- Email: [email protected]
A Convenient Tool to Profile Patients for Generalized Cardiovascular Disease Risk in Primary Care
A Convenient Tool to Profile Patients for Generalized Cardiovascular Disease Risk in Primary Care S. Rasika Wickramasinghe, PhDa,*, Andrew P. DeFilippis, MD, MSca, Donald M. Lloyd-Jones, MD, ScMb, and Roger S. Blumenthal, MDa The early identification of and intervention in patients with increased risk factors for generalized cardiovascular disease greatly reduces their long-term mortality from hard coronary artery disease (CAD) events and other related co-morbidities. Thus, a number of multivariate risk factor analyses based on large-scale population studies have led to various risk-scoring models aimed at screening for those at high risk for CAD. These assessment systems, by virtue of novel diagnostic markers and better population data, have become increasingly adept at accurately predicting CAD risk in individual patients. Nevertheless, their practical application in the setting of primary care has lagged, because of a reluctance of many primary care physicians to adopt these methods. An effective risk assessment system should not only encompass the risk for hard CAD events but also include risks for related co-morbid conditions, such as stroke and heart failure, and be simple and accurate enough to be efficiently used in primary care clinics. In conclusion, in an attempt to simplify 1 of the more effective risk assessment devices for generalized cardiovascular disease risk that meets all these requirements, the investigators strongly support the model proposed by D’Agostino and colleagues and provide here a commentary on its utility in identifying patients at high risk for cardiovascular disease. © 2009 Published by Elsevier Inc. (Am J Cardiol 2009;103:1174 –1177)
Asymptomatic patients with risk factors for cardiovascular disease (CVD) do not come directly to the attention of cardiologists; they are first seen by primary care physicians in the setting of routine office visits or unexpectedly, while being assessed for other illnesses. Thus, the availability of rapid screening tools for primary care physicians to quickly and reliably stratify patients for CVD risk during office visits can enhance the detection of patients at increased risk for CVD. In the February 12, 2008, issue of Circulation, D’Agostino et al1 published the results of a study using participants of the Framingham Heart Study and the Framingham Offspring Study. In these analyses, multivariate risk assessment tools were refined to allow office-based physicians to rapidly and easily identify high-risk candidates for several different CVD outcomes, including global CVD events, as well as subtypes of coronary artery disease (CAD), cerebrovascular disease, peripheral arterial disease, and heart failure. The utility of this tool resides in its ability to provide absolute risk estimates for all clinically relevant forms of CVD in patients and to do so using data points and criteria that can be readily collected in the office setting. We believe that this risk assessment tool can be conveniently adopted in the primary care setting in the form of a risk scorecard that can be used by physicians and nursing staff a The Johns Hopkins Ciccarone Center for the Prevention of Heart Disease, Division of Cardiology, The Johns Hopkins University School of Medicine, Baltimore, Maryland; and bDepartment of Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois. Manuscript received December 3, 2008; revised manuscript received and accepted December 21, 2008. *Corresponding author: Tel: 443-799-8498; fax: 410-614-9190. E-mail address: [email protected] (S.R. Wickramasinghe).
0002-9149/09/$ – see front matter © 2009 Published by Elsevier Inc. doi:10.1016/j.amjcard.2008.12.040
members as part of routine patient evaluation, or even by patients in the waiting room before their visits. Use of Traditional Cardiovascular Disease Risk Assessment Tools Multivariate risk stratification for the assessment of CVD risk has been repeatedly used across multiple studies. For instance, the data from the Framingham Heart Study formulation for “hard” CAD events (myocardial infarction and CAD death) have been incorporated into the third report of the Expert Panel of Detection, Evaluation, and Treatment of Cholesterol in Adults (Adult Treatment Panel [ATP] III).2 Similarly, other countries have either integrated adaptations of the Framingham data or have designed their own risk assessment systems on the basis of CVD incidence statistics and risk factors prevalent in their specific populations: the Systematic Coronary Risk Evaluation (SCORE) project in Europe and the QRESEARCH cardiovascular RISK algorithm (QRISK) in the United Kingdom.3–5 The QRISK algorithm in particular uses traditional risk factors for CVD (age, systolic blood pressure, smoking, and ratio of total serum cholesterol to high-density lipoprotein cholesterol) but also incorporates family history, antihypertensive treatment, and body mass index.6 In addition, recent evidence suggests that coronary calcium scoring provides clear incremental predictive information on CAD risk beyond that of the Framingham risk estimate across multiple ethnic and racial groups.7 Given the high specificity of coronary calcium scoring in the prediction of CVD events, these data argue that noninvasive diagnostic modalities are best used in conjunction with traditional risk assessment systems, when there is uncerwww.AJConline.org
Editorial/Profiling Patients for Generalized CVD Risk
1175
Table 1 Cardiovascular disease risk and heart age scorecard based on office-based parameters Points
Age (yrs) M
⫺3 ⫺2 ⫺1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ⬎20
F
30–34
30–34
35–39
35–39
40–44 45–49 50–54 55–59 60–64 65–69 70–74 ⬎75
40–44 45–49
BMI (kg/m2) M
⬍25 25–30 ⬎30
F
⬍25 25–30 ⬎30
SBP Untreated (mm Hg)
SBP Treated (mm Hg)
Smoker?
Diabetic?
M
M
M
M
120–129 130–139 140–159 ⬎160
F
120–129 130–139 140–149 150–159 ⬎160
⬍120 120–129 130–139 140–150 ⬎160
F
⬍120
120–129 130–139 140–149 150–159 ⬎160
50–54 55–59 60–64 65–69 70–74 ⬎75
F
No
No
Yes
Yes
No
F
No
Yes Yes
CVD Risk
Heart Age (yrs)
M
F
M
F
1% 1% 2% 2% 3% 3% 4% 5% 6% 7% 8% 10% 11% 13% 15% 18% 22% 25% 29% ⬎30% ⬎30% ⬎30% ⬎30% ⬎30% ⬎30%
⬍1% ⬍1% 1% 1% 2% 2% 2% 3% 3% 4% 4% 5% 5% 6% 7% 8% 10% 11% 13% 15% 18% 20% 24% 27% ⬎30%
⬍30 ⬍30 31 33 35 37 39 41 44 46 49 52 55 58 62 65 69 73 78 80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
⬍30 ⬍30 ⬍30 32 34 36 38 41 43 46 48 51 54 58 61 65 69 73 77 80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
This table enables physicians to quickly and accurately calculate generalized CVD risk on the basis of D’Agostino et al’s1 formula, using simple, office-based parameters. BMI ⫽ body mass index; SBP ⫽ systolic blood pressure.
tainty regarding initiation of therapy, or when there is concern that the Framingham score may not be capturing a patient’s risk adequately.8 One such example is a middleaged patient with borderline or elevated risk factors (who will have a very low Framingham 10-year risk estimate despite risk factors, simply because of age) who has a family history of premature CVD. Given that the specificity of coronary calcium scoring translates into a high positive predictive value only when the population being screened has a moderate or high underlying disease incidence, officebased screening tools may allow the selection of a population more likely to benefit from more involved diagnostics such as coronary calcium scoring. Thus, at present, with the prohibitive cost of coronary calcium screening for the entire population, and possible long-term risks that are largely unknown, we believe that it is advantageous for physicians to routinely use rapid officebased screening methods that are accurate and convenient. These office-based risk assessment tools can then be used as initial screening devices to identify intermediate- or highrisk patient populations, who can then be selectively subjected to atherosclerosis imaging procedures with greater degrees of specificity and positive predictive values for predicting CVD risk. Nevertheless, it is important to remember that screening tools such as those derived from Framingham are not infallible. Traditional risk scoring systems are effective and discriminate risk well when applied
to large populations, but these should not be substituted for good clinical judgment at the individual patient level. Many previous risk assessment tools based on United States populations, including the standard Framingham risk score, assess the risk for CAD to the exclusion of other CVD processes, such as stroke, heart failure, and intermittent claudication.9 These other disease processes share common risk factors with CAD. In terms of physician-patient communication, it may be more meaningful to a patient to hear that his or her risk for a major CVD event is 15% over the next 10 years rather than a 5% risk for a hard CAD event over the next decade. In this context, the availability of a tool that can quickly and accurately calculate the risk for generalized CVD (not just CAD) in patients using simple criteria should become an important instrument in adult primary care. Despite the availability and demonstrated benefit of many risk prediction algorithms, their use in primary care has been quite limited, and the use of risk scores will not translate into better patient outcomes unless they are used appropriately by physicians.10 A number of potential reasons exist for the apparent reluctance of primary care physicians to adopt such risk assessment tools. Foremost among them is the fact that some risk assessment tools focus on predictive accuracy at the cost of simplicity, often involving long and tenuous risk score systems that require significant effort. In addition, most risk assessment systems necessitate laboratory testing results that may not
1176
The American Journal of Cardiology (www.AJConline.org)
Table 2 Cardiovascular disease risk and heart age scorecard based on laboratory-based parameters Points
⫺3 ⫺2 ⫺1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ⬎20
Age (yrs)
HDL Cholesterol (mg/dl)
M
M
F
F
Total Cholesterol (mg/dl) M
F
SBP Untreated (mm Hg) M
F
SBP Treated (mm Hg)
Smoker? Diabetic?
CVD Risk
M
M
M
F
F
M
F
⬍120 ⬎60 ⬎60 ⬍120 50–59 50–59 ⬍120 30–34 30–34 45–49 45–49 ⬍160 ⬍160 120–129 120–129 ⬍120 No No No No 35–44 35–44 160–199 160–199 130–139 130–139 35–39 35–39 ⬍35 ⬍35 200–239 140–159 140–149 120–129 120–129 240–279 200–239 ⬎160 130–139 130–139 Yes 40–44 ⬎280 240–279 150–159 140–159 Yes Yes Yes 40–44 45–49 ⬎280 ⬎160 ⬎160 140–149 45–49 150–159 50–54 ⬎160 50–54 55–59 60–64 55–59 65–69 60–64 70–74 65–69 ⬎75 70–74 ⬎75
F
⬍1% ⬍1% 1% ⬍1% 1% 1% 1% 1% 2% 1% 2% 2% 3% 2% 3% 2% 4% 3% 5% 3% 6% 4% 7% 4% 8% 5% 9% 6% 11% 7% 13% 9% 15% 10% 18% 12% 21% 14% 25% 16% 29% 18% ⬎30% 22% ⬎30% 25% ⬎30% 29% ⬎30% ⬎30%
Heart Age (yrs) M
F
⬍30 ⬍30 ⬍30 30 32 34 36 38 40 42 45 48 51 54 57 60 64 68 72 76 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
⬍30 ⬍30 ⬍30 30 31 34 36 39 42 45 48 51 55 59 64 68 73 79 80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
This table enables the calculation of generalized CVD risk on the basis of D’Agostino et al’s1 formula, using laboratory values such as total cholesterol and HDL cholesterol. HDL ⫽ high-density lipoprotein; other abbreviation as in Table 1.
be available during routine clinic visits. We believe that a risk score must be simple to calculate and based on readily available patient data to be used appropriately. The new Framingham CVD risk score system meets those criteria. D’Agostino and Colleagues’ Risk Assessment System The general CVD risk prediction model of D’Agostino et al1 is an important advance in patient screening. Using original and offspring cohorts of the Framingham study, the investigators collected CVD risk factor data to generate gender-specific Cox proportional-hazards regressions, relating risk factors to the first incidence of CVD during a follow-up period of 12 years.1 Using these models, a mathematical function that predicts general CVD risk in the next 10 years was constructed. The investigators then presented this risk prediction model in the form of a risk score system, similar to the risk assessment system published by the National Cholesterol Education Program ATP III.1,2 The new Framingham general cardiovascular risk profile has several features that make it attractive. First, it uses 2 separate risk-scoring methods based either on data collected during a general office visit (age, body mass index, blood pressure, treatment of hypertension, smoking, and diabetes) or on data incorporating laboratory studies (including highdensity lipoprotein and total cholesterol). These assessment tools have discrimination statistics in the range of 0.75 to
0.79 for the prediction of generalized CVD risk, which is considered to be “very good” discrimination between cases and noncases.11 This risk assessment tool also incorporates the concept of “heart age.” In this concept, the CVD risk of a candidate is transformed into the age of a subject without any risk factors, but with the same CVD risk. Heart age may not be as useful to physicians in the diagnostic or therapeutic process, but can greatly affect patients’ impressions of their CVD risk. A 45-year-old man with significant risk factors will likely be better able to appreciate the notion that his heart has the age equivalent of a typical 70-year-old man than the quantitative concept that he carries a 20% CVD risk over the next 10 years. Unfortunately, current treatment guidelines are not based on D’Agostino et al’s1 scoring system, and simply applying their score in place of the Framingham or ATP III risk score to guide treatment, assuming the same thresholds are used as in ATP III, may result in an overuse of aspirin and lipid-lowering therapy on the basis of current treatment guidelines.12 For example, ATP III treatment recommendations are based on risk estimates of hard CAD. Because the newer risk equations published by D’Agostino et al1 incorporate all CVD end points, the risk estimates will be substantially higher than for hard CAD alone. Therefore, the use of this comprehensive CVD risk tool may potentially
Editorial/Profiling Patients for Generalized CVD Risk
lead to the overtreatment of patients; in contrast, statins and aspirin are generally well tolerated and may impart a reasonable risk/benefit ratio in subjects with a 10% risk for major cardiovascular events over the next decade. Interestingly, global CVD risk is always greater than the risk for a component of CVD, such as CAD or stroke. The risks for these individual components (CAD, heart failure, stroke) can also be calculated using this model by multiplying generalized CVD risk in D’Agostino et al’s1 scoring system by a calibration factor specific to the component of interest. Because of these powerful features and ease of use, we reproduce D’Agostino et al’s1 risk assessment tools into a single table that can be conveniently carried by physicians in their coat pockets and used with ease in an office setting (Table 1). In its consolidated form, each column of the table is divided by gender and represents a single risk factor: age, body mass index, systolic blood pressure treated, systolic blood pressure untreated, smoking, and diabetes. The last 2 columns represent cumulative CVD risk and heart age. To quickly estimate a patient’s 10-year total CVD risk, a physician need only read the score in the far-left column for each risk factor and then, after calculating the total score, read the cumulative CVD risk and heart age from the last 2 columns for the total score. This gives a reliable measure of a patient’s risk for CVD and, within a matter of minutes, can identify patients who require further diagnostic workup on the basis of physicians’ estimates of acceptable 10-year risk. A similar table that incorporates laboratory data for the assessment of generalized CVD risk can be designed (Table 2). A logical advance to this assessment tool will be the inclusion of other newer criteria that improve the predictive power of this risk function as new data become available. The introduction of such new predictors to this function should be reserved, however, because the additional criteria are only likely to undermine the simplicity that makes this model so appealing. Coronary calcium scoring is an attractive new addition to this tool, but as we have shown, its benefits are likely to be greatest in specific prescreened patient groups older than a certain age, not across entire populations.7,8 We propose that future treatment guidelines include recommendations based on calculated risk for global CVD events. Nevertheless, physicians should be wary of using arbitrary risk cutoffs for the initiation of treatment in patients, because such guidelines generally rely on consensus rather than on evidence garnered from clinical trials. Despite the clear utility and benefit of risk prediction algorithms as initial screening tools, they should be used in treatment decisions only after careful validation. Perhaps a more meaningful way of using these risk tables may be to base treatment decisions on the calculated relative risk for a patient, with respect to the cardiovascular risk for normal patients of the same age group. Although current treatment guidelines do not incorporate such calculated relative risks, the ability to calculate heart and vascular age in D’Agostino et al’s1 model provides important information in understanding the relative age- and gender-matched CVD risk of a patient. At this point, D’Agostino et al’s1 criteria are not meant to be exhaustive in the factors that contribute to CVD risk.
1177
Clearly, a number of other predictors that are not factored into this assessment increase the risk for CVD.2 Similarly, some of the predictors of this tool (such as body mass index) may not be reliable in the setting of specific body habitus.13 From a design standpoint, there are other limitations to this risk assessment tool: notably, the study population is exclusively Caucasian, and thus, the tool needs to be validated across other ethnic groups. Furthermore, data used in the original Framingham Heart Study were better suited for populations in the 1970s to 1990s, before the rise of the obesity epidemic. Although the relative contributions of risk factors and the risk factors themselves are likely to have changed over the ensuing 30 to 40 years, the data collected do not account for this. Thus, further validation studies must establish whether the risk assessment models based on Framingham data are equally applicable today as they were in the past. Nevertheless, from the point of view of providing a reliable, convenient, and simple risk assessment tool, D’Agostino et al1 and have devised a winning product. 1. D’Agostino RB, Vasan RS, Pencina MJ, Wold PA, Cobain M, Massaro JM, Kannel WB. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation 2008;117: 743–753. 2. Executive summary of the third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285:2486 –2497. 3. De Backer G, Ambrosioni E, Borch-Johnsen K, Brotons C, Cifkova R, Dallongeville J, Ebrahim S, Faergeman O, Graham I, Mancia G, et al. European guidelines on cardiovascular disease prevention in clinical practice. Third Joint Taskforce of European and Other Societies on Cardiovascular Disease Prevention in Clinical Practice. Eur Heart J 2003;24:1601–1610. 4. Savill P. QRISK—a new CVD risk score. Practitioner 2007;251:7. 5. Hippisley-Cox J, Coupland C, Vinogradova Y, Robson J, May M, Brindle P. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: prospective open cohort study. BMJ 2007;335:136. 6. Hippisley-Cox J, Coupland C, Vinogradova Y, Robson J, Brindle P. Performance of the QRISK cardiovascular risk prediction algorithm in an independent UK sample of patients from general practice: a validation study. Heart 2008;94:34 –39. 7. Detrano R, Guerci AD, Carr JJ, Bild DE, Burke G, Folsom AR, Liu K, Shea S, Szklo M, Bluemke DA, et al. Coronary calcium scoring as a predictor of coronary events in four racial or ethnic groups. N Engl J Med 2008;358:1336 –1345. 8. Lakoski SG, Greenland P, Wong ND, Schreiner PJ, Herrington DM, Kronmal RA, Liu K, Blumenthal RS. Coronary artery calcium scores and risk for cardiovascular events in women classified at “low risk” based on Framingham risk score: the Multi-Ethnic Study of Atherosclerosis (MESA). Arch Intern Med 2007;167:2437–2442. 9. Beswick A, Brindle P. Risk scoring in the assessment of cardiovascular risk. Curr Opin Lipidol 2006;17:375–386. 10. Brindle P, Beswick A, Fahey T, Ebrahim S. Accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart 2006;92:1752–1759. 11. Hosmer DW, Lemeshow S. Applied Logistics Regression. 2nd ed. New York, New York: John Wiley, 2000. 12. Hall LM, Jung RT, Leese GP. Controlled trial of effect of documented cardiovascular risk scores on prescribing. BMJ 2003;326:251–252. 13. Savva SC, Tornaritis M, Savva ME, Kourides Y, Panagi A, Silikiotou N, Georgiou C, Karafatos A, Waist circumference and waist-to-height ratio are better predictors of cardiovascular disease risk factors than body mass index. Int J Obes Relat Metab Disord 2000;24:1453–1458.
Asymptomatic patients with risk factors for cardiovascular disease (CVD) do not come directly to the attention of cardiologists; they are first seen by primary care physicians in the setting of routine office visits or unexpectedly, while being assessed for other illnesses. Thus, the availability of rapid screening tools for primary care physicians to quickly and reliably stratify patients for CVD risk during office visits can enhance the detection of patients at increased risk for CVD. In the February 12, 2008, issue of Circulation, D’Agostino et al1 published the results of a study using participants of the Framingham Heart Study and the Framingham Offspring Study. In these analyses, multivariate risk assessment tools were refined to allow office-based physicians to rapidly and easily identify high-risk candidates for several different CVD outcomes, including global CVD events, as well as subtypes of coronary artery disease (CAD), cerebrovascular disease, peripheral arterial disease, and heart failure. The utility of this tool resides in its ability to provide absolute risk estimates for all clinically relevant forms of CVD in patients and to do so using data points and criteria that can be readily collected in the office setting. We believe that this risk assessment tool can be conveniently adopted in the primary care setting in the form of a risk scorecard that can be used by physicians and nursing staff a The Johns Hopkins Ciccarone Center for the Prevention of Heart Disease, Division of Cardiology, The Johns Hopkins University School of Medicine, Baltimore, Maryland; and bDepartment of Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois. Manuscript received December 3, 2008; revised manuscript received and accepted December 21, 2008. *Corresponding author: Tel: 443-799-8498; fax: 410-614-9190. E-mail address: [email protected] (S.R. Wickramasinghe).
0002-9149/09/$ – see front matter © 2009 Published by Elsevier Inc. doi:10.1016/j.amjcard.2008.12.040
members as part of routine patient evaluation, or even by patients in the waiting room before their visits. Use of Traditional Cardiovascular Disease Risk Assessment Tools Multivariate risk stratification for the assessment of CVD risk has been repeatedly used across multiple studies. For instance, the data from the Framingham Heart Study formulation for “hard” CAD events (myocardial infarction and CAD death) have been incorporated into the third report of the Expert Panel of Detection, Evaluation, and Treatment of Cholesterol in Adults (Adult Treatment Panel [ATP] III).2 Similarly, other countries have either integrated adaptations of the Framingham data or have designed their own risk assessment systems on the basis of CVD incidence statistics and risk factors prevalent in their specific populations: the Systematic Coronary Risk Evaluation (SCORE) project in Europe and the QRESEARCH cardiovascular RISK algorithm (QRISK) in the United Kingdom.3–5 The QRISK algorithm in particular uses traditional risk factors for CVD (age, systolic blood pressure, smoking, and ratio of total serum cholesterol to high-density lipoprotein cholesterol) but also incorporates family history, antihypertensive treatment, and body mass index.6 In addition, recent evidence suggests that coronary calcium scoring provides clear incremental predictive information on CAD risk beyond that of the Framingham risk estimate across multiple ethnic and racial groups.7 Given the high specificity of coronary calcium scoring in the prediction of CVD events, these data argue that noninvasive diagnostic modalities are best used in conjunction with traditional risk assessment systems, when there is uncerwww.AJConline.org
Editorial/Profiling Patients for Generalized CVD Risk
1175
Table 1 Cardiovascular disease risk and heart age scorecard based on office-based parameters Points
Age (yrs) M
⫺3 ⫺2 ⫺1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ⬎20
F
30–34
30–34
35–39
35–39
40–44 45–49 50–54 55–59 60–64 65–69 70–74 ⬎75
40–44 45–49
BMI (kg/m2) M
⬍25 25–30 ⬎30
F
⬍25 25–30 ⬎30
SBP Untreated (mm Hg)
SBP Treated (mm Hg)
Smoker?
Diabetic?
M
M
M
M
120–129 130–139 140–159 ⬎160
F
120–129 130–139 140–149 150–159 ⬎160
⬍120 120–129 130–139 140–150 ⬎160
F
⬍120
120–129 130–139 140–149 150–159 ⬎160
50–54 55–59 60–64 65–69 70–74 ⬎75
F
No
No
Yes
Yes
No
F
No
Yes Yes
CVD Risk
Heart Age (yrs)
M
F
M
F
1% 1% 2% 2% 3% 3% 4% 5% 6% 7% 8% 10% 11% 13% 15% 18% 22% 25% 29% ⬎30% ⬎30% ⬎30% ⬎30% ⬎30% ⬎30%
⬍1% ⬍1% 1% 1% 2% 2% 2% 3% 3% 4% 4% 5% 5% 6% 7% 8% 10% 11% 13% 15% 18% 20% 24% 27% ⬎30%
⬍30 ⬍30 31 33 35 37 39 41 44 46 49 52 55 58 62 65 69 73 78 80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
⬍30 ⬍30 ⬍30 32 34 36 38 41 43 46 48 51 54 58 61 65 69 73 77 80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
This table enables physicians to quickly and accurately calculate generalized CVD risk on the basis of D’Agostino et al’s1 formula, using simple, office-based parameters. BMI ⫽ body mass index; SBP ⫽ systolic blood pressure.
tainty regarding initiation of therapy, or when there is concern that the Framingham score may not be capturing a patient’s risk adequately.8 One such example is a middleaged patient with borderline or elevated risk factors (who will have a very low Framingham 10-year risk estimate despite risk factors, simply because of age) who has a family history of premature CVD. Given that the specificity of coronary calcium scoring translates into a high positive predictive value only when the population being screened has a moderate or high underlying disease incidence, officebased screening tools may allow the selection of a population more likely to benefit from more involved diagnostics such as coronary calcium scoring. Thus, at present, with the prohibitive cost of coronary calcium screening for the entire population, and possible long-term risks that are largely unknown, we believe that it is advantageous for physicians to routinely use rapid officebased screening methods that are accurate and convenient. These office-based risk assessment tools can then be used as initial screening devices to identify intermediate- or highrisk patient populations, who can then be selectively subjected to atherosclerosis imaging procedures with greater degrees of specificity and positive predictive values for predicting CVD risk. Nevertheless, it is important to remember that screening tools such as those derived from Framingham are not infallible. Traditional risk scoring systems are effective and discriminate risk well when applied
to large populations, but these should not be substituted for good clinical judgment at the individual patient level. Many previous risk assessment tools based on United States populations, including the standard Framingham risk score, assess the risk for CAD to the exclusion of other CVD processes, such as stroke, heart failure, and intermittent claudication.9 These other disease processes share common risk factors with CAD. In terms of physician-patient communication, it may be more meaningful to a patient to hear that his or her risk for a major CVD event is 15% over the next 10 years rather than a 5% risk for a hard CAD event over the next decade. In this context, the availability of a tool that can quickly and accurately calculate the risk for generalized CVD (not just CAD) in patients using simple criteria should become an important instrument in adult primary care. Despite the availability and demonstrated benefit of many risk prediction algorithms, their use in primary care has been quite limited, and the use of risk scores will not translate into better patient outcomes unless they are used appropriately by physicians.10 A number of potential reasons exist for the apparent reluctance of primary care physicians to adopt such risk assessment tools. Foremost among them is the fact that some risk assessment tools focus on predictive accuracy at the cost of simplicity, often involving long and tenuous risk score systems that require significant effort. In addition, most risk assessment systems necessitate laboratory testing results that may not
1176
The American Journal of Cardiology (www.AJConline.org)
Table 2 Cardiovascular disease risk and heart age scorecard based on laboratory-based parameters Points
⫺3 ⫺2 ⫺1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ⬎20
Age (yrs)
HDL Cholesterol (mg/dl)
M
M
F
F
Total Cholesterol (mg/dl) M
F
SBP Untreated (mm Hg) M
F
SBP Treated (mm Hg)
Smoker? Diabetic?
CVD Risk
M
M
M
F
F
M
F
⬍120 ⬎60 ⬎60 ⬍120 50–59 50–59 ⬍120 30–34 30–34 45–49 45–49 ⬍160 ⬍160 120–129 120–129 ⬍120 No No No No 35–44 35–44 160–199 160–199 130–139 130–139 35–39 35–39 ⬍35 ⬍35 200–239 140–159 140–149 120–129 120–129 240–279 200–239 ⬎160 130–139 130–139 Yes 40–44 ⬎280 240–279 150–159 140–159 Yes Yes Yes 40–44 45–49 ⬎280 ⬎160 ⬎160 140–149 45–49 150–159 50–54 ⬎160 50–54 55–59 60–64 55–59 65–69 60–64 70–74 65–69 ⬎75 70–74 ⬎75
F
⬍1% ⬍1% 1% ⬍1% 1% 1% 1% 1% 2% 1% 2% 2% 3% 2% 3% 2% 4% 3% 5% 3% 6% 4% 7% 4% 8% 5% 9% 6% 11% 7% 13% 9% 15% 10% 18% 12% 21% 14% 25% 16% 29% 18% ⬎30% 22% ⬎30% 25% ⬎30% 29% ⬎30% ⬎30%
Heart Age (yrs) M
F
⬍30 ⬍30 ⬍30 30 32 34 36 38 40 42 45 48 51 54 57 60 64 68 72 76 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
⬍30 ⬍30 ⬍30 30 31 34 36 39 42 45 48 51 55 59 64 68 73 79 80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80 ⬎80
This table enables the calculation of generalized CVD risk on the basis of D’Agostino et al’s1 formula, using laboratory values such as total cholesterol and HDL cholesterol. HDL ⫽ high-density lipoprotein; other abbreviation as in Table 1.
be available during routine clinic visits. We believe that a risk score must be simple to calculate and based on readily available patient data to be used appropriately. The new Framingham CVD risk score system meets those criteria. D’Agostino and Colleagues’ Risk Assessment System The general CVD risk prediction model of D’Agostino et al1 is an important advance in patient screening. Using original and offspring cohorts of the Framingham study, the investigators collected CVD risk factor data to generate gender-specific Cox proportional-hazards regressions, relating risk factors to the first incidence of CVD during a follow-up period of 12 years.1 Using these models, a mathematical function that predicts general CVD risk in the next 10 years was constructed. The investigators then presented this risk prediction model in the form of a risk score system, similar to the risk assessment system published by the National Cholesterol Education Program ATP III.1,2 The new Framingham general cardiovascular risk profile has several features that make it attractive. First, it uses 2 separate risk-scoring methods based either on data collected during a general office visit (age, body mass index, blood pressure, treatment of hypertension, smoking, and diabetes) or on data incorporating laboratory studies (including highdensity lipoprotein and total cholesterol). These assessment tools have discrimination statistics in the range of 0.75 to
0.79 for the prediction of generalized CVD risk, which is considered to be “very good” discrimination between cases and noncases.11 This risk assessment tool also incorporates the concept of “heart age.” In this concept, the CVD risk of a candidate is transformed into the age of a subject without any risk factors, but with the same CVD risk. Heart age may not be as useful to physicians in the diagnostic or therapeutic process, but can greatly affect patients’ impressions of their CVD risk. A 45-year-old man with significant risk factors will likely be better able to appreciate the notion that his heart has the age equivalent of a typical 70-year-old man than the quantitative concept that he carries a 20% CVD risk over the next 10 years. Unfortunately, current treatment guidelines are not based on D’Agostino et al’s1 scoring system, and simply applying their score in place of the Framingham or ATP III risk score to guide treatment, assuming the same thresholds are used as in ATP III, may result in an overuse of aspirin and lipid-lowering therapy on the basis of current treatment guidelines.12 For example, ATP III treatment recommendations are based on risk estimates of hard CAD. Because the newer risk equations published by D’Agostino et al1 incorporate all CVD end points, the risk estimates will be substantially higher than for hard CAD alone. Therefore, the use of this comprehensive CVD risk tool may potentially
Editorial/Profiling Patients for Generalized CVD Risk
lead to the overtreatment of patients; in contrast, statins and aspirin are generally well tolerated and may impart a reasonable risk/benefit ratio in subjects with a 10% risk for major cardiovascular events over the next decade. Interestingly, global CVD risk is always greater than the risk for a component of CVD, such as CAD or stroke. The risks for these individual components (CAD, heart failure, stroke) can also be calculated using this model by multiplying generalized CVD risk in D’Agostino et al’s1 scoring system by a calibration factor specific to the component of interest. Because of these powerful features and ease of use, we reproduce D’Agostino et al’s1 risk assessment tools into a single table that can be conveniently carried by physicians in their coat pockets and used with ease in an office setting (Table 1). In its consolidated form, each column of the table is divided by gender and represents a single risk factor: age, body mass index, systolic blood pressure treated, systolic blood pressure untreated, smoking, and diabetes. The last 2 columns represent cumulative CVD risk and heart age. To quickly estimate a patient’s 10-year total CVD risk, a physician need only read the score in the far-left column for each risk factor and then, after calculating the total score, read the cumulative CVD risk and heart age from the last 2 columns for the total score. This gives a reliable measure of a patient’s risk for CVD and, within a matter of minutes, can identify patients who require further diagnostic workup on the basis of physicians’ estimates of acceptable 10-year risk. A similar table that incorporates laboratory data for the assessment of generalized CVD risk can be designed (Table 2). A logical advance to this assessment tool will be the inclusion of other newer criteria that improve the predictive power of this risk function as new data become available. The introduction of such new predictors to this function should be reserved, however, because the additional criteria are only likely to undermine the simplicity that makes this model so appealing. Coronary calcium scoring is an attractive new addition to this tool, but as we have shown, its benefits are likely to be greatest in specific prescreened patient groups older than a certain age, not across entire populations.7,8 We propose that future treatment guidelines include recommendations based on calculated risk for global CVD events. Nevertheless, physicians should be wary of using arbitrary risk cutoffs for the initiation of treatment in patients, because such guidelines generally rely on consensus rather than on evidence garnered from clinical trials. Despite the clear utility and benefit of risk prediction algorithms as initial screening tools, they should be used in treatment decisions only after careful validation. Perhaps a more meaningful way of using these risk tables may be to base treatment decisions on the calculated relative risk for a patient, with respect to the cardiovascular risk for normal patients of the same age group. Although current treatment guidelines do not incorporate such calculated relative risks, the ability to calculate heart and vascular age in D’Agostino et al’s1 model provides important information in understanding the relative age- and gender-matched CVD risk of a patient. At this point, D’Agostino et al’s1 criteria are not meant to be exhaustive in the factors that contribute to CVD risk.
1177
Clearly, a number of other predictors that are not factored into this assessment increase the risk for CVD.2 Similarly, some of the predictors of this tool (such as body mass index) may not be reliable in the setting of specific body habitus.13 From a design standpoint, there are other limitations to this risk assessment tool: notably, the study population is exclusively Caucasian, and thus, the tool needs to be validated across other ethnic groups. Furthermore, data used in the original Framingham Heart Study were better suited for populations in the 1970s to 1990s, before the rise of the obesity epidemic. Although the relative contributions of risk factors and the risk factors themselves are likely to have changed over the ensuing 30 to 40 years, the data collected do not account for this. Thus, further validation studies must establish whether the risk assessment models based on Framingham data are equally applicable today as they were in the past. Nevertheless, from the point of view of providing a reliable, convenient, and simple risk assessment tool, D’Agostino et al1 and have devised a winning product. 1. D’Agostino RB, Vasan RS, Pencina MJ, Wold PA, Cobain M, Massaro JM, Kannel WB. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation 2008;117: 743–753. 2. Executive summary of the third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 2001;285:2486 –2497. 3. De Backer G, Ambrosioni E, Borch-Johnsen K, Brotons C, Cifkova R, Dallongeville J, Ebrahim S, Faergeman O, Graham I, Mancia G, et al. European guidelines on cardiovascular disease prevention in clinical practice. Third Joint Taskforce of European and Other Societies on Cardiovascular Disease Prevention in Clinical Practice. Eur Heart J 2003;24:1601–1610. 4. Savill P. QRISK—a new CVD risk score. Practitioner 2007;251:7. 5. Hippisley-Cox J, Coupland C, Vinogradova Y, Robson J, May M, Brindle P. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: prospective open cohort study. BMJ 2007;335:136. 6. Hippisley-Cox J, Coupland C, Vinogradova Y, Robson J, Brindle P. Performance of the QRISK cardiovascular risk prediction algorithm in an independent UK sample of patients from general practice: a validation study. Heart 2008;94:34 –39. 7. Detrano R, Guerci AD, Carr JJ, Bild DE, Burke G, Folsom AR, Liu K, Shea S, Szklo M, Bluemke DA, et al. Coronary calcium scoring as a predictor of coronary events in four racial or ethnic groups. N Engl J Med 2008;358:1336 –1345. 8. Lakoski SG, Greenland P, Wong ND, Schreiner PJ, Herrington DM, Kronmal RA, Liu K, Blumenthal RS. Coronary artery calcium scores and risk for cardiovascular events in women classified at “low risk” based on Framingham risk score: the Multi-Ethnic Study of Atherosclerosis (MESA). Arch Intern Med 2007;167:2437–2442. 9. Beswick A, Brindle P. Risk scoring in the assessment of cardiovascular risk. Curr Opin Lipidol 2006;17:375–386. 10. Brindle P, Beswick A, Fahey T, Ebrahim S. Accuracy and impact of risk assessment in the primary prevention of cardiovascular disease: a systematic review. Heart 2006;92:1752–1759. 11. Hosmer DW, Lemeshow S. Applied Logistics Regression. 2nd ed. New York, New York: John Wiley, 2000. 12. Hall LM, Jung RT, Leese GP. Controlled trial of effect of documented cardiovascular risk scores on prescribing. BMJ 2003;326:251–252. 13. Savva SC, Tornaritis M, Savva ME, Kourides Y, Panagi A, Silikiotou N, Georgiou C, Karafatos A, Waist circumference and waist-to-height ratio are better predictors of cardiovascular disease risk factors than body mass index. Int J Obes Relat Metab Disord 2000;24:1453–1458.