_________________________________
1Department of Internal Medicine and Clinical Epidemiology, Princess Alexandra Hospital, Brisbane, QLD 4102, Australia
Correspondence to: I A Scott ian_scott@health.qld.gov.au
Cite this as: BMJ 2009;338:a2844
<DOI> 10.1136/bmj.a2844 http://www.bmj.com/content/338/bmj.a2844
Clinician assessment of future risk of cardiovascular events in asymptomatic individuals is often inaccurate without formal risk prediction tools.1 Prediction tools that are easy to use and that integrate Framingham criteria of age, sex, serum cholesterol, blood pressure, smoking status, diabetes, and left ventricular hypertrophy into one global risk score have evolved to aid risk assessment.2 In recent decades, new putative risk factors have been described, including clinical risk factors such as chronic kidney disease and metabolic syndrome, as well as numerous laboratory markers of disease (box 1).3, 4, 5 Each new risk factor attracts studies evaluating whether a certain ancillary test shows an incrementally higher cardiovascular risk after adjusting for traditional risk factors.6 This review asks whether ancillary testing in asymptomatic adults improves the accuracy of predicting cardiovascular risk in individuals, and what effects it might have on patients’ care, behaviour, and clinical outcomes.
Indiscriminate testing could waste resources on tests that will not affect intended management (and that may cause anxiety) and on treatment, started as a result of positive tests, in populations for whom efficacy has not been proved. In the 35% of adults deemed to be at low risk of cardiovascular events (10 year risk <5%), further tests, even with abnormal results, may not justify further investigation or treatment.7 Conversely, in the 25% of patients at high risk (10 year risk ≥20%), even normal test results should not necessarily prompt discontinuation of aggressive preventive strategies. The 40% of patients at intermediate risk (5% to 20%) are the most likely to benefit from testing. This review does not consider investigations used to screen for or diagnose existing disease (rather than risk) in asymptomatic people.
SOURCES AND SELECTION CRITERIA
I searched Medline, Cochrane Library, BMJ Updates, and BMJ Clinical Evidence over the past 10 years using keywords “cardiovascular risk”, “primary prevention”, “screening”, “risk prediction”, “asymptomatic” and “prospective cohort studies”, “controlled trials”, and “systematic reviews”. The content dealing with risk prediction is mainly based on prospective cohort studies; that dealing with effect of risk assessment and feedback on health outcomes and risk profiles is based on controlled trials
SUMMARY POINTS
A positive or abnormal value for a new risk factor test will be more useful in estimating disease risk if:
In general, tests are clinically useful if they correctly predict an event or non-event in 70% or more of cases (C-index ≥0.70), which requires a strong association between test value and disease risk (unadjusted relative risk or odds ratio ≥10.0).8
Conventional risk factors and Framingham models The Framingham risk score underpins most contemporary models of cardiovascular risk prediction that use conventional risk factors (box 2).9 Models based on the Framingham risk score discriminate best in white Western females (C-statistic 0.79) and worst in non-white African males (0.63).10
New clinical risk models and clinical risk factors
New prediction models such as QRISK11 and ASSIGN12 aim to supersede Framingham based models by using data from more contemporary cohorts of individuals, many of whom are receiving treatment for known risk factors, and by including additional risk variables (box 2). The C-indices do not differ much between the models, but QRISK is better calibrated than the Framingham risk score and ASSIGN, producing more accurate estimates in patients at low (<5%) and high (≥20%) 10 year risk (table).11
BOX 1 NEW DISEASE MARKERS FOR ASSESSING CARDIOVASCULAR RISK
Markers of inflammation, endothelial dysfunction, oxidative stress
Cardiac injury markers
Markers of neurohormonal activation
Renal injury markers
Procoagulant markers
Dyslipidaemic markers
Glycaemic markers
The clinical label of metabolic syndrome (defined as varying combinations of fasting plasma glucose, triglycerides, lipoprotein ratios, blood pressure, and measures of visceral adiposity) increases cardiovascular risk by 50% (relative risk 1.5) after adjustment for traditional risk factors.13 However, most studies note little or no added improvement in predicting risk in individuals,14 and its value in this context remains unclear.
Major abnormalities on the resting electrocardiogram—such as Q waves, left bundle branch block, and atrial fibrillation—are associated with up to a 3.5-fold increased risk of cardiovascular death at 10 years, but outcomes are incorrectly classified in a third of cases (C-statistics ≤0.67).15 Similarly, exercise stress tests that report a positive or negative result based simply on ST segment deviation are little better than Framingham based models in predicting event risk. Among 25 927 asymptomatic men at low risk (mean age 43 years; 8 year coronary mortality rate 0.6%), an abnormal stress electrocardiogram, which occurred in only 4.3% of cases, incorrectly predicted death in almost 40% of cases, although accuracy improved with increasing number of risk factors (relative risk 21 for no risk factors, 80 for three or more factors).16 More sophisticated stress electrocardiogram scores based on multiple variables (exercise time, extent of ST deviation, provocation of chest pain) and clinical risk factors (age, sex, diabetes, smoking status) show more promise, with C-statistics for 5 year survival up to 0.83.17
BOX 2 COMPARISON OF MODELS OF CLINICAL RISK
Framingham risk score9
Assigns points to each of several risk factors identified by a Cox proportional hazards regression analysis of the previously untreated Framingham cohort (5209 white, middle class individuals who were studied in the US over two decades, starting in 1948. Number of points (positive or negative) assigned to each variable accords with its β-coefficient in the original Framingham regression equations. Variables include age (with points varying by sex), high density lipoprotein cholesterol, total cholesterol, systolic blood pressure, cigarette smoking, diabetes, and left ventricular hypertrophy on electrocardiography. Points are summed to give an estimate of the 5 year and 10 year probability of a coronary event. More recent modifications include existing treatment for hypertension and family history of premature coronary disease as risk factors.
QRISK11
Derived from Cox regression modelling of data from two million patients attending primary care clinics in the United Kingdom between 1995 and 2007. Variables included in the first version were age, sex, smoking status, systolic blood pressure, ratio of total serum cholesterol to high density lipoprotein, body mass index, family history of coronary heart disease in first degree relatives younger than 60, area measure of deprivation (Townsend score), and existing treatment with antihypertensive agent. A more recent version (QRISK2) has additional variables of self assigned ethnicity, type 2 diabetes, rheumatoid arthritis, renal disease, and atrial fibrillation.
Event risk at 10 years is calculated with interactive web based software. Recently endorsed by the National Institute for Health and Clinical Excellence (NICE) as the preferred risk prediction tool for UK citizens.
ASSIGN12
Derived from Cox regression modelling of data from 13<thin >297 individuals aged 30-74 who were initially free of cardiovascular disease and randomly chosen from 25 districts of Scotland between 1984 and 1987. Includes all Framingham variables along with family history of premature cardiovascular disease, cigarette dosage in smokers, and a measure of social deprivation (Scottish index of multiple deprivation).
In patients who cannot undergo exercise stress electrocardiography (uninterpretable baseline electrocardiogram; inability to exercise; characteristics such as anaemia or left ventricular hypertrophy that predispose to false-positive ST segment changes), the prognostic value of pharmacologically induced stress imaging (echocardiography, myocardial perfusion scanning) has not been well studied but, to date, seems to be low even in populations at high risk.18
Biomarkers
Attention has shifted to readily assayable biomarkers, either singly or in combinations known as multimarker panels (box 1). Several studies have assessed the association of such markers with risk in populations, but the few that have assessed the predictive use of single biomarkers in individual patients have all shown no incremental value beyond traditional risk factors.5 Multimarker panels have been claimed to have greater discriminatory power than single markers, but most studies,19, 20, 21 although not all,22 have reported disappointing results.
Non-invasive imaging tests
Can prediction be improved by directly imaging subclinical vascular atherosclerosis?
Coronary artery imaging
Using electron beam or multislice computed tomography, coronary artery calcium scanning visualises calcium accretions in atherosclerotic plaque lining the walls of coronary arteries, and computes a volumetric score (Agatston score) between 0 (no accretions) and over 1000 (heavy wall calcification).23 Other studies using multidetector tomography identify substantial obstructive coronary disease (>50% luminal stenosis), which includes non-calcified plaque.24
The largest study to date of coronary artery calcium scanning measured scores and all cause mortality over seven years in 25<thin space>253 asymptomatic individuals, in whom scores above 10 were independently predictive of mortality after adjustment for traditional risk factors, with relative risk increasing from 3.6 for a score of 11-100 to 9.4 for a score of ≥1000.23 These values corresponded to 12 year event rates of ≤5.5% for scores of 11-400, 7.0% for scores of 401-1000, and 23.1% for scores >1000. However, only 10% of all patients, in whom multiple risk factors were present, had scores greater than 400. After adjustment for conventional risk factors (dichotomised sex adjusted age, family history, hypertension, lipids, smoking, diabetes), scores retained greater accuracy in predicting risk than did risk factors alone (C-statistic 0.76 versus 0.61; P=0.03). However, when scores were compared with age alone as a continuous variable, the discrimination difference was attenuated (C-statistic 0.81 v 0.77), implying age is the single most important predictor of both death and extent of coronary artery calcification.
Current expert consensus recommends avoiding computed angiography in risk assessment for asymptomatic people at low risk, because of the carcinogenic risk associated with radiation (up to 2% for annual scanning over a lifespan),25 and the possibility of iatrogenic morbidity incurred by invasive investigation—pulmonary lesions are incidentally detected in 20% of people who undergo scanning, but most are benign.26
Carotid and peripheral artery imaging
In asymptomatic populations, thickening of the intima media of the carotid artery or reduced arterial flow in the leg (low ankle brachial index), as assessed by ultrasound, is associated with as much as a threefold increase in cardiovascular risk after adjustment for traditional risk factors.27 However, studies of incremental predictive value in individuals are few and no recommendations exist for the routine use of vascular ultrasound in risk prediction.
What is the evidence that risk prediction alters clinical management or outcomes?
A systematic review found eleven studies, including five randomised controlled trials with reasonable methodological quality, that had evaluated the clinical benefits and harms of applying Framingham based risk scores to the care of asymptomatic people.28 None of the five showed a significant overall change in care or risk factors, although no harmful effects were seen. In one trial, an a priori subgroup analysis of patients at high risk showed significant improvements in prescribing of lipid lowering agents (absolute increase 11%) and antihypertensive drugs (absolute increase 13%).29 Two other trials showed improvements in systolic blood pressure (4.6 mm Hg decrease) 30 and uptake of physical exercise (11% increase).31 A more recent trial not included in the above review found a significant but small (0.08 mmol/l) decrease in low density lipoprotein cholesterol.32 No effects on event rates have been reported; studies have been underpowered to detect such effects.
No trials have directly assessed the effects of stress testing or laboratory biomarkers on control of risk factors or on clinical outcomes compared with no investigation. As many as 3% of patients undergoing stress electrocardiography in cohort studies subsequently showed severe coronary disease prompting revascularisation,12 but no trials have confirmed benefit from revascularisation in asymptomatic populations. Moreover, in populations at lower risk (<10% event risk at 5 years), at least 75% of positive test results are false positives, potentially resulting in unwarranted testing and anxiety.
One trial assessed the effect of coronary artery calcium scanning in 450 asymptomatic US army personnel (mean age 42 years, mean 10 year risk of coronary events 5.9%), of whom 15% had positive scores.33 Using a factorial design that randomised participants to usual care or intensive case management with or without feedback to doctor or patient about the coronary artery calcium score, awareness of the score had no effect on risk profile or lifestyle. Another trial, which randomised 1005 people with raised coronary artery calcium scores (above 80th percentile) to a statin and vitamins C and E or matching placebo, found no treatment effect on scores or cardiovascular event rates at four year follow-up.34
According to the available evidence, ancillary testing in most asymptomatic people adds little to the prediction of individual cardiovascular risk, and does not affect care, lifestyle, adherence, or clinical outcomes. Randomised trials are needed to assess whether the improved accuracy of risk categorisation provided by some tests leads to therapeutic reductions of risk in patients not previously identified, or to reductions in the number of patients needing treatment by identifying low risk groups. Other areas in need of studies are the potential harms and cost effectiveness of additional testing, generalisability of results to populations with different ethnicities and comorbidities, standardisation of test assays in maximising precision and validation of population norms to guide interpretation of results, and observer bias and reliability.
CASE SCENARIO
A 58 year old man who smokes and is being treated for hypertension with an angiotensin converting enzyme inhibitor comes in for review, driven by new concern about his cardiovascular risk since his 54 year old brother recently had a heart attack. You have been caring for him for some years and have been trying to get him to stop smoking. He is also overweight (body mass index 27) and has an unhealthy diet, with a ratio of total cholesterol to high density lipoprotein cholesterol of 4.0 while receiving statin therapy. He has read in the lay press about new blood tests that might give a better indication of his cardiovascular risk, and he is aware of new radiography techniques that can apparently show calcified plaque in coronary arteries. He asks whether these tests might be relevant to him. You hesitate, unsure about how useful these tests would be, and noting that no reference is made to them in recent NICE guidelines on risk assessment (see “Additional educational resources”). You are also worried that if such tests yield a negative result he might be even less motivated to alter his lifestyle than he is already. On the other hand, if the tests show a positive result, you are not sure what part of your current management plan you would change in response.
ADDITIONAL EDUCATIONAL RESOURCES
For healthcare professionals
For patients
A case may be made for further testing in patients at intermediate or high risk who are younger (<35 years), older (>75 years), or come from population groups not included in cohorts used to derive and validate current clinical risk models. However, ancillary testing in most middle aged people is premature and potentially wasteful of resources. An alternative strategy, which needs to be studied in randomised trials, is to ensure that all adult patients (and their doctors) are aware of their cardiovascular risk by using simple, accessible clinical risk calculators (such as QRISK) and adopt a management plan appropriate to their level of risk. In the case scenario presented, the 10 year cardiovascular event risk using QRISK2 is 33%, which should provide sufficient motivation for this patient and his doctor to optimise lifestyle and preventive treatments, making further testing unnecessary.
I thank Chris del Mar, Jenny Doust, and Rod Jackson for their helpful comments on earlier drafts, and Peter Hawkins for giving me the idea for this review.
Contributors: JAS is the sole contributor.
Funding: None received.
Competing interests: None declared.
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent not required (patient anonymised, dead, or hypothetical).
1 Friedmann PD, Brett AS, Mayo-Smith MF. Differences in generalists’ and cardiologists’ perceptions of cardiovascular risk and the outcomes of preventive therapy in cardiovascular disease. Ann Intern Med 1996; 124: 414-21.
2 Jackson R. Updated New Zealand cardiovascular disease risk-benefit prediction guide. BMJ 2000; 320: 709-10.
3 Astor BC, Coresh J, Heiss G, Pettitt D, Sarnak MJ. Kidney function and anemia as risk factors for coronary heart disease and mortality: the Atherosclerosis Risk in Communities (ARIC) Study. Am Heart J 2006; 151: 492-500.
4 Wilson PW, D’Agostino RB, Parise H, Sullivan L, Meigs JB. Metabolic syndrome as a precursor of cardiovascular disease and type 2 diabetes mellitus. Circulation 2005; 112: 3066-72.
5 Tsimikas S, Willerson JT, Ridker PM. C-reactive protein and other emerging blood biomarkers to optimize risk stratification of vulnerable patients. J Am Coll Cardiol 2006; 47 (suppl 8): C19-31.
6 Hackham DG, Anand SS. Emerging risk factors for atherosclerotic vascular disease. A critical review of the evidence. JAMA 2003; 290: 932-40.
7 Greenland P, Smith SC Jr, Grundy SM. Improving coronary heart disease risk assessment in asymptomatic people: role of traditional risk factors and noninvasive cardiovascular tests. Circulation 2001; 104: 1863-7.
8 Pepe MS, Janes H, Longton G, Leisenring W, Newcomb P. Limitations of the odds ratio in gauging the performance of a diagnostic, prognostic, or screening marker. Am J Epidemiol 2004; 159: 882-90.
9 Anderson KM, Odell PM, Wilson PW, Kannel WB. Cardiovascular disease risk profiles. Am Heart J 1991; 121: 293-8.
10 D’Agostino RB Sr, Vasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation 2008; 117: 743-53.
11 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.
12 Woodward M, Brindle P, Tunstall-Pedoe H, SIGN group on risk estimation. Adding social deprivation and family history to cardiovascular risk assessment: the ASSIGN score from the Scottish Heart Health Extended Cohort (SHHEC). Heart 2007; 93: 172-6.
13 Gami AS, Witt BJ, Howard DE, Erwin PJ, Gami LA, Somers VK, et al. Metabolic syndrome and risk of incident cardiovascular events and death: a systematic review and meta-analysis of longitudinal studies. J Am Coll Cardiol 2007; 49: 403-14.
14 Kahn R, Buse J, Ferrannini E, Stern M, American Diabetes Association; European Association for the Study of Diabetes. The metabolic syndrome: time for a critical appraisal: joint statement from the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 2005; 28: 2289-304.
15 Fowler-Brown A, Pignone M, Pletcher M, Tice JA, Sutton SF, Lohr KN, et al. Exercise tolerance testing to screen for coronary heart disease: a systematic review for the technical support for the U.S. Preventive Services Task Force. Ann Intern Med 2004; 140: W9-24.
16 Gibbons LW, Mitchell TL, Wei M, Blair SN, Cooper KH. Maximal exercise test as a predictor of risk for mortality from coronary heart disease in asymptomatic men. Am J Cardiol 2000; 86: 53-8.
17 Lauer MS, Pothier CE, Magid DJ, Smith SS, Kattan MW. An externally validated model for predicting long-term survival after exercise treadmill testing in patients with suspected coronary artery disease and a normal electrocardiogram. Ann Intern Med 2007; 147: 821-8.
18 Blumenthal RS, Becker DM, Yanek LR, Aversano TR, Moy TF, Kral BG, Becker LC. Detecting occult coronary disease in a high-risk asymptomatic population. Circulation 2003; 107: 702-7.
19 Wang TJ, Gona P, Larson MG, Tofler GH, Levy D, Newton-Cheh C, et al. Multiple biomarkers for the prediction of first major cardiovascular events and death. N Engl J Med 2006; 355: 2631-9.
20 Folsom AR, Chambless LE, Ballantyne CM, Coresh J, Heiss G, Wu KK, et al. An assessment of incremental coronary risk prediction using C-reactive protein and other novel risk markers: the atherosclerosis risk in communities study. Arch Intern Med 2006; 166: 1368-73.
21 Shlipak MG, Fried LF, Cushman M, Manolio TA, Peterson D, Stehman-Breen C, et al. Cardiovascular mortality risk in chronic kidney disease: comparison of traditional and novel risk factors. JAMA 2005; 293: 1737-45.
22 Zethelius B, Berglund L, Sundstr≥m J, Ingelsson E, Basu S, Larsson A, et al. Use of multiple biomarkers to improve the prediction of death from cardiovascular causes. N Engl J Med 2008; 358: 2107-16.
23 Budoff MJ, Shaw LJ, Liu ST, Weinstein SR, Mosler TP, Tseng PH, et al. Long-term prognosis associated with coronary calcification: observations from a registry of 25,253 patients. J Am Coll Cardiol 2007; 49: 1860-70.
24 Bachar GN, Atar E, Fuchs S, Dror D, Kornowski R. Prevalence and clinical predictors of atherosclerotic coronary artery disease in asymptomatic patients undergoing coronary multidetector computed tomography. Coron Art Dis 2007; 18: 353-60.
25 Brenner DJ, Elliston CD. Estimated radiation risks potentially associated with full-body CT screening. Radiology 2004; 232: 735-8.
26 Burt JR, Iribarren C, Fair JM, Norton LC, Mahbouba M, Rubin GD, et al. Atherosclerotic Disease, Vascular Function, and Genetic Epidemiology (ADVANCE) Study. Incidental findings on cardiac multidetector row computed tomography among healthy older adults: prevalence and clinical correlates. Arch Intern Med 2008; 168: 756-61.
27 Van der Meer IM, Bots ML, Hofman A, del Sol AI, van der Kuip DAM, Witteman JCM. Predictive value of non-invasive measures of atherosclerosis for incident myocardial infarction. The Rotterdam Study. Circulation 2004; 109: 1089-94.
28 Sheridan SL, Crespo E. Does the routine use of global coronary heart disease risk scores translate into clinical benefits or harms? A systematic review of the literature. BMC Health Services Research 2008; 8: 60-73.
29 Hall LM, Jung RT, Leese GP. Controlled trial of effect of documented cardiovascular risk scores on prescribing. BMJ 2003; 326: 251-2.
30 Montgomery AA, Fahey T, Peters TJ, MacIntosh C, Sharp DJ. Evaluation of computer based clinical decision support system and risk chart for management of hypertension in primary care: randomised controlled trial. BMJ 2000; 320: 686-90.
31 Van Steenkiste B, van der Weijden T, Stoffers HE, Kester AD, Timmermans DR, Grol R. Improving cardiovascular risk management: a randomised, controlled trial on the effect of a decision support tool for patients and physicians. Eur J Cardiovasc Prev Rehabil 2007; 14: 44-50.
32 Grover SA, Lowensteyn I, Joseph L, Kaouache M, Marchand S, Coupal L, et al. Patient knowledge of coronary risk profile improves the effectiveness of dyslipidemia therapy: the CHECK-UP study: a randomized controlled trial. Arch Intern Med 2007; 167: 2296-303.
33 O’Malley PG, Feuerstein IM, Taylor AJ. Impact of electron beam tomography, with or without case management, on motivation, behavioral change, and cardiovascular risk profile: a randomized controlled trial. JAMA 2003; 289: 2215-23.
34 Arad Y, Spadaro LA, Roth M, Newstein D, Guerci AD. Treatment of asymptomatic adults with elevated coronary calcium scores with atorvastatin, vitamin C, and vitamin E: the St. Francis Heart Study randomized clinical trial. J Am Coll Cardiol 2005; 46: 166-72.