This site is intended for health professionals only

At the heart of general practice since 1960

Calculating cardiovascular disease risk

Recommendations on cardiovascular risk assessment are under review. Dr Anthony Wierzbicki, Professor Tim Reynolds and Dr Rubin Minhas consider the implications for GPs

Recommendations on cardiovascular risk assessment are under review. Dr Anthony Wierzbicki, Professor Tim Reynolds and Dr Rubin Minhas consider the implications for GPs

Which patients should be treated because they are at high risk of cardiovascular disease? It is a common question for GPs and one where they may reach for a risk calculator, such as the one in the BNF.

But there are questions about the accuracy of the calculations used to assess risk. The development of atherosclerosis is complex, driven as it is by the interaction of many risk factors. It is perhaps not surprising that the estimation of the likelihood of cardiovascular disease by risk calculators is not a simple matter.

Of the dozens of variables that are statistically correlated with cardiovascular disease, most are not robustly independent and so are of limited use in calculating risk. Epidemiological studies and sophisticated statistical analysis have now reduced the risk factors used in most calculators to a handful of major risk factors, with individual calculators using some additional risk factors (see box left), some identified from population studies and some added as adjustments to help stratify risk.

Methods of assessing cardiovascular risk are a policy as well as a scientific problem. Officials from the Department of Health are currently considering proposals from the National Screening Committee on the options for a risk assessment programme (see box overleaf).

The clinical controversy is whether adjustments to calculators are valid and how are they made. Nearly all the calculators agree on the major but differ in the lesser risk factors, suggesting the minor risk factors make a relatively small contribution to the overall population risk. The problem is that for some groups and individuals triglycerides, family history, obesity or ethnicity – which do not feature in many risk calculations – may play a major role in cardiovascular disease development, illustrating the problem of relying on population-based equations to predict individual risk.

Risk assessment is not an exact science. All CVD risk estimates have wide confidence intervals because of daily biological variation. A 20% CHD risk has a 95% confidence interval of 14-26%. Multiple measurements can refine the accuracy of measurement, but not much. Measurements in triplicate measures only reduce the estimate to 16-24%. The statistical methods used in the calculations mean that risk estimates are actually even less accurate than this.

Use of risk calculators
Risk calculators also assume that their equations are valid for all populations. This is not true. The Framingham calculator, for example, overestimates risk in patients from southern European countries, but appears to be reasonably accurate for northern European populations. When it was applied to the West of Scotland Coronary Outcomes Project (WOSCOPS) trial population it predicted the number of patients developing CHD events with great accuracy.

The Framingham risk score also underestimates risk in patients with diabetes (because of the small number of patients with diabetes in the original study and the limited therapy they received at that date). More specific calculators, such as the one derived from the UK Prospective Diabetes Study, are more appropriate for patients with diabetes.

The Joint British Guidelines (JBS2) argue that the CVD risk is so high in diabetes that all patients should be treated anyway. Other guidelines disagree. The solution may be to include blood glucose or glycated haemoglobin in risk calculators, which would take account of the excess risk from insulin resistance and the metabolic syndrome. This would also deal with some of the excess risk seen in insulin-resistant ethnic populations, including people from south Asia, Africa and Turkey.

Other problems arise from seemingly trivial adjustments to the risk calculation pathway. Introducing adjustments for hypertension, obesity and ethnicity can all result in exaggeration of predicted risk. Sometimes these adjustments seem to have a spurious precision. Risk calculations that differentiate between Bengalis and Gujaratis are more likely to reflect differences in the small population samples used to derive the data, rather than an innate difference in genetic predisposition to risk.

Finally, and most significantly, age is the strongest driver of cardiovascular risk. As treatment thresholds are reset ever lower (a 10-year 20% risk of developing cardiovascular disease and a 15% risk of developing coronary heart disease is the threshold in the JBS2 guidelines), the number of people who qualify for medical treatment increases. As the population ages, more and more patients will achieve risk levels above the treatment threshold. Given the limitations to current methods of risk assessment, there is now interest in alternative methods of identifying higher-risk individuals within populations at overall moderate risk.

These may be biochemical such as markers of inflammation including C-reactive protein (CRP) and lipoprotein associated phospholipase A2 (LpPLA2), or measures of vascular function such as asymmetric dimethylarginine. Other methods include physiological measures associated with atherosclerosis (endothelial dysfunction) or measurement of athero-sclerosis by direct imaging.

A number of techniques are being investigated including measures of endothelial function and carotid intima-media thickness, and coronary calcium scores that can predict risk better than conventional risk calculators. Their validity is being investigated in large-scale prospective studies but none are yet recommended for routine use.

CV risk factors
Major risk factors:
• Age
• Sex
• Smoking status
• Diabetes status
• Total: HDL cholesterol ratio
• Systolic blood pressure

Used by individual risk calculators:
• ECG left ventricular hypertrophy (Framingham)
• Triglycerides (PROCAM)
• Microalbuminuria (UKPDS for diabetes)
• Family history of coronary heart disease (NSF)
• Ethnicity (such as Asian) (NSF)
• Social deprivation index (SIGN)

Screening for vascular risk
• The National Screening Committee (NSC) has drafted guidance on vascular risk assessment
• The Department of Health Vascular Programme Board is reviewing NSC recommendations
• Options under consideration are– self-assessment of risk through the Health Check programme– record-based assessment to identify people at highest risk who are not receiving comprehensive risk advice and management– primary care population-based risk assessment using the primary care populations to offer those who are not at highest risk, as identified by record-based assessment, the opportunity of risk assessment

Practical Implications
The practical implication of the limitations of CVD risk assessment is that clinical judgement is still required. Atherosclerosis is a chronic disease and once decisions are made about treatment then life-long drug therapy is recommended.
• Decisions on drug therapy can be postponed until multiple measurements have been made over a period of time and lifestyle advice has been fully implemented.
• Certain groups – those with atherosclerosis, those with familial hyperlipidaemias and patients with type 2 diabetes – require automatic treatment because of their high current or predicted future risk.
• The minimum number of adjustments should be made to any risk estimate.
• Assessments of CVD risk should be interpreted in the light of a wide confidence interval.• If in doubt, assess and treat the patient, not the number, and consult with a local expert if unsure.

Risk calculators
• Framingham risk score: www.nhlbi.nih.gov/about/framingham/riskabs.htm
• PROCAM risk score : www.chd-taskforce.com
• UKPDS risk engine :www.dtu.ox.ac.uk and select 'risk engine'
• Joint British Guidelines on cardiovascular prevention :heart.bmj.com/cgi/content/extract/91/suppl_5/v1

Dr Rubin Minhas is a GP in Medway, Kent. He is chair of the Cardiovascular Working Group of the South Asian Health Foundation and a member of the NICE Technology Appraisal Committee
Professor Tim Reynolds is consultant chemical pathologist at Queen's Hospital, Burton-on-Trent, Staffordshire
Dr Anthony Wierzbicki is consultant chemical pathologist and director of the lipid unit at Guy's and St Thomas' Hospitals, London

Competing interests Dr Wierzbicki has received grant support, lecture honoraria and travel grants from Abbott, Fournier-Solvay, GlaxoSmithKline, LifeCycle Pharma, Merck kGA, Merck-Sharp & Dohme, Pfizer, Sanofi-Aventis and Takeda pharmaceuticals

Rate this article 

Click to rate

  • 1 star out of 5
  • 2 stars out of 5
  • 3 stars out of 5
  • 4 stars out of 5
  • 5 stars out of 5

0 out of 5 stars

Have your say