|Home | About | Journals | Submit | Contact Us | Français|
Current guideline statements for primary and secondary prevention of cardiovascular disease rely on estimates of absolute risk of coronary events. For example, the American Heart Association guidelines on primary prevention state that persons with ≥10% risk over ten years of myocardial infarction (MI) or coronary death should be considered for antiplatelet therapy with aspirin.1 Similarly, the National Cholesterol Education Program Adult Treatment Panel III (ATP III) guidelines2 state that target low-density lipoprotein (LDL) level should be based on projected absolute risk of future coronary events rather than on presence or absence of specific risk factors. These guidelines state that patients at high risk of MI and coronary death, defined as an absolute 10-year risk of 20% or more, should have a target LDL level <100 mg/dL and should receive statin therapy if needed to achieve this goal. Stroke, however, is not included as one of the outcomes contributing to these absolute risk levels.
Included in the group of patients with elevated risk, moreover, are those who already have ischemic heart disease, as well as patients deemed to be "coronary heart disease (CHD) risk equivalents," indicating those at the same elevated risk as patients with ischemic heart disease. CHD risk equivalents include patients with diabetes mellitus, those with multiple risk factors that put them at elevated risk based on calculation of their Framingham Score, and patients with "other forms of symptomatic atherosclerotic disease." The latter group is further defined to include those with peripheral arterial disease (PAD), abdominal aortic aneurysm (AAA), and carotid artery disease. The category of "risk equivalents" in the ATP III guidelines, however, does not include the vast majority (~80%3) of ischemic stroke patients without carotid artery disease as cause of their stroke.
Ischemic stroke is therefore notably excluded from the list of outcomes contributing to the absolute risk estimates used in these statements, and stroke patients are also excluded from the category of CHD risk equivalents. Evolving evidence suggests that ischemic stroke should be included in estimation of vascular risk and as a CHD risk equivalent. This inclusion could have implications for primary and secondary prevention.
The goal of this review is to summarize the evidence related to the following two questions: 1) Among stroke patients, is the risk of MI/coronary death high enough to justify calling stroke itself a risk equivalent?; and 2) Should stroke be included along with coronary heart disease in the vascular disease outcome cluster for purposes of absolute estimation of risk in primary and secondary prevention? In other words, should we speak of “coronary and stroke risk equivalents” rather than just “CHD risk equivalents”?
The answers to these questions have important clinical implications, as current guidelines may underestimate risk, leaving untreated patients who might be eligible for primary and secondary preventive therapies. These issues may be especially important for race-ethnic groups for whom stroke risk is as great as or exceeds coronary risk.4, 5
We first reviewed the ATP III guidelines for the definition of a CHD risk equivalent and the discussion supporting the inclusion of DM, PAD, AAA, and carotid artery disease as risk equivalents.6
To address the first question, we reasoned that the most appropriate data would be derived from prospective studies that examine risk or event rates of MI and vascular or coronary death among patients who have had a stroke. A structured, step-wise database search involving PubMed, the Cochrane Database of Systematic Reviews, and the ISI Web of Knowledge was performed under the limits of English language, human subjects, age ≥19, and publication date between 1/1/1980 and 1/1/2010, with search terms mapped to medical subject heading (MeSH) as well as queried as text words. Search terms included “stroke,” “risk,” “cardiac,” “myocardial infarction,” and “coronary death.” All citation summaries were reviewed for relevance, and studies were selected if: original, prospective data or meta-analysis of prospective data were reported; follow-up period was greater than 3 months; patients had a first stroke or transient ischemic attack (TIA) as a qualifying event; and risk or event rates of MI and vascular or coronary death were reported.
To address the second question, we reasoned that the most appropriate data would be derived from prospective studies examining stroke-free populations followed over time for the outcomes of MI, vascular or coronary death, and nonfatal and fatal stroke. Comparing the 10-year risk of CHD in such studies with the 10-year risk of CHD + stroke would reveal the effect of including stroke on current guideline recommendations that assign ‘high’ risk based upon a particular level of absolute risk. A second search was performed as above, using the same databases and limits. Search terms included “myocardial infarction,” “vascular,” “coronary death,” and “stroke.” All citation summaries were reviewed for relevance, and studies were selected if: original, prospective data or meta-analysis of prospective data were reported; follow-up period was greater than 3 months; patients were free of cardiovascular disease at entry; and MI, vascular or coronary death, and nonfatal and fatal stroke were reported separately as outcomes.
For both questions, the reference lists of selected studies were reviewed to identify studies that may have been missed in the initial database searches. Experts in the field were also asked to suggest relevant articles. Unpublished data were not sought, nor were authors contacted to provide additional information. In the case of multiple publications from the same study, the data most relevant to the questions of this review were selected.
All studies meeting the inclusion criteria were reviewed in detail, and the following information was abstracted: study design, inclusion and exclusion criteria, sample size, follow-up period, definition of outcomes, and rates or risk of outcome events. Data from relevant studies is presented in summary tables that describe study participants, follow-up period, and risk or rates of outcome events. Studies are grouped by design: meta-analysis, observational studies, and clinical trials. For studies related to the first question, risk or event rate data are presented in 2 groups: cardiac and cardiac+stroke. For clinical trials, data for the placebo group is presented, and exceptions are noted in the tables. For studies related to the second question, risk or event rate data are presented in 3 groups: cardiac, stroke-related, and combined. Since outcome definitions and follow-up times vary among studies, the outcome definition used in each study is specified, along with time period. Annualized risks were calculated by dividing total risk by the follow-up period. We present a narrative overview rather than a meta-analysis since, for the first question, a meta-analysis was performed recently (2005), which we review along with studies published since 2005, and for the second question, there were significant differences in methodology among studies.
According to ATP III guidelines,6 persons with established CHD in the United States have a risk of recurrent MI and CHD death of >20% over 10 years. Hence, a CHD risk equivalent is defined as a condition conferring a risk of MI or coronary death equivalent to that among those with established CHD, i.e. >20% over 10 years. These guidelines define four diseases as CHD risk equivalents: PAD, carotid disease, AAA, and DM (Table 1). Five studies are cited to support PAD as a risk equivalent, with sample sizes ranging from 567 to 1592 and follow-up ranging from 3 to 10 years. Seven studies are cited to support carotid disease as a risk equivalent, with sample sizes ranging from 158 to 3024 and follow-up ranging from 2.5 to 8 years. Only one study was cited to support including AAA as a risk equivalent, including 343 participants with 6–11 years of follow-up. Three studies are cited to illustrate the risk of CHD among those with DM, with sample sizes ranging from 1059 to 4075 with 2–11 years of follow-up. Further evidence is cited to illustrate an increased case fatality rate for MI among diabetics and worse prognosis for survival among diabetics who develop CHD compared to non-diabetics. The CHD risk for each of the traditional risk equivalents is approximately 2% per year (Table 1).
A total of 1844 articles were identified through database queries, and 1311 met language, age, and date range criteria. Of these, 41 met inclusion criteria, and 3 more studies were identified through bibliographic review and suggestions from experts. Since a recent meta-analysis7 (2005) summarized results of 39 of these studies, we present data from this meta-analysis and 4 other studies that have been published since 2005.
In a recent publication, Touze et al7 performed a systematic review and a meta-analysis of the absolute risk of MI and vascular death after stroke or transient ischemic attack (TIA) in 39 studies. Inclusion criteria included prospective cohort study or randomized controlled trial (RCT) design, published after 1979, reporting on long-term follow-up of ≥100 patients, with follow-up ≥1 year with <5% loss to follow-up, written in English, with outcome data for MI or vascular death. Exclusion criteria included reporting hemorrhagic strokes only, having a highly selected population (e.g., single sex, young subjects, or specific race), or patients with a “specific unusual cause of stroke.” Overall, there were 25 RCTs, 8 population-based cohorts, and 6 single-center hospital-based cohorts, including a total of 65,996 patients with a mean follow-up of 3.5 years. Overall, meta-regression showed annual risks of total MI of 2.2% (95% CI, 1.7 to 2.7, 22 studies), nonfatal MI of 0.9% (95% CI, 0.7 to 1.2, 16 studies), and fatal MI of 1.1% (95% CI, 0.8 to 1.5, 19 studies).
In the Northern Manhattan Study (NOMAS),8 a population-based cohort of first ischemic stroke patients ≥40 years old was prospectively followed annually for recurrent stroke, MI and cause-specific mortality. The 5-year risk of MI or vascular death was 17.4% (95% CI, 14.2% to 20.6%). In the lowest risk group, those ≤70 years old without coronary artery disease, 5-year risk of MI or vascular death was 9.7%. The 5-year risk of MI, recurrent stroke, or vascular death was 29.0% (95% CI, 25.2% to 32.7%). In another observational study of administrative databases with follow-up of 1.2–1.3 years, the annual rate of MI was 1.7%, and the annual rate of MI and stroke was 5.7%.9
In the placebo group of the SPARCL trial of lipid lowering therapy among patients with stroke or TIA,10 the risk of major coronary events (including death from cardiac causes, nonfatal MI, and resuscitation after cardiac arrest) was 5.1% over the median of 4.9 years, less than the 2% per year cutoff for a CHD risk equivalent, likely due to the fact that patients without coronary disease were enrolled. However, in the composite outcome of major coronary event plus stroke, the risk was 17.2% over 5 years, well over the 10% cutoff over 5 years for a risk equivalent. In the placebo group of the PRoFESS study,11 a secondary stroke prevention trial, a composite of major cardiovascular events (death from cardiovascular causes, recurrent stroke, myocardial infarction, or new or worsening heart failure) occurred in 1463 patients (14.4%) over a mean of 2.5 years.
The studies reviewed here, which examine recurrent stroke and cardiac events after stroke, show that the risk of cardiac events after stroke is approximately 2% per year, and a large meta-analysis7 suggests that the risk may even be >2% annually. When recurrent stroke is added to the outcomes for absolute risk estimates, the risk of vascular events after stroke is increased further.
A total of 722 articles were identified through database queries, and 449 met language, age, and date range criteria. Of these, 10 met inclusion criteria, and 3 more studies were identified through bibliographic review and suggestions from experts; we review the resulting 13 studies here.
In 2008, the Framingham Heart Study published a general cardiovascular risk profile scoring system.12 Participants were free of cardiovascular disease (CVD) and aged 30 to 74 years. CVD was defined as a composite of CHD (coronary death, MI, coronary insufficiency, and angina), cerebrovascular events (ischemic stroke, hemorrhagic stroke, and TIA), PAD, and heart failure. Maximum follow-up was 12 years. The authors compared risk of individual components with the composite outcomes. For women in the 5th decile of risk, the mean 10-year risk of CVD was approximately 4%, corresponding to a risk of CHD of ~2.4%, a risk of stroke of 0.95%, and a combined risk of CHD or stroke of ~3.4%. For men in the 5th decile of risk, the mean 10-year risk of CVD was approximately 12%, corresponding to a risk of CHD of ~7.3%, a risk of stroke of ~2.9%, and a combined risk of CHD or stroke of ~10.2%.
In the “Riskard 2005” study,13 9 population-based studies in Italy comprising 12,045 men and 5,108 women aged 35–74 were analyzed. Follow-up ranged from 5–15 years, and outcomes assessed included mortality, causes of death, and non-fatal cardiovascular events. Three categories of outcomes were considered: major coronary events (sudden coronary death, non-sudden coronary death, definite non-fatal myocardial infarction, fatal myocardial infarction, definite fatal chronic ischemic heart disease, surgery of coronary arteries), major cerebrovascular events (definite fatal and non-fatal hemorrhagic and thrombotic stroke, surgery of carotid arteries), and major cardiovascular events (major coronary and cerebrovascular events as defined above, plus major peripheral artery events comprising fatal and non-fatal aortic aneurysms, fatal lower limbs artery disease, surgery of aorta or lower limb arteries). Estimating from the figures depicting risk of these 3 outcomes by sex, age, and follow-up time, the 10-year risk of first major coronary events was ~6% in males and ~3% in females aged 60 years, whereas the 10-year risk of first major cardiovascular events was ~11% in males and ~4% in females.
In the Reduction of Atherothrombosis for Continued Health (REACH) study,14 participants were enrolled with either 1) a history of coronary artery disease (CAD), cerebrovascular disease, or peripheral arterial disease (PAD), or 2) at least 3 atherothrombotic risk factors. Participants were enrolled from multiple international outpatient sites and followed at 1 year for cardiovascular outcomes. Among the 11,766 participants without a history of CVD but with multiple risk factors, the 1-year event rate of cardiovascular death + nonfatal MI was 1.51%, whereas the rate of stroke was 0.80%. The 1-year event rate of the combined outcome cardiovascular death, MI, or stroke was 2.15%.
Clinical trial data is limited by selection bias and short-term follow-up but can provide stratified risk of strictly defined outcome events. In the LIFE trial,15 the rate of MI was 4%, the rate of stroke 6%, and the rate of the combined outcome cardiovascular mortality, stroke, or MI was 12%. Although there was a significant reduction in risk with losartan treatment, the results from both treatment groups are pooled here to reflect anti-hypertensive treatment in the community, which may not be in accordance with the stipulations of a clinical trial. In the ACCORD trial,16 in the standard therapy group, the rate of nonfatal MI and CV death was 6.4%, the rate of nonfatal and fatal stroke was 1.4%, and the risk of nonfatal MI or nonfatal stroke or CV death was 7.2%. In the ADVANCE trial,17 in the standard therapy group, 9.1% of participants had a history of stroke at study entry. In the standard therapy group, the rate of nonfatal MI and CV death was 8.0%, the rate of all cerebrovascular events was 5.9%, and the rate of nonfatal MI, nonfatal stroke, and CV death was 10.6%. In the CHARISMA trial,18 in the placebo + aspirin group, the rate of nonfatal MI and CV death was 4.9%, the rate of nonfatal stroke was 2.4%, and the risk of nonfatal MI, nonfatal stroke, or CV death was 7.3%. In the ONTARGET trial,19 in the combination therapy group, 20.9% had a history of stroke or TIA. In the combination therapy group, the rate of fatal and nonfatal MI and CV death was 12.5%, the rate of fatal and nonfatal stroke was 4.4%, and the risk of death from CV causes, MI, or stroke was 14.1%. Table 4 summarizes 5 further trials20–25 that show similar increases in combined risk compared to cardiac risk or stroke risk alone.
In summary, observational studies and clinical trial data show that adding risk of stroke to risk of cardiac events significantly elevates risk and in most cases results in crossing the threshold of absolute risk of 20% over 10 years that defines “high” risk.
Guidelines from other international groups already include stroke in the outcome cluster for risk calculation. In the European guidelines on cardiovascular disease prevention in clinical practice (Third Joint Task Force of European and other Societies on Cardiovascular Disease Prevention in Clinical Practice),26 the stated rationale for changing the focus from CHD to CVD prevention was a similar etiology for MI, ischemic stroke, and PAD, as well as treatments that are effective for all of these forms of vascular disease.
The Joint British Societies’ (JBS) guidelines on prevention of CVD27 present a risk prediction chart to estimate risk of developing CVD over 10 years. In the latest version of the JBS guidelines, CHD risk was replaced by CVD risk, a combined end point of CHD (fatal and non-fatal myocardial infarction and new angina) plus stroke (fatal and non-fatal stroke and cerebral haemorrhage) and transient cerebral ischemia.
The latest New Zealand guidelines28 provide a chart to estimate 5-year risk of “a cardiovascular event…defined as a death related to coronary disease, non-fatal myocardial infarction, new angina, fatal or non-fatal stroke or transient ischaemic attack, or the development of congestive heart failure or peripheral vascular disease.”
A review of evidence from hospital- and population-based studies supports: 1) the designation of stroke as a risk equivalent, and 2) the inclusion of stroke in the outcome cluster for purposes of absolute estimation of risk in primary and secondary prevention. Among patients who have had a stroke, the risk of subsequent cardiac events is high enough to designate stroke a CHD “risk equivalent” according to definitions in current guidelines. Including recurrent stroke in the outcome cluster leads to even higher absolute risk levels after first stroke. Furthermore, in observational studies and clinical trials of stroke-free subjects, including stroke in the calculation of cardiovascular risk often results in a crossing of the 10 year 20% threshold designating elevated risk. The amount and quality of the data supporting these estimations is at least as significant as the data that supported the designation of PAD, AAA, carotid artery disease, and DM as CHD risk equivalents (Table 1). As a result, many recent, large-scale studies (for example, the Women’s Health Study,29 QRISK,30 ASSIGN,31 and SCORE32) and international guidelines for primary and secondary prevention have already included stroke in the risk calculation of cardiovascular disease. Similarly, many recent clinical trials evaluating treatments for prevention of vascular disease have used primary outcomes that are a combined endpoint of cardiac events, stroke, and vascular death (for example, CHARISMA,18 MATCH,33 ESPRIT,34 ADVANCE,17 ONTARGET,19 and ACCORD16).
Excluding stroke as an outcome in risk calculation could have significant clinical implications, due to an underestimation of risk and failure to treat patients who would otherwise qualify for preventive therapy. For example, for an individual whose risk score using only coronary events puts him or her just below the threshold for treatment (10% over ten years for aspirin or 20% over ten years for statin therapy), inclusion of stroke in the outcome cluster would likely push him or her over the threshold.
Although many of the studies that report cerebrovascular events include both ischemic and hemorrhagic strokes, for which treatment strategies differ, a large population-based study that included only ischemic stroke patients showed that cardiac risk exceeded the threshold designating elevated risk.8 Furthermore, hemorrhagic strokes form a minority of total strokes, and the aggregation of ischemic and hemorrhagic strokes is unlikely to significantly affect the risk estimations reviewed here. However, the approach to the primary and secondary prevention of hemorrhagic stroke is not addressed by this review.
Ischemic stroke and CAD have clearly different and specific treatment approaches. For example, warfarin for atrial fibrillation is indicated for stroke prevention, carotid endarterectomy for carotid disease, and beta-blockers for CAD. However, for many current and possibly future therapies, ischemic stroke and CAD are likely to be treated in a similar manner. It is with a view of this overlap that practitioners are likely to get the greatest impact in prevention, provided that accurate risk estimates are used. For example, well-documented modifiable risk factors that are common to both CAD and ischemic stroke are hypertension, cigarette smoking, diabetes, dyslipidemia, and physical inactivity.1, 5 Despite debates about the sex-specific effects of aspirin in primary prevention,35 anti-platelet therapy is currently a mainstay in the prevention of both cardiac and cerebrovascular diseases. Furthermore, trials such as HPS36 and SPARCL10 have shown that statins provide effective primary and secondary prevention of cardiac events as well as stroke. Secondary analysis from SPARCL showed that atorvastatin was similarly effective in reducing risk of stroke and cardiovascular outcomes regardless of baseline stroke subtype,37 further lending support to the inclusion of all-cause stroke in risk calculation.
For providers interested in primary and secondary stroke prevention, there is also a current class of treatments that falls within a gray zone, in which there is little guidance about how to treat stroke patients. For example, the flu vaccine is recommended by the AHA for patients with “coronary artery disease and other forms of atherosclerotic disease.”38 These patients presumably include symptomatic carotid disease, but this definition leaves out the other 80–90% of stroke patients without symptomatic carotid disease, including, for example, those with lacunar or cryptogenic strokes. Including stroke patients in the category of CHD risk equivalents, as suggested in this review, however, would emphasize the importance of treating stroke patients with flu vaccination, along with those with atherosclerotic heart disease, diabetes, and chronic kidney disease.39
The authors thank Dr. Ralph Sacco for providing suggestions for studies to review and for helpful comments on the manuscript.
Funding for this work comes from the NIH/NINDS.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflict of Interest
Dr. Elkind receives research support from BMS-Sanofi and diaDexus, Inc; compensation for consulting work from Boehringer-Ingelheim, Daiichi-Sankyo, GlaxoSmithKline, Jarvik Heart, Novartis, Tethys Biosciences, and Pfizer; and honoraria for lecturing from BMS-Sanofi and Boehringer-Ingelheim. Dr. Dhamoon has no disclosures.