVD risk over ten years. JBS2 emphasises that individuals with any symptomatic manifestation of CVD, including diabetes, are assumed to be at high risk of cardiovascular events and do not require formal risk estimation. Evidence level 4

This assumption of high risk means that many more individuals (around 635,000 in Scotland) will fall into the high risk category. If implemented widely, this definition of high risk will have a significant impact on health professional workload and resource expenditure (see NHSQIS CVD Clinical and Resource Impact Assessment Report).29 It may also result in unnecessary treatment for many individuals. Evidence level 4

The evidence supporting the decision of JBS2 to reduce the intervention threshold for high CVD risk over ten years is not clear. Where evidence is lacking, thresholds are often determined by balancing workload against projected medium to long term costs. Existing guidelines which specify treatment thresholds for statin prescribing have been influenced by the costs of these drugs and have tended to set intervention levels relatively high with respect to the actual risks observed in those with coronary heart disease.27 Any risk assessment method that demonstrates a low specificity and high false positive rate will necessarily cause an inevitable increase in total prescribing costs.30

2.3.4 ASSIGN

The ASSIGN score (ASsessing cardiovascular risk using SIGN guidelines to ASSIGN preventive treatment) has been developed to include social deprivation as a risk factor. The inclusion of family history provides an indirect approach to ethnic susceptibility.

ASSIGN is based on the Scottish Heart Health Extended Cohort, a series of population studies from the 1980s and 1990s followed up until the end of 2005. The Scottish Heart Health study recruited men and women across 25 districts of Scotland in 1984-87 and the Scottish MONICA Project recruited in Edinburgh and Glasgow in 1986 and in Glasgow alone in 1989, 1992 and 1995.31

ASSIGN uses similar classic risk factors to Framingham, entered as continuous variables rather than categories. It includes the SIMD (Scottish Index of Multiple Deprivation) score for residential postcode. It also includes family history of cardiovascular disease, defined as coronary disease or stroke in parents or siblings below age 60 or in several close relatives. Like Framingham it does not include obesity as a risk factor; unlike Framingham it excludes left ventricular hypertrophy as a risk factor.

Results from ASSIGN are similar to those from the Framingham cardiovascular score in many respects but the overall estimation of ten year cardiovascular risk is rather lower, consistent with some overestimation in the Framingham score.18

ASSIGN tends to classify more people with a positive family history and who are socially deprived as being at high risk. When used in its own host population it abolished a large social gradient infutureCVDvictimsnotidentifiedforpreventivetreatmentbytheFraminghamcardiovascular score. It therefore improved social equity, although overall discrimination of future events was not greatly improved.18

A demonstration of the ASSIGN tool is available at http://assign-score.com

2.4 What is meant by high risk?

There are considerable variations in the definitions of the categories of risk. Both JBS228 and the current European guidelines in CVD prevention32 include patients with established coronary heart disease, peripheral arterial disease or cerebrovascular arterial sclerotic disease or diabetes in their definitions of high risk. The European guidelines are based on assessments of a symptomatic patients. In the European guidelines high risk is calculated as a ten year risk of 5% or greater for developing a fatal CVD event. The JBS2 guideline defines high risk as at least 20% risk of developing a first cardiovascular event over ten years.

In the great majority of cases, an individual's risk is the product of multiple risk factors and there is a need for an absolute risk estimation to be made for individuals believed to be at risk who have not presented as high risk by the presence of established disease.

The main debate around what constitutes high risk relates to the vast majority of the asymptomatic population who have no history of CVD or diabetes. The onset of statins has raised fundamental questions about the risk and prevention of CVD. The cardiovascular benefit of treatment with a statin is observed among people with annual levels of risk as low as 1%33 and the annual CHD risk may be nearing 1% in the US and in Northern European countries. In this scenario, most middle aged men and women could benefit from a statin and CVD risk reduction.34 The long term safety profile of statin therapy in relatively healthy adults has not yet been established.

section 3>

Guideline Index Page | SIGN Methodology

Scottish Intercollegiate Guidelines Network, Healthcare Improvement Scotland, Gyle Square, 1 South Gyle Crescent, Edinburgh EH12 9EB
Tel. 0131 623 4720 Web contact duncan.service@nhs.net
Last modified 13/09/13 © SIGN 2001-2013

Risk estimation and the prevention of CHD <Guidelines <Home