The research team, led by Dr Peter Brindle, a Bristol University researcher and city GP, also found that 84% of the coronary heart disease deaths that actually occurred during the study period, were in men predicted to be at low risk.
Doctors base the decision to offer risk-reducing treatment, such as cholesterol and blood pressure-lowering drugs, on their patient's coronary 'risk'. If it is higher than it is in reality, it makes it difficult for both patients and doctors to make an informed choice about starting life-long treatment that may have side effects, requires regular monitoring and causes unnecessary anxiety.
Current methods for assessing this risk are based on data collected between 1968-74 from Framingham, a small town in the north east of the USA. The relevance of the 'Framingham score' to the British population is uncertain.
Dr Brindle and colleagues from London University, tested the accuracy of the Framingham coronary risk predictions in 6,643 men from the British regional heart study. They calculated each man's coronary risk and compared these predictions with the number of actual coronary events that occurred over ten years.
During this ten-year period only 2.8% of men died from coronary heart disease, compared with the 4.1% predicted by the Framingham score - an overestimation of 47%. For fatal or non-fatal coronary heart disease events, such as a heart attack or angina, 10.2% occurred, compared with the 16% predicted - an overestimation of 57%.
The authors also found that 84% of the coronary heart disease deaths occurred in the 93% of men classified as being at low risk (i.e. those that had a less than 30% chance of having a coronary event within 10 years). According to current guidelines this means they would be unlikely to have been offered cholesterol-lowering treatment.
These findings have considerable implications for the prevention of coronary heart disease. The National Service Framework for coronary heart disease recommends the use of Framingham-based methods to identify high-risk people, and most people attending their GP for unrelated reasons are subject to this opportunistic screening.
Dr Brindle said: 'While we acknowledge that performing a coronary risk assessment using the Framingham score, which takes into account up to eight different risk factors, is better than using purely clinical judgement, it is important that GPs and their patients understand the limits of the accuracy of the Framingham coronary risk score.'