Summary of National Heart, Lung, and
Blood Institute Workshop on Cardiovascular Risk Assessment
Scott M. Grundy, MD, PhD, Ralph B. D'Agostino, Sr.,
PhD, Lori Mosca, MD, PhD, MPH, Gregory L. Burke, MD, Peter W. F. Wilson, MD,
Daniel J. Rader, MD, James I. Cleeman, MD, Edward J. Roccella, PhD, Jeffrey A.
Cutler, MD, Lawrence M. Friedman, MD From: The Center for Human
Nutrition, University of Texas Southwestern Medical Center at Dallas, Dallas ,
TX (SMG), Framingham Heart Study, Boston University, Boston, MA (RBD, PWFW),
Preventive Cardiology, Columbia and Cornell Universities, New York, NY (LM),
Wake Forest School of Medicine, Winston-Salem, NC (GLB), Preventive Cardiology
and Lipid Research Center, University of Pennsylvania Health System,
Philadelphia, PA (DJR), and the Office of Prevention, Education, and Control
(JIC, EJR), Division of Epidemiology and Clinical Applications (JAC), and
Office of the Director (LAF), National Heart, Lung, and Blood Institute,
Bethesda, MD. Correspondence to: Jeffrey A. Cutler,
M.D. National Heart, Lung, and Blood Institute 6701 Rockledge Drive, RM
8130 Bethesda, Maryland 20892-7936 Tele: (301) 435-0414 Fax: (301)
480-1773 Email:
CUTLERJ@NHLBI.NIH.GOV
Abstract
Background: The National Heart, Lung, and
Blood Institute conducted a Workshop in January 1999 to assess the
applicability to other U.S. populations of coronary heart disease (CHD) risk
prediction algorithms generated from the Framingham Heart Study (FHS). This
report presents major findings from the workshop, including consideration of
applications of risk assessment in practice. Methods and
Results: Longitudinal cohorts were identified for testing the accuracy of
the FHS function. The function--based on age and categories of blood pressure,
total cholesterol, HDL cholesterol, smoking and diabetic status--was applied
(separately by gender and race) to each of the other cohorts. Accuracy of
5-year predictions of non-fatal myocardial infarction or CHD death were
compared to those using functions developed with the study's own data. Other
than in the older subjects in one cohort, agreement between FHS-based
predictions and observed results for relative risk associated with each risk
factor was good, except that hypertension was a somewhat stronger predictor in
black subjects, especially women. Discrimination between cases and non-cases
based on the FHS function as a whole was also satisfactory, but generally not
quite as good as the study's own functions. For three cohorts, the FHS function
over-predicted absolute CHD risk and some recalibration of the function would
be required for optimal use. Conclusions: From a quantitative
viewpoint, the applicability of the FHS risk algorithm using traditional risk
factors appears satisfactory for most populations. The Workshop also identified
unresolved issues with regard to 1) further development of risk assessment
tools, 2) effects on physician and patient behavior, and 3) the role of global
risk assessment in clinical guidelines. Keywords: coronary disease,
epidemiology, prevention, risk factors
The complete article is available
as a PDF file [800K] This is an expanded version of the article first
published in Circulation, 2001;104:491-496.
Information
for visitors using screen readers
Information about PDF
|