Policy-making bodies increasingly agree that the most efficient and effective clinical CHD prevention requires a global assessment of CHD risk.5,13
Fortunately, a variety of user-friendly tools based on the Framingham equation are available to help clinicians perform CHD risk assessment for patients with no known history of cardiovascular disease. Our review suggests that, in general, the categorical charts and tables derived from the Framingham equation are accurate and feasible for use in clinical settings and can be used in lieu of the continuous Framingham calculators when necessitated by the clinical environment. This supports findings by chart developers who report similar discriminatory ability between their charts and the full Framingham equations.20
Some features of the computer or PDA-based tools, however, may make them a better choice for providers with access to such devices.
In deciding among available tools, providers may wish to choose tools that provide risk information in a format that can be used with current guidelines for risk reduction (see ). For instance, to allow risk-based decision making about lipid-lowering therapy, providers need a tool that allows stratification of risk into <10%, 10%−20%, and >20%.1,37
All of the spreadsheets, PDA, and web-based calculators have this capability because all use the continuous Framingham equations or the original Framingham categorical charts. Many of the risk charts also have this capability; the notable exception is the Modified Sheffield table, which uses only 15% and 30% cutoffs. To adhere to evidence-based guidelines on aspirin use, providers need a tool with finer gradations of risk because the risk/benefit ratio for aspirin use transitions from helpful to harmful at a 10-year risk of CHD events between 3% to 5% and 10%.2,5
This again reduces the number of useful risk charts, but still allows many acceptable options. At present, it is unclear how providers should address risk calculation in patients with diabetes. The National Cholesterol Education Program and the American Heart Association currently recommend that physicians treat patients with diabetes as though they have a risk for subsequent CHD events that is equivalent to that in patients with known CHD.1,13
In accordance with this, they have recommended that their Framingham risk calculators be used only in patients without diabetes. At present, however, we are unaware of direct evidence that suggests this strategy is more effective than relying on calculated risk assessment, and many calculators continue to request input of diabetes status for risk calculations.
Current Guidelines for Cardiovascular Risk Reduction
In choosing which risk tool to use, providers should consider their practice environment and who will be performing the risk assessments. Providers who have access to a computer with an available spreadsheet program or dedicated high-speed Internet access line should consider spreadsheet and web-based programs for risk calculation. These tools allow calculation of fine gradations of risk, frequently provide comparisons to individuals with low risk (e.g., BMJ Cardiorisk Manager, Birmingham Heartlands Calculator, National Cholesterol Education Program Risk Calculator, RiskCalculator from the Center for Cardiovascular Sciences at the University of Edinburgh, Healing Hearts Risk Calculator, Medical-decisions.com calculator, and American Heart Association Calculator), and, in some cases, provide targeted advice on treatment and allow exploration of the effects of treatment on calculated risk (e.g. BMJ Cardiorisk Manager, Birmingham Heartlands Calculator, Heart-to-Heart Calculator). Additionally, at least one of these tools (Heart-to-Heart Calculator) is targeted to patients and can be used independently of the clinician visit. For providers who do not have access to these tools, current PDA tools and risk charts offer an acceptable option.
Some providers may find that a combination of products is most useful, particularly if the outcome of interest varies according to patient concerns. Most tools provide information on the combined risk of stable and unstable angina, myocardial infarction, and CHD death. Some tools, however, report only the risk of myocardial infarction and CHD death; these tools will produce smaller numeric estimates of risk than tools that also include angina. The current NCEP risk calculator, for example, uses a set of newly revised Framingham equations that only predict the risk of myocardial infarction and CHD death. To our knowledge, these equations have not been published in the peer-reviewed literature. Other tools allow calculation of all CVD events by adding stroke outcomes (e.g., New Zealand Risk Table, British Cardiac Risk Assessor, BMJ Cardiorisk Manager, Risk Calculator from the Center for Cardiovascular Sciences at the University of Edinburgh) or by allowing independent calculation of the risk of stroke and peripheral vascular disease (e.g., Birmingham Heartlands Calculator).
In addition to choosing which type of risk tool to use, providers must ensure that they have sufficient information to complete the risk assessment. Some information, such as age, smoking status, and presence or absence of diabetes, can be obtained by interview at the time the risk calculation is performed. Other information, such as blood pressure, cholesterol levels, and presence or absence of left ventricular hypertrophy on electrocardiogram must be obtained prior to risk calculation.
Our review identified several limitations among the available Framingham tools. First, existing tools do not predict risk beyond 12 years. This is a limitation imposed by the published data available from the Framingham Heart Study. Although Framingham investigators have published data on the lifetime risk of developing coronary heart disease,38
they have not incorporated lifetime risk into tools for clinical risk estimation. Presentation of lifetime risks may have different effects on perceived threat and motivation to undertake risk-reducing behavior for some patients, particularly younger ones, who are making longer-term prevention decisions,38–41
although to date this has not been empirically studied. Second, none of the tools specify how electrocardiographic LVH is to be defined, although available evidence suggests that LVH with repolarization abnormality (strain pattern) provides the best predictive ability, and LVH by voltage criteria alone is not associated with clearly increased risk.42
Third, none of the tools provide confidence intervals around risk estimates. Their absence may convey a false sense of precision. Finally, most tools do not provide accurate information about the benefits and adverse effects of risk-lowering interventions, which may limit their clinical utility.
Aside from the limitations of the tools, we acknowledge the limitations of the Framingham equations themselves. Although the Framingham equations predict the degree of risk well in white and African-American men and women between the ages of 30 and 65 in the United States, they predict the degree of risk less well in non-U.S. populations, certain U.S. ethnic groups (Japanese men, Hispanic men, and Native-American women), men and women younger than age 30 or older than age 65, and diabetic persons.25,27,28
One approach to the Framingham equations' limits is to recalibrate the tool for use in designated target populations.28
At present, we are not aware of any Framingham-based risk calculation tools that have attempted to do this.
The current Framingham equations have additionally attempted to balance accuracy and feasibility43
and hence have limited the number of risk factors required for risk estimation. They do not include the following established and potential risk factors, which may be of interest: blood glucose level, hemoglobin A1C, triglycerides, lipoprotein A, small dense low-density lipoprotein particles, homocysteine, c-reactive protein, microalbuminuria, coagulation factors, weight or body mass index, physical activity, and family history of premature cardiovascular disease. The effect of adding additional risk factors to risk calculation tools has been little studied.
As of April 2003, our searches of the medical literature also show that the effect of risk calculators on clinical practice and outcomes has not been well studied. Two studies6,44
suggest that providing physicians with computerized risk calculators has had little impact on CHD risk. These studies, however, provided no link to evidence-based guidelines and had important methodological limitations including high attrition rates44
and use in populations who already have existing CHD.6
A third study, in which researchers alternately wrote patient risk scores on the front of patient charts or not, also suggests the limited effects of providing physicians with only risk estimates.45
Whether calculating and communicating global CHD risk to patients affects their willingness or ability to change their lifestyle and use preventive medications, such as aspirin, antihypertensive drugs, or cholesterol-lowering medication, has not been well studied. Although a recent pilot study46
testing the combined effects of a self-guided workbook and physician visit on global CHD risk reported that 68% of users planned to make a variety of interventions on their risk as a result of using the book, traditional CHD risk appraisal has had only modest impact on actual patient behavior in the areas of diet and exercise.47–49
One recent study has shown reductions in CHD risk, body mass index, and cholesterol levels at 5 years follow-up in intervention groups that received CHD risk appraisal with or without physician consultation,50
but conclusions were limited by high attrition rates and poor participation in follow-up consultations throughout most of the study. Further research is still needed.
Research should also determine whether the inclusion of newer risk factors for CHD (i.e., lipoprotein a, homocysteine, micro-albuminuria, or c-reactive protein), or noninvasive measures of atherosclerosis, such as electron-beam computerized tomography (EBCT) or carotid Doppler ultrasound, improves risk assessment and leads to better use of CHD risk-reducing treatments. Some have suggested that these novel risk factors may be best used to modify the pretest probability estimate from the Framingham risk score, particularly for those with intermediate risk.43