Study Suggests Model Used for Cholesterol Guidelines May Lead to Misclassification of Risk for Heart Attack and Coronary Death
- Compared with the original Framingham model used to calculate the risk of myocardial infarction or coronary death, the point-based version of the tool misclassifies millions of Americans into different risk groups, with 25-46% of affected individuals experiencing potential impacts on drug treatment recommendations for cholesterol control.
- Compared with the original model, the point-based system misclassified 15% of adults into different risk groups, disproportionately shifting patients into higher risk groups: 10% (representing 3.9 million people) were misclassified into higher-risk groups, while 5% (representing 1.8 million people) were misclassified into lower-risk groups.
- Patterns of misclassification varied significantly by gender, age, and underlying CHD risk. For example, among those misclassified, 64% of men and 80% of women were shifted into higher-risk groups by the point-based system. Misclassification affected 7% of people ages 20-44 years, 17% ages 45-64 years, and 27% ages 65-79 years.
National cholesterol guidelines use the “Framingham model” to calculate a person’s 10-year risk of myocardial infarction or coronary death. Based on this risk, patients are categorized into different risk groups, which are used to guide treatment decisions. Both the original and point-based versions of the Framingham model are endorsed by the National Cholesterol Education Project’s Adult Treatment Panel III (ATP III) guidelines. Given that approximately 36 million persons in the U.S. are eligible for lipid-lowering therapy, differences in risk classification depending on which model is used could result nin millions receiving different lipid-lowering therapy. This study compared differences in predicted risk between the original and point-based Framingham calculations. Using nationally representative data from 2001 to 2006 for 2,543 adults (representing 39 million adults ages 20-79), investigators calculated the 10-year risk of major coronary events using each model, and then determined whether differences in risk estimates would place individuals into different ATP-III risk groups.
- Estimates of how many subjects would have recommended changes in lipid-lowering therapy based on misclassification are approximate due to limited sample sizes, absence of data on potential lifestyle interventions, and potential inaccuracies in self-reported use of lipid-lowering medications.
- Investigators did not have access to actual cardiovascular outcomes, and so were unable to determine the accuracy of these models for predicting cardiovascular events.
The next generation of cholesterol guidelines (ATP IV) is expected to be released in the near future, and will likely predict risk using a new model incorporating a broader range of cardiovascular outcomes. The score-sheet versions of this model have already been developed, and if applied to guidelines, the authors suggest may result in problems similar to those observed in this study.
Dr. Steinman was supported by an HSR&D Career Development Award; he is part of HSR&D’s Program to Improve Care for Veterans with Complex Comorbid Conditions, San Francisco.
Gordon W, Polansky J, Boscardin W, Fung K, and Steinman M. Coronary Risk Assessment by Point-Based vs. Equation-Based Framingham Models: Significant Implications for Clinical Care. Journal of General Internal Medicine November 2010;25(11):1145-51. E-pub Sep 8, 2010.