Ten-year risk prediction in French men using the Framingham coronary score: Results from the national SU.VI.MAX cohort

https://doi.org/10.1016/j.ypmed.2008.02.023Get rights and content

Abstract

Objective

To evaluate the ability of the Framingham risk function to predict the 10-year coronary heart disease (CHD) risk in French men.

Methods

3440 men, aged 45 to 60years, free of CHD at baseline, were selected from the SU.VI.MAX cohort. The expected number of event, obtained from applying the Framingham risk score to the baseline SU.VI.MAX biological and clinical data of 1994/1996, were compared to the actual risks observed in the cohort. The accuracy of the Framingham risk function was assessed using the area under the receiver operating characteristic (ROC) curve.

Results

The overall Framingham risk function predicted twice as many CHD events than observed. The area under the ROC curve for Framingham risk score was 74%.

Conclusion

The Framingham risk function may discriminate between high risk from low risk subjects, but it is not valid for estimating absolute 10-year CHD risk in this French population.

Introduction

Despite a decrease in incidence during the last decades (Immonen-Raiha et al., 1996, Lang et al., 1999, McGovern et al., 2001), coronary heart diseases (CHD) still account for a majority of deaths worldwide (Murray and Lopez, 1997, Sans et al., 1997) and are therefore an issue of primary concern in public health policies. Risk functions have been developed in order to estimate individual CHD risk depending on risk factors levels. These equations would be useful in clinical practice for determining which patients might need medical intervention or for educating patients about the necessity of risk factor reduction.

The most commonly used risk function is based on data from the Framingham cohort, which was initiated in the 1970's in an American population (Anderson et al., 1991). The Framingham equation estimates the probability of CHD within 4–12years according to age, sex, systolic blood pressure, total cholesterol/HDL-cholesterol ratio, diabetes, smoking status and electric left ventricular hypertrophy. High risk subjects have been defined as those with estimated 10-year CHD risk over 20%. However, several studies that assessed the applicability of Framingham risk model in various populations showed mixed results and its validity in low CHD rate countries is particularly questioned (Eichler et al., 2007). Few Several studies have investigated this issue in French populations (Bastuji-Garin et al., 2002, Empana et al., 2003, Laurier et al., 1994), but none with a follow-up of over 10years.

The aim of this study was to evaluate the predictive accuracy (discrimination, calibration) of the Framingham risk function in middle-aged men from a large French prospective study.

Section snippets

Study population

Subjects were participants in the SU.VI.MAX (“SUpplémentation en VItamines et Minéraux AntioXydants”) Study, a randomized primary prevention trial initially designed to test the effect of antioxidant vitamins and minerals, at nutritional doses, on the incidence of cardiovascular diseases and cancers (Hercberg et al., 1998, Hercberg et al., 2004). The 13017 subjects included were followed yearly since 1994/1995. They were invited to a medical visit each year, for either blood sampling (from

Results

Table 1 summarizes the sample characteristics. At baseline, subjects were 52.0 ± 4.7years, about 14% of men were smokers, 2.4% were diabetics and 9.3% were under antihypertensive drug treatment. There were 128 (3.7%) subjects with incident CHD event within 10years of follow-up and the majority of them were angina pectoris (n = 59) and non-fatal myocardial infarction (n = 52). Age-adjusted and multivariate-adjusted standardized HR for CHD risk in the SU.VI.MAX. population according to Framingham risk

Discussion

The present study provides additional evidence that the Framingham risk function is not directly applicable in countries where CHD incidence is low, especially in France. Despite its good discrimination, the Framingham prediction model largely overestimated absolute CHD risk, resulting in a predicted CHD events twice as high as the observed CHD events in our study population.

These results are consistent with previous studies in which the Framingham equation discrimination was shown to be

Conclusion

The present study with a well-grounded validation of coronary events is the only one to assess the validity of the Framingham equation in French populations after 10-year of follow-up (Bastuji-Garin et al., 2002, Empana et al., 2003, Laurier et al., 1994). We showed that the Framingham equation cannot be used in the French population and in low CHD rate populations in general, due to a large overestimation of absolute CHD risk. To improve CHD prediction, in addition to risk equations dedicated

Acknowledgments

The SU.VI.MAX Study received a specific funding for these analyses from the Sanofi-aventis France Laboratory.

We wish to thank Stacie Chat-Yung, MS, RD for her assistance with editing and translation in English.

References (34)

  • BrindleP. et al.

    Predictive accuracy of the Framingham coronary risk score in British men: prospective cohort study

    BMJ

    (2003)
  • ConroyR.M. et al.

    Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project

    Eur. Heart J.

    (2003)
  • CookN.R.

    Use and misuse of the receiver operating characteristic curve in risk prediction

    Circulation

    (2007)
  • Eichler, K., Puhan, M.A., Steurer, J., & Bachmann, L.M., 2007, Prediction of first coronary events with the Framingham...
  • EmpanaJ.P. et al.

    Are the Framingham and PROCAM coronary heart disease risk functions applicable to different European populations? The PRIME Study

    Eur. Heart J.

    (2003)
  • FerrarioM. et al.

    Prediction of coronary events in a low incidence population. Assessing accuracy of the CUORE Cohort Study prediction equation

    Int. J. Epidemiol.

    (2005)
  • FriedewaldW.T. et al.

    Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge

    Clin. Chem.

    (1972)

    • Cited by (14)

    • MonEco: a Novel Health Monitoring Ecosystem to Predict Respiratory and Cardiovascular Disorders

      2023, IRBM

    • A novel cardiovascular death prediction model for Chinese individuals: A prospective cohort study of 381,963 study participants

      2017, Atherosclerosis
      Citation Excerpt :

      The present study is the first to incorporate an increase in resting heart rate as variable into the CVD death risk prediction model, specifically for Chinese individuals. Increasing risk prediction models for CVD and associated hard endpoint have been developed [15–20,23–27]. In addition to conventional CVD risk factors, emerging novel biomarkers have been incorporated into the prediction models [28–33].

    • Model-observational bridging study on the effectiveness of ezetimibe on cardiovascular morbidity and mortality in France: A population-based study

      2016, Journal of Clinical Lipidology
      Citation Excerpt :

      The over-estimation of the unadjusted Framingham equation for predicting secondary CV events was made on the assumption of applicability of the equation developed for a North-American population. It is well established that predictions based on the Framingham equation overestimate by 1.75 the probability of primary CV events in France and that a consequent adjustment is systematically used in French studies but no correction is currently recommended for secondary CV events.8 This study estimated that the Framingham equation to predict repeated (secondary) CV events overestimated the real observations by 1.55, and equations were calibrated with this latest factor.

    • External validation of the 2008 Framingham cardiovascular risk equation for CHD and stroke events in a European population of middle-aged men. The PRIME study

      2013, Preventive Medicine
      Citation Excerpt :

      The 1998 Framingham risk equation has been validated in multiple ethnic American populations (D'Agostino et al., 2001). Regarding non-American populations, several studies have reported overestimation of CHD risk in European populations, while discrimination ability varied from modest (Buitrago et al., 2011; Cooper et al., 2005; Empana et al., 2003; Koller et al., 2007) to good (Ferrario et al., 2005; Hense et al., 2003; Simmons et al., 2008; Thomsen et al., 2002; Vergnaud et al., 2008). Interestingly, the 1998 Framingham equation, which does not take into account antihypertensive treatment, already revealed similar overestimation and poor discrimination for CHD risk in PRIME as in the present study (Empana et al., 2003).

    • SCORE should be preferred to Framingham to predict cardiovascular death in French population

      2009, European Journal of Cardiovascular Prevention and Rehabilitation

    • Online risk numbers – helpful, meaningless or simply wrong? Reflections on online risk calculators

      2019, Health (United Kingdom)
    View all citing articles on Scopus
    View full text
    Copyright © 2008 Published by Elsevier Inc.