The applicability of Framingham data to Britain has been questioned. Dr Matthew Lockyer discusses the implications

I was interested to read the news story in the last issue ‘Framingham coronary risk score overestimates risk in British men’ (see News, Guidelines in Practice, December 2003).

A couple of years ago, while on holiday in New England, I spotted ‘Framingham, Mass.’ on a signpost. To the annoyance of my family, I made a detour to drive through this home of one of the longest epidemiological studies ever undertaken.

For more than 50 years the inhabitants of this town have been studied, supplying the world with detailed data on cardiovascular risk factors and outcomes. Disappointingly, it was no different from any other American town, with shopping malls and fast food outlets dominating the view.

It occurred to me then that the American lifestyle is significantly different from ours. The recently published paper in the BMJ by Brindle et al. (Br Med J 2003; 327: 1267-70) on which your news story is based shows how important this may be. The cardiovascular risk in healthy individuals is calculated by an equation that weights various risk factors. The absolute risk is calculated in many models, including the Sheffield tables, from Framingham data. The paper suggests that the Framingham models may give a significant overestimation of cardiovascular risk for British populations.

Outcome data from long-term studies of British men were compared with the predicted outcomes from Framingham models. There was a relative overestimation of risk across all levels, of 47% for deaths from CHD and 57% for CHD events.

This means that the greatest exaggeration of absolute risk occurs in the groups at highest risk. The paper suggests not only that current models may be exaggerating cardiovascular risk, but also that most events occur in individuals whose risk is below 30%, the threshold at which the NSF for CHD recommends lipid lowering treatment. The implication is that the cardiovascular risk calculation lacks sensitivity, failing to detect those who will benefit from lipid lowering therapy.

We GPs are likely to find ourselves at the sharp end of this problem, facing questions from patients already on treatment and those needing advice on primary prevention.

I believe we should concentrate on good secondary prevention and targeting the known high-risk groups, such as patients with diabetes, while we await clarification of the significance of this paper. There is still evidence that secondary prevention could be improved.

Many patients eligible for primary prevention could have their risk decreased by other interventions, such as smoking cessation and effective management of hypertension.

Dr Matthew Lockyer, GP, Suffolk

Guidelines in Practice, January 2004, Volume 7(1)
© 2004 MGP Ltd
further information | subscribe