This article is more than 5 years old. Information may no longer be current.
Commonly used clinical risk calculators overestimate CV events
Four commonly used risk scores, including the newer American College of Cardiology/American Heart Association atherosclerotic CVD risk score, may overestimate CV events by up to 67% in women and 154% in men, according to a report published in Annals of Internal Medicine.
Researchers evaluated the calibration and discrimination of the ACC/AHA atherosclerotic CVD (ASCVD) risk score released in 2013 compared with four other risk-prediction tools in a cohort of 4,227 participants in the MESA study aged 50 to 74 years who were followed for more than 10 years. All participants were free of CV symptoms and had no history of MI or stroke at baseline.
“Overestimation dominated and was noted throughout the continuum of risk for these tools, including the intermediate risk groups in which treatment decisions balancing therapeutic risk and benefit are more challenging,” Andrew P. DeFilippis, MD, MSc, from the division of cardiovascular medicine at University of Louisville, and colleagues wrote.
Estimation of risk
Use of the original Framingham Risk Score (FRS), which incorporates age, sex, diabetes, BP, cholesterol and smoking to estimate 10-year risk for angina, MI or CHD-related death, overestimated the risk for CV endpoints including MI, angina, cardiac arrest, stroke and CV-related death by 53% in men and 48% in women. Use of the Adult Treatment Panel III risk score (ATP-III, 2001), which includes a modified FRS and thresholds and goals for lipid-lowering therapy based on 10-year CHD risk, overestimated risk by 154% in men and 46% in women. Use of an updated FRS (2008), which includes other common ASCVD events, such as stroke, to predict CVD, overestimated risk by 37% in men and 8% in women. Use of the ACC/AHA ASCVD risk score, which utilizes risk factors in the original FRS and includes separate equations for white and black individuals to guide therapy, overestimated risk by 86% in men and 67% in women.
However, use of the Reynolds Risk Score (2007), which includes family history of premature CHD and high-sensitivity C-reactive protein in addition to standard risk factors, overestimated risk by 9% in men and underestimated risk by 21% in women.
In the group with a predicted risk score of 7.5% to 10% using the ACC/AHA ASCVD risk score, the threshold at which statin initiation is recommended, the acute rate of CV events was 3% in men and 5% in women.
The overestimation was not explained by aspirin, lipid-lowering therapy or antihypertensive therapy, or interim revascularization, according to the researchers.
Roger S. Blumenthal
DeFilippis and colleagues concluded that these findings highlight the importance of individualized risk assessment that includes additional variables, such as other medical conditions, family history, physical activity and CRP.
“Additional testing could be a much-needed tiebreaker in the all too common ‘to treat or not to treat’ dilemmas. Such testing should be considered in all patients with marginal risk scores — those in whom the decision to treat with long-term statin and aspirin remains unclear,” Roger S. Blumenthal, MD, professor of medicine and director of the Johns Hopkins Ciccarone Center for the Prevention of Heart Disease and Cardiology Today Section Editor, said in a press release.
Although not examined in the current analysis, the researchers said the risk overestimation may be related to the fact that calculators, including the newest one released in 2013, use risk reference data obtained decades ago.
Michael Blaha
“The less-than-ideal predictive accuracy of these calculators may be a manifestation of the changing face of heart disease. Cardiac risk profiles have evolved in recent years with fewer people smoking, more people having early preventive treatment and fewer people having heart attacks or having them at an older age,” Michael Blaha, MD, MPH, director of clinical research at the Johns Hopkins Ciccarone Center for the Prevention of Heart Disease, said in the release.
Physician response to overestimation
In an accompanying editorial, Paul M. Ridker, MD, MPH, and Nancy R. Cook, ScD, from Brigham and Women’s Hospital and Harvard Medical School, noted that the researchers “have demonstrated clearly the complexity involved in thoughtful development and external validation of cardiovascular prediction algorithms.”
Ridker and Cook questioned the impact on physicians using the ACC/AHA ASCVD risk score to determine statin use.
Paul M. Ridker
“Clinicians might recalibrate the algorithm so that it tracks more closely with contemporary evidence, simultaneously calculate multiple risk algorithms as currently done in some Mayo Clinic prevention programs, or elect to ignore the problem and accept that more persons will be treated with a class of drugs proven to reduce vascular event rates. Physicians might also consider including revascularization procedures as an endpoint because they are relevant to our patients, are expensive and are part of the trial endpoints used for efficacy evaluation,” they wrote. – by Katie Kalvaitis
Disclosures: Blaha reports receiving personal fees from Pfizer outside of this work. Blumenthal and Cook report no relevant financial disclosures. DeFilippis reports consulting for Roche and grants/payment for lectures from AstraZeneca. Ridker reports receiving grants from Amgen, AstraZeneca, Novartis and Pfizer and other financial relationships with AstraZeneca and Siemens; he is also listed as a co-investor on patents held by Brigham and Women’s Hospital related to the use of inflammatory biomarkers in CVD and diabetes that have been licensed to AstraZeneca and Siemens.