Implications of a newer Framingham model

Dr Bosomworth’s integration of risk assessment and clinical practice guideline recommendations into a tool that generates patient-specific numbers needed to treat¹ has the potential to bridge an important gap in clinical decision making. The practicality is clearly appreciated, as evidenced by responses published in this journal in July 2011.^2,3 It is important to identify why, as one response noted, use of this tool might “increase ... prescription of statin drugs.”³ The Framingham general cardiovascular disease 10-year risk model (FRS-CVD), use of which was recommended in the 2009 Canadian dyslipidemia guidelines,⁴ provides a risk estimate that incorporates a larger and more pathophysiologically diverse number of events. In addition to estimating the risk of “soft” and “hard” coronary artery disease (CAD) events (CAD death, myocardial infarction, coronary insufficiency, angina), it also incorporates the risk of cerebrovascular events (ischemic stroke, hemorrhagic stroke, transient ischemic attack), peripheral artery disease (intermittent claudication), and heart failure. Earlier Canadian dyslipidemia guidelines⁵ advocated the use of the Framingham hard CAD 10-year risk model (FRS-CAD), which estimated only “hard” coronary events (CAD death, myocardial infarction).

For most patients, their estimated risk is greater using FRS-CVD than it is using FRS-CAD.⁶ For example, in the case study that Dr Bosomworth presents, the 10-year risk using FRS-CVD is approximately 14%, while using FRS-CAD the risk estimate is 8%. In a small cohort study conducted in Ontario, the 2009 Canadian dyslipidemia guidelines’ advocacy of FRS-CVD rather than FRS-CAD was shown to increase the number of patients recommended for lipid-lowering therapy by 2.3-fold.⁷ In a cross-sectional analysis conducted in the United States, the use of FRS-CVD rather than FRS-CAD was shown to significantly diminish the low-risk category for both men and women.⁶ If use of the FRS-CVD is adopted by upcoming US dyslipidemia guidelines, the investigators of the US analysis anticipate the effect to be profound and one that warrants “close economic and disease management evaluation.”⁶ In addition, because statins have not been shown to be beneficial in reducing the risk of all of the cardiovascular end points comprising the FRS-CVD risk estimate, numbers needed to treat derived from these risk estimates will for most patients inflate treatment benefit further (in addition to the extrapolation to a 10-year time period). For example, statins do not reduce the risk of hemorrhagic stroke; rather, a nonsignificant increase in risk was documented in a recent meta-analysis.⁸ As it relates to clinical decision making surrounding a particular drug therapy, a risk assessment tool might be informative if it identifies a risk shown to be reduced by the intervention. In this regard, estimates of benefit extrapolated from the earlier FRS-CAD risk model would at least be more consistent with the statin evidence base in the setting of primary prevention.

• Copyright© the College of Family Physicians of Canada