utility of framingham coronary heart disease risk score...

9
2942 S troke patients often harbor asymptomatic coronary atherosclerosis. 1 Beyond the acute period, stroke sur- vivors are at a higher risk for death attributable to car- diac rather than recurrent cerebrovascular events. 2,3 The high risk of coronary events among stroke survivors 4 has prompted a discussion of whether stroke should be con- sidered a coronary heart disease (CHD) risk equivalent, along with symptomatic carotid artery disease, peripheral arterial disease, abdominal aortic aneurysm, and diabetes mellitus. 5–9 Although symptomatic carotid artery disease is among the listed CHD risk equivalents, atherosclerosis develops in many sites within the aortocervicocephalic cir- culation other than the carotid artery, 10 and several race–eth- nic minorities who tend to have more strokes because of atherosclerosis in the intracranial arterial circulation 11 may be overlooked by this approach. A review of hospital-based and population-based studies appears to support the des- ignation of stroke as an additional CHD risk equivalent. 7 However, the review included individuals with known CHD, giving little information for the risk of future cardiac events among those without known CHD. 12,13 In addition, consider- ing every stroke as a CHD risk equivalent could potentially expose individuals with a low risk of subsequent coronary events to unnecessary treatments. Stroke is substantially more heterogeneous in etiology than myocardial infarction (MI) and includes diverse nonatherosclerotic mechanisms such as vasculitis, migraine, dissection, and hypercoagu- lable states. Intensively treating all stroke patients with therapies like statins could have untoward clinical impli- cations without conferring benefit. Therefore, there have been calls to develop strategies to identify patients at high risk who could benefit from CHD-specific treatment. 8,9,12 A recent American Heart Association scientific statement highlighted the need to develop a validated risk prediction instrument for risk of cardiac disease after stroke. 14 Coronary risk stratification among stroke patients would allow practitioners to develop a more effective, targeted, and cost-effective preventive strategy. 8 Although recurrent stroke and CHD preventive therapies are similar, key differences exist. For example, β-blockers are an integral part of CHD Background and Purpose—Coronary heart disease (CHD) is a major cause of mortality among stroke patients after the acute period. Simple risk stratification of stroke patients without known CHD may permit prompt implementation of CHD-specific management strategies for those who are at high risk for cardiac events. We assessed the utility of the Framingham Coronary Heart Disease Risk Score (FCRS) as a prognosticator in stroke patients without known CHD. Methods—Post hoc analysis of a trial dataset of 3509 recent ischemic stroke patients who were aged 35 years or older, recruited from 56 centers, and followed-up for 2 years. Patients were categorized as having known CHD, high FCRS (20%), and low/intermediate FCRS (<20%). The predictive values between baseline FCRS and primary (myocardial infarction [MI]), secondary (MI or vascular death), and tertiary (recurrent stroke) outcomes were assessed in multivariate analyses. Results—Rates of first MI at 2 years were 6.34%, 4.65%, and 1.44% for the known CHD, high FCRS, and low/intermediate FCRS groups. Compared with stroke patients with low/intermediate FCRS, individuals with high FCRS had a higher risk of MI (adjusted hazard ratio, 3.70; 95% confidence interval, 2.14–6.38) and MI or vascular death (adjusted hazard ratio, 2.21; 95% confidence interval, 1.48–3.28). High FCRS did not predict recurrent stroke. Conclusion—Among patients with a recent ischemic stroke without known CHD, high FCRS was associated with a higher risk of MI and vascular death, but not stroke. FCRS could be a simple way to identify recent stroke patients who may benefit from additional CHD-specific management. (Stroke. 2012;43:2942-2947.) Key Words: coronary atherosclerosis coronary heart disease myocardial infarction outcome risk stroke vascular death Utility of Framingham Coronary Heart Disease Risk Score for Predicting Cardiac Risk After Stroke Amytis Towfighi, MD; Daniela Markovic, MS; Bruce Ovbiagele, MD, MS Received June 18, 2012; accepted July 11, 2012. From the Department of Neurology, University of Southern California, Los Angeles, (A.T.); Department of Neurology, Rancho Los Amigos National Rehabilitation Center, Los Angeles, CA (A.T.); Department of Biomathematics, University of California, Los Angeles, (D.M.); Department of Neurosciences Medical University of South Carolina, Charleston, SC (B.O). The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.112. 668319/-/DC1. Correspondence to Amytis Towfighi, MD, 7601 E Imperial Highway, HB145, Downey, CA 90242. E-mail towfi[email protected] © 2012 American Heart Association, Inc. Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.112.668319 by guest on July 12, 2018 http://stroke.ahajournals.org/ Downloaded from by guest on July 12, 2018 http://stroke.ahajournals.org/ Downloaded from by guest on July 12, 2018 http://stroke.ahajournals.org/ Downloaded from by guest on July 12, 2018 http://stroke.ahajournals.org/ Downloaded from

Upload: dodiep

Post on 25-Jun-2018

216 views

Category:

Documents


1 download

TRANSCRIPT

2942

Stroke patients often harbor asymptomatic coronary atherosclerosis.1 Beyond the acute period, stroke sur-

vivors are at a higher risk for death attributable to car-diac rather than recurrent cerebrovascular events.2,3 The high risk of coronary events among stroke survivors4 has prompted a discussion of whether stroke should be con-sidered a coronary heart disease (CHD) risk equivalent, along with symptomatic carotid artery disease, peripheral arterial disease, abdominal aortic aneurysm, and diabetes mellitus.5–9 Although symptomatic carotid artery disease is among the listed CHD risk equivalents, atherosclerosis develops in many sites within the aortocervicocephalic cir-culation other than the carotid artery,10 and several race–eth-nic minorities who tend to have more strokes because of atherosclerosis in the intracranial arterial circulation11 may be overlooked by this approach. A review of hospital- based and population- based studies appears to support the des-ignation of stroke as an additional CHD risk equivalent.7 However, the review included individuals with known CHD, giving little information for the risk of future cardiac events

among those without known CHD.12,13 In addition, consider-ing every stroke as a CHD risk equivalent could potentially expose individuals with a low risk of subsequent coronary events to unnecessary treatments. Stroke is substantially more heterogeneous in etiology than myocardial infarction (MI) and includes diverse nonatherosclerotic mechanisms such as vasculitis, migraine, dissection, and hypercoagu-lable states. Intensively treating all stroke patients with therapies like statins could have untoward clinical impli-cations without conferring benefit. Therefore, there have been calls to develop strategies to identify patients at high risk who could benefit from CHD- specific treatment.8,9,12 A recent American Heart Association scientific statement highlighted the need to develop a validated risk prediction instrument for risk of cardiac disease after stroke.14

Coronary risk stratification among stroke patients would allow practitioners to develop a more effective, targeted, and cost- effective preventive strategy.8 Although recurrent stroke and CHD preventive therapies are similar, key differences exist. For example, β-blockers are an integral part of CHD

Background and Purpose—Coronary heart disease (CHD) is a major cause of mortality among stroke patients after the acute period. Simple risk stratification of stroke patients without known CHD may permit prompt implementation of CHD- specific management strategies for those who are at high risk for cardiac events. We assessed the utility of the Framingham Coronary Heart Disease Risk Score (FCRS) as a prognosticator in stroke patients without known CHD.

Methods—Post hoc analysis of a trial dataset of 3509 recent ischemic stroke patients who were aged 35 years or older, recruited from 56 centers, and followed- up for 2 years. Patients were categorized as having known CHD, high FCRS (≥20%), and low/intermediate FCRS (<20%). The predictive values between baseline FCRS and primary (myocardial infarction [MI]), secondary (MI or vascular death), and tertiary (recurrent stroke) outcomes were assessed in multivariate analyses.

Results—Rates of first MI at 2 years were 6.34%, 4.65%, and 1.44% for the known CHD, high FCRS, and low/intermediate FCRS groups. Compared with stroke patients with low/intermediate FCRS, individuals with high FCRS had a higher risk of MI (adjusted hazard ratio, 3.70; 95% confidence interval, 2.14–6.38) and MI or vascular death (adjusted hazard ratio, 2.21; 95% confidence interval, 1.48–3.28). High FCRS did not predict recurrent stroke.

Conclusion—Among patients with a recent ischemic stroke without known CHD, high FCRS was associated with a higher risk of MI and vascular death, but not stroke. FCRS could be a simple way to identify recent stroke patients who may benefit from additional CHD- specific management. (Stroke. 2012;43:2942-2947.)

Key Words: coronary atherosclerosis ◼ coronary heart disease ◼ myocardial infarction ◼ outcome ◼ risk ◼ stroke ◼ vascular death

Utility of Framingham Coronary Heart Disease Risk Score for Predicting Cardiac Risk After Stroke

Amytis Towfighi, MD; Daniela Markovic, MS; Bruce Ovbiagele, MD, MS

Received June 18, 2012; accepted July 11, 2012.From the Department of Neurology, University of Southern California, Los Angeles, (A.T.); Department of Neurology, Rancho Los Amigos National

Rehabilitation Center, Los Angeles, CA (A.T.); Department of Biomathematics, University of California, Los Angeles, (D.M.); Department of Neurosciences Medical University of South Carolina, Charleston, SC (B.O).

The online- only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.112. 668319/-/DC1.

Correspondence to Amytis Towfighi, MD, 7601 E Imperial Highway, HB145, Downey, CA 90242. E-mail [email protected]© 2012 American Heart Association, Inc.

Stroke is available at http://stroke.ahajournals.org DOI: 10.1161/STROKEAHA.112.668319

preventive regimens but may be less effective for stroke prevention compared with other classes of antihypertensive agents.15,16 Furthermore, although statins are recommended for CHD prevention, their utility in nonatherosclerotic stroke remains in question.17

The Framingham Coronary Heart Disease Risk Score (FCRS) is a simple, widely recognized, readily available tool for estimating 10-year risk for CHD outcomes in the general population.18 The utility of FCRS in predicting risk of coronary events in the short- term to intermediate-term among stroke patients is unknown. This study aimed to assess whether base-line FCRS could delineate the risk of subsequent (1) MI, (2) MI or vascular death, and (3) recurrent stroke among recent ischemic stroke patients.

Subjects and MethodsPatient PopulationWe conducted a post hoc analysis of the Vitamin Intervention for Stroke Prevention (VISP) trial, a double- blind, randomized, con-trolled trial assessing whether best medical therapy and high- dose folic acid, pyridoxine, and cobalamin administered to lower homo-cysteine levels would reduce the incidence of recurrent stroke in patients with a nondisabling ischemic stroke within the preceding 120 days.19 Details of the trial protocol and main results have been published previously.19,20 In brief, from September 1996 to May 2003, VISP enrolled 3680 recent ischemic stroke patients aged 35 years or older from centers across the United States (n=45), Canada (n=10), and Scotland (n=1). Ischemic stroke was defined as brain infarction characterized by the sudden onset of a neurological deficit lasting ≥24 hours or evident on computed tomography or magnetic resonance imaging. Infarcts attributable to cardioembo-lism (atrial fibrillation within 30 days of stroke, prosthetic cardiac valve, intracardiac thrombus or neoplasm, or valvular vegetation) were excluded.19,20 One month after randomization, at 6 months, and every 6 months thereafter (up to 24 months), participants returned for evaluation including interview, examination, medication use as-sessment, stroke symptom questionnaire, stroke scales, and medical follow- up questionnaire. All subjects received best available medi-cal and surgical management as determined by their primary physi-cian, including risk factor modification and often aspirin 325 mg daily. A Cardiovascular End Point Review Committee adjudicated CHD end points. Myocardial infarction was defined by new ECG changes including Q waves or marked ST- T changes plus abnor-mal cardiac enzymes, cardiac symptoms plus abnormal enzymes, or symptoms plus hyperacute ECG changes resolving with thromboly-sis. A Cerebrovascular End Point Review Committee adjudicated recurrent stroke end points. Recurrent stroke was defined by sudden onset of neurological symptoms lasting ≥24 hours with an increase in the National Institutes of Health Stroke Scale on a section not abnormal in the previous examination. When symptoms were not accompanied by an increased National Institutes of Health Stroke Scale score in an area that was previously normal, recurrent stroke was diagnosed using cranial computed tomography or magnetic resonance imaging evidence of new infarction consistent with the clinical presentation. The Ethics Committee or Institutional Review Board at each site approved the trial, and all participants provided written informed consent. VISP did not find high- dose vitamin ther-apy superior to the low- dose regimen in recurrent stroke preven-tion, so data for all enrolled patients were combined and included in these analyses.19

Statistical AnalysisPatients were categorized into 3 categories: known CHD (defined as history of MI, angina, coronary angioplasty or stenting, or coro-nary artery bypass graft surgery); high FCRS (FCRS ≥20%); and low/intermediate FCRS (FCRS <20%). FCRS was determined using

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

Towfighi et al Cardiac Risk Prediction after Stroke 2943

preventive regimens but may be less effective for stroke prevention compared with other classes of antihypertensive agents.15,16 Furthermore, although statins are recommended for CHD prevention, their utility in nonatherosclerotic stroke remains in question.17

The Framingham Coronary Heart Disease Risk Score (FCRS) is a simple, widely recognized, readily available tool for estimating 10-year risk for CHD outcomes in the general population.18 The utility of FCRS in predicting risk of coronary events in the short- term to intermediate-term among stroke patients is unknown. This study aimed to assess whether base-line FCRS could delineate the risk of subsequent (1) MI, (2) MI or vascular death, and (3) recurrent stroke among recent ischemic stroke patients.

Subjects and MethodsPatient PopulationWe conducted a post hoc analysis of the Vitamin Intervention for Stroke Prevention (VISP) trial, a double- blind, randomized, con-trolled trial assessing whether best medical therapy and high- dose folic acid, pyridoxine, and cobalamin administered to lower homo-cysteine levels would reduce the incidence of recurrent stroke in patients with a nondisabling ischemic stroke within the preceding 120 days.19 Details of the trial protocol and main results have been published previously.19,20 In brief, from September 1996 to May 2003, VISP enrolled 3680 recent ischemic stroke patients aged 35 years or older from centers across the United States (n=45), Canada (n=10), and Scotland (n=1). Ischemic stroke was defined as brain infarction characterized by the sudden onset of a neurological deficit lasting ≥24 hours or evident on computed tomography or magnetic resonance imaging. Infarcts attributable to cardioembo-lism (atrial fibrillation within 30 days of stroke, prosthetic cardiac valve, intracardiac thrombus or neoplasm, or valvular vegetation) were excluded.19,20 One month after randomization, at 6 months, and every 6 months thereafter (up to 24 months), participants returned for evaluation including interview, examination, medication use as-sessment, stroke symptom questionnaire, stroke scales, and medical follow- up questionnaire. All subjects received best available medi-cal and surgical management as determined by their primary physi-cian, including risk factor modification and often aspirin 325 mg daily. A Cardiovascular End Point Review Committee adjudicated CHD end points. Myocardial infarction was defined by new ECG changes including Q waves or marked ST- T changes plus abnor-mal cardiac enzymes, cardiac symptoms plus abnormal enzymes, or symptoms plus hyperacute ECG changes resolving with thromboly-sis. A Cerebrovascular End Point Review Committee adjudicated recurrent stroke end points. Recurrent stroke was defined by sudden onset of neurological symptoms lasting ≥24 hours with an increase in the National Institutes of Health Stroke Scale on a section not abnormal in the previous examination. When symptoms were not accompanied by an increased National Institutes of Health Stroke Scale score in an area that was previously normal, recurrent stroke was diagnosed using cranial computed tomography or magnetic resonance imaging evidence of new infarction consistent with the clinical presentation. The Ethics Committee or Institutional Review Board at each site approved the trial, and all participants provided written informed consent. VISP did not find high- dose vitamin ther-apy superior to the low- dose regimen in recurrent stroke preven-tion, so data for all enrolled patients were combined and included in these analyses.19

Statistical AnalysisPatients were categorized into 3 categories: known CHD (defined as history of MI, angina, coronary angioplasty or stenting, or coro-nary artery bypass graft surgery); high FCRS (FCRS ≥20%); and low/intermediate FCRS (FCRS <20%). FCRS was determined using

formulas published by Wilson et al.18 The low and intermediate FCRS groups were combined for the purpose of this analysis, because clini-cal management is typically similar for low and intermediate groups. Of 3680 participants in the trial, 171 participants were missing a com-ponent of FCRS and were excluded from the final analysis, yielding a total of 3509 subjects. We compared baseline clinical factors and as-sessed rates of MI, stroke, vascular death, and all- cause death among the 3 groups.

We assessed the prevalence and clinical predictors of high FCRS in VISP subjects without known CHD. Baseline demographic and clinical covariates were preselected based on previous studies of factors that influence vascular events after ischemic stroke. We used multivariate logistic regression to examine the association between the following variables simultaneously and cardiac risk status: race (black, white, other); history of stroke; severity of index stroke (National Institutes of Health Stroke Scale score: 0, 1–5, ≥5); history of congestive heart failure; history of carotid endarterectomy; body mass index (BMI; normal, overweight, obese); alcohol use in the previous year; antidyslipidemic use; antithrombotic use; low- density lipoprotein cholesterol level >100 mg/dL; and triglyceride level >150 mg/dL. Final models were selected using backwards stepwise search with P<0.25 as the retention criterion. We examined the as-sociation between baseline factors and high FCRS in the following ways: (1) without adjustment for any of the individual FCRS com-ponents (model I); (2) after adding age to the model (model II); and (3) after adding both age and sex to the model (model III). Age was added to the models because FCRS places a large weight on age as a risk factor.21

We used Cox models with competing risks to assess the indepen-dent association of baseline cardiac risk status with the primary (MI), secondary (MI or vascular death), and tertiary (stroke) outcomes, adjusting for covariates using backward stepwise search with <0.25 as the retention criterion. The Cox models for MI and stroke were adjusted for the competing risk of all- cause mortality. The Cox model for MI or vascular death was adjusted for the competing risk of non-vascular mortality. Subjects who did not experience the event of in-terest or the corresponding competing event were censored. For each outcome, we assessed the association with and without age added to the model. We computed the area under the curve (concordance) as a measure of prediction accuracy for each model.

To determine if assessing FCRS as a continuous variable would provide more useful information than assessing FCRS as a categori-cal variable, we assessed linearity between the FCRS continuous score vs the risk of MI by fitting a restricted cubic splines model with 5 knots. To compare the nonlinear model containing the spline terms to the linear model not containing the spline terms, we com-puted the probability value using the likelihood ratio test with 3 de-grees of freedom. Results showed that the nonlinear model tended to fit the data better than the linear model (P=0.056). Although lin-earity was approximately satisfied for a FCRS range of ≈10% or higher, linearity did not seem to hold at FCRS <10%, where slope was close to zero. Therefore, we decided to use a categorical model with a clinically defined FCRS threshold of low/intermediate risk vs high risk.

To evaluate whether FCRS conveys, as originally postulated, greater information than the sum of its parts, we compared mul-tivariate models containing 6 non- FCRS vascular risk factors sig-nificantly associated with outcome using backwards stepwise search plus: (1) individual FCRS components most strongly associated with outcome (model I); (2) FCRS (model II); and (3) FCRS plus FCRS components most strongly associated with outcome (model III). Models were compared using Akaike information criteria. The overall significance level for the study was P<0.05 using a 2-sided test. We computed the concordance statistic (area under the curve) to assess prediction accuracy of each model as per the procedure proposed by Wolbers et al.22 The aforementioned model accuracy statistics were calculated after 5-fold cross- validation to obtain more accurate estimates of model performance while allowing us to use all the data. Model fit also was assessed using modifications of methods by Hosmer and Lemeshow.23 We assessed the interac-tion between VISP treatment category and high FCRS in predicting

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

2944 Stroke November 2012

the risk of MI by including the appropriate interaction terms in the model.All analyses were performed with SAS statistical software version 9.2.

ResultsComparing baseline characteristics between patients with and without missing FCRS, those with missing values had a slightly higher systolic blood pressure (143.9 vs 141.2 mm Hg; P=0.01), were more likely to have diabetes (32% vs 26%; P=0.07), and were less likely to use lipid modifiers (29% vs 38%; P=0.07). Of the 3509 individuals included in this analysis, 31% had known CHD. Among subjects without known CHD, prevalence of high FCRS was 37% (n=933).

Table 1 shows the baseline characteristics of individu-als with low/intermediate FCRS, high FCRS, and known CHD. Compared with those with low/intermediate FCRS, individuals with high FCRS were less likely to have National Institutes of Health Stroke Scale score 0 (29.8% vs 35.7%; P=0.01), more likely to be overweight (44.0% vs 39.4%; P=0.02), and had higher mean triglyceride (191.8 vs 162.0 mg/dL; P<0.001) and mean low- density lipoprotein cholesterol levels (131.3 vs 120.9 mg/dL; P<0.001; Table 1).

Several baseline factors were associated with high FCRS (Supplementary Table I). The following factors were simultaneously associated with high FCRS in all 3 multivariate

Table 1. Baseline Characteristics of Recent Ischemic Stroke Patients by Cardiac Risk Category (n=3509)

Variable(s)

Low/Intermediate Framingham Coronary Risk Score (n=1614)

High Framingham Coronary Risk Score (n=933)

Known Coronary Heart Disease

(n = 962)

n % n % n % P Value

Female 905 56.1 133 14.3 276 28.7 <0.001

Race 0.06

Black 262 16.2 132 14.1 116 12.1

Other 88 5.5 57 6.1 57 5.9

White 1264 78.3 744 79.7 789 82.0

Current smoker 228 14.1 216 23.2 145 15.1 <0.001

Stroke >90 d 506 31.4 258 27.7 308 32.0 0.08

History of stroke 336 20.8 218 23.4 255 26.5 0.004

NIHSS* 0.05

0 577 35.7 278 29.8 324 33.7

1–4 913 56.6 579 62.1 559 58.1

≥5 124 7.7 76 8.1 79 8.2

History of congestive heart failure 39 2.4 24 2.6 124 12.9 <0.001

Antidyslipidemic at baseline 614 38.0 357 38.3 583 60.6 <0.001

Antithrombotic at baseline 754 46.7 457 49.0 613 63.7 <0.001

History of carotid endarterectomy 76 4.7 60 6.4 105 10.9 <0.001

History of diabetes 290 18.0 367 39.4 359 37.4 <0.001

History of hypertension 1154 71.7 705 75.7 738 76.8 0.008

BMI, kg/m2* 0.02

<25 486 30.5 235 25.4 253 26.5

25–30 628 39.4 407 44.0 389 40.7

>30 480 30.1 282 30.5 314 32.8

Alcohol use in previous year 950 60.5 560 61.5 520 55.4 0.01

Mean SD Mean SD Mean SD

Age, y 63.4 11.5 69.2 8.9 68.2 9.8 <0.001

SBP at baseline, mm Hg 137.1 17.8 148.2 17.5 139.4 19.2 <0.001

DBP at baseline, mm Hg 77.5 9.8 79.9 10.0 76.3 10.4 <0.001

LDL at baseline, mg/dL* 120.9 40.0 131.3 39.6 116.3 39.5 <0.001

Total cholesterol at baseline, mg/dL

202.2 48.6 210.5 42.8 192.4 45.4 <0.001

HDL at baseline, mg/dL 50.1 17.2 40.1 10.9 42.5 13.7 <0.001

Triglycerides at baseline, mg/dL* 162.0 183.8 191.8 116.2 179.6 127.0 <0.001

BMI indicates body mass index; CHD, coronary heart disease; DBP, diastolic blood pressure; FCRS, Framingham Coronary Heart Disease Risk Score; HDL, high- density lipoprotein; LDL, low- density lipoprotein; NIHSS, National Institutes of Health Stroke Scale; SBP, systolic blood pressure; SD, standard deviation.

NIHSS scores range from 0 to 42, with higher scores on the scale indicating worse stroke severity.*Significant difference between low/intermediate and high FCRS (P<0.05).

models (regardless of adjustment for age and sex): National Institutes of Health Stroke Scale score 1 to 4 vs 0; triglyceride level >150 mg/dL; and low- density lipoprotein level >100 mg/dL (P≤0.05). Persons of black/other race and obese individu-als were more likely to have high FCRS after also controlling for age and sex.

At 2 years, 104 individuals had MI, 190 had a stroke, and 166 had MI or died of vascular causes. At 2 years, risks of MI in individuals with low/intermediate FCRS, high FCRS, and known CHD were 1.44%, 4.65%, and 6.34%, risks of stroke were 7.63%, 9.72%, and 9.01%, and risks of vascular death were 1.87%, 3.05%, and 5.91% (Table 2).

The final multivariate model showed that 6 non- FCRS factors were associated with higher risk of MI: history of stroke; congestive heart failure; abstinence from alcohol in past year; history of carotid endarterectomy; lower BMI (<25 kg/m2); and not using lipid modifiers. Compared with individuals with low/intermediate FCRS, those with high

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

Towfighi et al Cardiac Risk Prediction after Stroke 2945

models (regardless of adjustment for age and sex): National Institutes of Health Stroke Scale score 1 to 4 vs 0; triglyceride level >150 mg/dL; and low- density lipoprotein level >100 mg/dL (P≤0.05). Persons of black/other race and obese individu-als were more likely to have high FCRS after also controlling for age and sex.

At 2 years, 104 individuals had MI, 190 had a stroke, and 166 had MI or died of vascular causes. At 2 years, risks of MI in individuals with low/intermediate FCRS, high FCRS, and known CHD were 1.44%, 4.65%, and 6.34%, risks of stroke were 7.63%, 9.72%, and 9.01%, and risks of vascular death were 1.87%, 3.05%, and 5.91% (Table 2).

The final multivariate model showed that 6 non- FCRS factors were associated with higher risk of MI: history of stroke; congestive heart failure; abstinence from alcohol in past year; history of carotid endarterectomy; lower BMI (<25 kg/m2); and not using lipid modifiers. Compared with individuals with low/intermediate FCRS, those with high

FCRS had a higher risk of the primary outcome (MI) after adjusting for these 6 factors (hazard ratio [HR], 3.70, 95% confidence interval [CI], 2.14–6.38). Further adjustment for age attenuated the results (HR, 3.33; 95% CI, 1.90–5.84; Table 3; Figure). The effect of high FCRS was slightly more pronounced for patients receiving low- dose vitamin therapy (HR, 5.63) than for patients receiving high- dose vitamin ther-apy (HR, 2.51); however, this interaction was not significant (P=0.14). Compared with those with low/intermediate FCRS, individuals with known CHD were more likely to have a sub-sequent MI (HR, 4.43; 95% CI, 2.58–7.59). The association persisted after further adjustment for age (HR, 4.05; 95% CI, 2.34–7.01; Table 3; Figure).

The final multivariate model assessing the association between high FCRS and risk of MI or vascular death adjusted for the following covariates that were significant in back-wards stepwise regression model: previous stroke; mean systolic blood pressure in trial; congestive heart failure;

Table 2. Unadjusted Clinical Outcomes at 2 Years by Cardiac Risk Category Among 3509 Patients With a Recent Ischemic Stroke

Cardiac Risk Status

Variable Low/Intermediate FCRS High FCRS Known CHD

N of patients 1614 933 962

MI (risk at 2 y, %; 95% CI) 1.44 (0.82−2.05) 4.65 (3.26–6.05) 6.34 (4.74–7.94)

Stroke (risk at 2 y, %; 95% CI) 7.63 (6.29–8.96) 9.72 (7.75–11.68) 9.01 (7.12–10.89)

Stroke, MI, or vascular death (risk at 2 y, %; 95% CI)

9.84 (8.35–11.33) 15.05 (12.69–17.40) 17.76 (15.27–20.26)

All- cause death (risk at 2 y, %; 95% CI) 4.19 (3.15–5.23) 6.72 (5.04–8.40) 10.01 (8.01–12.01)

Vascular death (risk at 2 y, %; 95% CI) 1.87 (1.16–2.57) 3.05 (1.91–4.19) 5.91 (4.34–7.48)

CHD indicates coronary heart disease; CI, confidence interval; FCRS, Framingham Coronary Heart Disease Risk Score; MI, myocardial infarction.

Table 3. Multivariate Models Assessing Risk of Myocardial Infarction, Myocardial Infarction and Vascular Death, and Stroke in Individuals With High Framingham Coronary Risk Score or Known Coronary Heart Disease vs Low/Intermediate Framingham Coronary Risk Score

Sample Size, n

N of Events

High Framingham Coronary Risk Score

Known Coronary Heart Disease

Cross- Validated AUCHR (95% CI) P Value HR (95% CI) P Value

Primary

MI* 3384 113 3.70 (2.14–6.38) <0.001 4.43 (2.58–7.59) <0.001 70.1%

MI§ 3384 113 3.33 (1.90–5.84) <0.001 4.05 (2.34–7.01) <0.001 71.0%

Secondary

MI or vascular death† 3470 190 2.21 (1.48–3.28) <0.001 3.37 (2.30–4.92) <0.001 69.3%

MI or vascular death§ 3470 190 1.96 (1.31–2.95) <0.001 3.06 (2.09–4.47) <0.001 70.5%

Tertiary

Stroke‡ 3416 274 1.19 (0.89–1.58) 0.24 1.06 (0.79–1.42) 0.71 63.4%

Stroke§ 3416 274 1.16 (0.87–1.55) 0.31 1.04 (0.77–1.40) 0.81 63.4%

AUC indicates area under the curve; CI, confidence interval; HR, hazard ratio; MI, myocardial infarction.*Adjusted for previoius stroke, coronary heart failure, alcohol intake, carotid endarterectomy, body mass index, and antidyslipidemic use.†Adjusted for prior stroke, mean in trial systolic blood pressure, coronary heart failure, carotid endarterectomy, body mass index, stroke severity, antidyslipidemic

use, and antithrombotic use.‡Adjusted for previous stroke, mean in trial systolic blood pressure, coronary heart failure, alcohol intake, carotid endarterectomy, stroke severity, and

antithrombotic use.§Adjusted for aforementioned covariates plus age.

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

2946 Stroke November 2012

carotid endarterectomy; BMI; stroke severity; antidyslipid-emic use; and antithrombotic use. After adjusting for those 8 covariates, high FCRS was associated with higher risk of MI or vascular death compared with low/intermediate FCRS (HR, 2.21; 95% CI, 1.48–3.28; Table 3). Further adjustment for age attenuated the magnitude (HR, 1.96; 95% CI, 1.31–2.95). Individuals with known CHD had a higher risk of subsequent MI or vascular death compared with those with low/interme-diate FCRS (HR, 3.37; 95% CI, 2.30–4.92). Adjustment for age attenuated the results (HR, 3.06; 95% CI, 2.09–4.47).

After multivariate analysis, neither high FCRS nor known CHD was associated with a higher risk of stroke compared with low/intermediate FCRS. The model for predicting the risk of MI had the highest discrimination ( cross- validated area under the curve, 70.1%), whereas the model for predicting the risk of stroke had the lowest discrimination ( cross- validated area under the curve, 63.4%; Table 3).

The individual FCRS components associated with risk of MI after adjusting for previous stroke, congestive heart fail-ure, alcohol intake, carotid endarterectomy, BMI, and anti-dyslipidemic use were age, male sex, diabetes, and smoking (Supplementary Table II, model I). Compared with low/inter-mediate FCRS, high FCRS was associated with higher inci-dence of MI, even after adjustment for the FCRS predictor (Supplementary Table II, model II) and FCRS components that were significantly associated with outcome (HR, 2.17; 95% CI, 1.16–4.09; Supplementary Table II, model III). The model that included the individual FCRS components without the FCRS predictor had slightly but not significantly better discrimination than the model that included the FCRS pre-dictor alone without including the individual FCRS compo-nents ( cross- validated area under the curve: 74.0% vs 70.1%, P=0.17; Supplementary Table II). The addition of FCRS to the model including the individual FCRS components did not further improve model prediction ( cross- validated area under the curve: 74.4% vs 74.0%, P=0.91; Supplementary Table II).

DiscussionIn this analysis of patients with a recent ischemic stroke, more than one- third of those without CHD had high FCRS. High

FCRS was associated with a nearly 4-fold higher risk of MI and ≈2-fold risk of MI or vascular death over a 2-year period, after adjusting for numerous sociodemographic and clinical factors. These findings suggest that beyond its established utility as a prognosticator of long- term CHD risk in the gen-eral population, FCRS also might be a useful tool for predict-ing intermediate- term MI risk among individuals with recent stroke. Even though all FCRS components are risk factors for both MI and stroke, and although we studied a population with known stroke without known CHD, high FCRS was not asso-ciated with risk of recurrent stroke. The null finding for any link between FCRS and stroke is not surprising because the score was originally developed and validated for CHD events. FCRS does not include stroke- specific risk factors such as atrial fibrillation (which is included in the Framingham Stroke Risk Score).24 Interestingly, the effect of high FCRS on inci-dence of MI remained even after adjusting for its individual components, suggesting that contribution of high FCRS to MI risk after stroke may be greater than the sum of its traditional vascular risk factor parts.

These findings suggest that FCRS may be a useful tool for identifying stroke survivors with a high risk of MI or vas-cular death within the next 2 years. These individuals may benefit from additional CHD- specific agents or further diag-nostic evaluation for silent cardiac ischemia, a condition that confers the same poor prognosis as symptomatic ischemia.25 Guidelines have not been established for treatment of stroke patients with asymptomatic CHD or high risk of future car-diac disease; however, counseling, diagnostic evaluation, and CHD- specific risk factor control have been suggested.25

Beyond the issue of additional CHD- specific treatment in stroke patients with high FCRS is whether tighter control of FCRS components or modifications of vascular risk factors linked to high FCRS may boost CHD outcomes after stroke. Among modifiable FCRS components, diabetes and smoking were independently linked to incident MI. In addition, high FCRS was associated with modifiable non- FCRS vascular risk factors, including higher BMI, triglyceride, and low- density lipoprotein cholesterol levels. The SPARCL trial supports an approach of intensive risk factor treatment to avert CHD events in recent stroke patients without known CHD. In SPARCL, individuals with a history of TIA/stroke without symptomatic CHD were treated with high- dose atorvastatin (vs placebo) and experienced a substantial reduction in risk of a major coronary events (defined as death from cardiac causes, nonfatal MI, or resuscitation after cardiac arrest; HR, 0.65; 95% CI, 0.49–0.87), an acute coronary event (HR, 0.65; 95% CI, 0.50–0.84), and any coronary event (odds ratio, 0.58; 95% CI, 0.46–0.73).26 Low- density lipoprotein cholesterol levels declined by 45% in the high- dose statin group vs 4% in the placebo group. Of note, the high- dose statin appeared more efficacious in preventing the secondary outcome of major CHD events than it was in preventing the primary outcome of recurrent stroke.

This study has limitations. First, because it is a retrospec-tive analysis of participants with noncardioembolic stroke enrolled in a clinical trial, the findings may not be general-izable to the general stroke population. Individuals with car-dioembolic stroke would be expected to have a higher risk

Figure. Adjusted cumulative incidence of myocardial infarction (MI) after recent ischemic stroke among individuals with low/intermediate Framingham Coronary Heart Disease Risk Score (FCRS), high FCRS, and known coronary heart disease (CHD).

of cardiac events, possibly because of underlying cardiac disease.14 Second, although we conducted multivariate anal-yses, unmeasured confounding is possible. For example, although we controlled for baseline medication use, we were unable to control for medication use during the trial. In addi-tion, the post hoc nature of the study limited us to examining only the variables collected in the original study. Furthermore, the VISP trial was conducted mainly in the early 2000s, so its relevance to stroke patients encountered today could be ques-tioned. However, the baseline characteristics (mean age, sex, pretrial symptomatic cerebrovascular disease, known diabetes or hypertension, mean BMI, current smoking, and baseline blood pressure) of a recent large secondary stroke prevention trial27 are comparable with those seen in VISP. Finally, the lack of association between FCRS and recurrent stroke risk is perhaps not surprising because FCRS has not been shown to predict stroke risk in the general population. Further stud-ies could assess if the Framingham Stroke Risk Score24 or FCRS28 are predictive of risk of recurrent stroke among stroke survivors. The strengths of the study include the rigorous procedures of the VISP trial design, multiple measurements/contacts for each subject, and consistency of effect direction across primary, secondary, and tertiary outcomes.

AcknowledgmentsThe authors are grateful to the Vitamin in Stroke Prevention (VISP) Investigators (NINDS R01 NS34447), especially Dr Chambless for making the dataset available to the authors.

Sources of Funding NIH- NINDS (U01 NS079179) and American Heart Association National Scientist Development Program.

DisclosuresThe abstract of this study was the winner of the 2012 Robert G. Siekert Award and was presented in oral form at the Plenary Session of American Heart Association International Stroke Conference on February 1, 2012, in New Orleans, Louisiana.

References 1. Amarenco P, Lavallee PC, Labreuche J, Ducrocq G, Juliard JM, Feldman

L, et al. Prevalence of coronary atherosclerosis in patients with cerebral infarction. Stroke. 2011;42:22–29.

2. Dhamoon MS, Sciacca RR, Rundek T, Sacco RL, Elkind MS. Recurrent stroke and cardiac risks after first ischemic stroke: The Northern Manhattan Study. Neurology. 2006;66:641–646.

3. Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, et al. Five- year survival after first- ever stroke and related prognostic factors in the Perth Community Stroke Study. Stroke. 2000;31:2080–2086.

4. Touze E, Varenne O, Chatellier G, Peyrard S, Rothwell PM, Mas JL. Risk of myocardial infarction and vascular death after transient isch-emic attack and ischemic stroke: A systematic review and meta- analysis. Stroke. 2005;36:2748–2755.

5. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). 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) final report. Circulation. 2002;106:3143–3421.

6. Dhamoon MS, Tai W, Boden- Albala B, Rundek T, Paik MC, Sacco RL, et al. Risk of myocardial infarction or vascular death after first ischemic stroke: The Northern Manhattan Study. Stroke. 2007;38:1752–1758.

7. Dhamoon MS, Elkind MS. Inclusion of stroke as an outcome and risk equivalent in risk scores for primary and secondary prevention of vascu-lar disease. Circulation. 2010;121:2071–2078.

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

Towfighi et al Cardiac Risk Prediction after Stroke 2947

of cardiac events, possibly because of underlying cardiac disease.14 Second, although we conducted multivariate anal-yses, unmeasured confounding is possible. For example, although we controlled for baseline medication use, we were unable to control for medication use during the trial. In addi-tion, the post hoc nature of the study limited us to examining only the variables collected in the original study. Furthermore, the VISP trial was conducted mainly in the early 2000s, so its relevance to stroke patients encountered today could be ques-tioned. However, the baseline characteristics (mean age, sex, pretrial symptomatic cerebrovascular disease, known diabetes or hypertension, mean BMI, current smoking, and baseline blood pressure) of a recent large secondary stroke prevention trial27 are comparable with those seen in VISP. Finally, the lack of association between FCRS and recurrent stroke risk is perhaps not surprising because FCRS has not been shown to predict stroke risk in the general population. Further stud-ies could assess if the Framingham Stroke Risk Score24 or FCRS28 are predictive of risk of recurrent stroke among stroke survivors. The strengths of the study include the rigorous procedures of the VISP trial design, multiple measurements/contacts for each subject, and consistency of effect direction across primary, secondary, and tertiary outcomes.

AcknowledgmentsThe authors are grateful to the Vitamin in Stroke Prevention (VISP) Investigators (NINDS R01 NS34447), especially Dr Chambless for making the dataset available to the authors.

Sources of Funding NIH- NINDS (U01 NS079179) and American Heart Association National Scientist Development Program.

DisclosuresThe abstract of this study was the winner of the 2012 Robert G. Siekert Award and was presented in oral form at the Plenary Session of American Heart Association International Stroke Conference on February 1, 2012, in New Orleans, Louisiana.

References 1. Amarenco P, Lavallee PC, Labreuche J, Ducrocq G, Juliard JM, Feldman

L, et al. Prevalence of coronary atherosclerosis in patients with cerebral infarction. Stroke. 2011;42:22–29.

2. Dhamoon MS, Sciacca RR, Rundek T, Sacco RL, Elkind MS. Recurrent stroke and cardiac risks after first ischemic stroke: The Northern Manhattan Study. Neurology. 2006;66:641–646.

3. Hankey GJ, Jamrozik K, Broadhurst RJ, Forbes S, Burvill PW, Anderson CS, et al. Five- year survival after first- ever stroke and related prognostic factors in the Perth Community Stroke Study. Stroke. 2000;31:2080–2086.

4. Touze E, Varenne O, Chatellier G, Peyrard S, Rothwell PM, Mas JL. Risk of myocardial infarction and vascular death after transient isch-emic attack and ischemic stroke: A systematic review and meta- analysis. Stroke. 2005;36:2748–2755.

5. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). 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) final report. Circulation. 2002;106:3143–3421.

6. Dhamoon MS, Tai W, Boden- Albala B, Rundek T, Paik MC, Sacco RL, et al. Risk of myocardial infarction or vascular death after first ischemic stroke: The Northern Manhattan Study. Stroke. 2007;38:1752–1758.

7. Dhamoon MS, Elkind MS. Inclusion of stroke as an outcome and risk equivalent in risk scores for primary and secondary prevention of vascu-lar disease. Circulation. 2010;121:2071–2078.

8. Touze E, Varenne O, Mas JL. Overestimation of coronary risk in stroke patients. Stroke. 2007;38:e98; author reply e99.

9. Chaturvedi S. Should stroke be considered both a brain attack and a heart attack? Stroke. 2007;38:1713–1714.

10. Roger VL, Go AS, Lloyd- Jones DM, Benjamin EJ, Berry JD, Borden WB, et al. Heart disease and stroke statistics–2012 update: A report from the American Heart Association. Circulation. 2012;125: e2–e220.

11. Rincon F, Sacco RL, Kranwinkel G, Xu Q, Paik MC, Boden- Albala B, et al. Incidence and risk factors of intracranial atherosclerotic stroke: The Northern Manhattan Stroke Study. Cerebrovasc Dis. 2009;28: 65–71.

12. Touze E, Varenne O, Calvet D, Mas JL. Coronary risk stratification in patients with ischemic stroke or transient ischemic stroke attack. Int J Stroke. 2007;2:177–183.

13. Amarenco P, Goldstein LB, Sillesen H, Benavente O, Zweifler RM, Callahan A III, et al. Coronary heart disease risk in patients with stroke or transient ischemic attack and no known coronary heart disease: Findings from the Stroke Prevention By Aggressive Reduction In Cholesterol Levels (SPARCL) Trial. Stroke. 2010;41:426–430.

14. Lackland DT, Elkind MS, D’Agostino R Sr, Dhamoon MS, Goff DC Jr, Higashida RT, et al. Inclusion of stroke in cardiovascular risk pre-diction instruments: A statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2012;43:1998–2027.

15. Khan N, McAlister FA. Re- examining the efficacy of beta- blockers for the treatment of hypertension: A meta- analysis. CMAJ. 2006;174: 1737–1742.

16. Wiysonge CS, Bradley H, Mayosi BM, Maroney R, Mbewu A, Opie LH, et al. Beta- blockers for hypertension. Cochrane Database Syst Rev. 2007;CD002003.

17. Furie KL, Kasner SE, Adams RJ, Albers GW, Bush RL, Fagan SC, et al. Guidelines for the prevention of stroke in patients with stroke or transient ischemic attack: A guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2011;42:227–276.

18. Wilson PW, D’Agostino RB, Levy D, Belanger AM, Silbershatz H, Kannel WB. Prediction of coronary heart disease using risk factor cat-egories. Circulation. 1998;97:1837–1847.

19. Toole JF, Malinow MR, Chambless LE, Spence JD, Pettigrew LC, Howard VJ, et al. Lowering homocysteine in patients with ischemic stroke to prevent recurrent stroke, myocardial infarction, and death: The Vitamin Intervention For Stroke Prevention (VISP) randomized con-trolled trial. JAMA. 2004;291:565–575.

20. Spence JD, Howard VJ, Chambless LE, Malinow MR, Pettigrew LC, Stampfer M, et al. Vitamin Intervention for Stroke Prevention (VISP) trial: Rationale and design. Neuroepidemiology. 2001;20:16–25.

21. Grundy SM, Pasternak R, Greenland P, Smith S Jr, Fuster V. Assessment of cardiovascular risk by use of multiple- risk- factor assessment equa-tions: A statement for healthcare professionals from the American Heart Association and the American College of Cardiology. Circulation. 1999;100:1481–1492.

22. Wolbers M, Koller MT, Witteman JC, Steyerberg EW. Prognostic models with competing risks: Methods and application to coronary risk predic-tion. Epidemiology. 2009;20:555–561.

23. D’Agostino RB, Nam B. Evaluation of the performance of survival analysis models: Discrimination and calibration measures. Handbook of statistics. New York: Elsevier; 2004:1–25.

24. Wolf PA, D’Agostino RB, Belanger AJ, Kannel WB. Probability of stroke: A risk profile from the Framingham Study. Stroke. 1991;22:312–318.

25. Adams RJ, Chimowitz MI, Alpert JS, Awad IA, Cerqueria MD, Fayad P, et al. Coronary risk evaluation in patients with transient ischemic attack and ischemic stroke: A scientific statement for healthcare professionals from the Stroke Council and the Council on Clinical Cardiology of the American Heart Association/American Stroke Association. Circulation. 2003;108:1278–1290.

26. Amarenco P, Bogousslavsky J, Callahan A III, Goldstein LB, Hennerici M, Rudolph AE, et al. High- dose atorvastatin after stroke or transient ischemic attack. N Engl J Med. 2006;355:549–559.

27. Yusuf S, Diener HC, Sacco RL, Cotton D, Ounpuu S, Lawton WA, et al. Telmisartan to prevent recurrent stroke and cardiovascular events. N Engl J Med. 2008;359:1225–1237.

28. 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;117:743–753.

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

Amytis Towfighi, Daniela Markovic and Bruce OvbiageleAfter Stroke

Utility of Framingham Coronary Heart Disease Risk Score for Predicting Cardiac Risk

Print ISSN: 0039-2499. Online ISSN: 1524-4628 Copyright © 2012 American Heart Association, Inc. All rights reserved.

is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231Stroke doi: 10.1161/STROKEAHA.112.668319

2012;43:2942-2947; originally published online September 4, 2012;Stroke. 

http://stroke.ahajournals.org/content/43/11/2942World Wide Web at:

The online version of this article, along with updated information and services, is located on the

http://stroke.ahajournals.org/content/suppl/2012/10/23/STROKEAHA.112.668319.DC1Data Supplement (unedited) at:

  http://stroke.ahajournals.org//subscriptions/

is online at: Stroke Information about subscribing to Subscriptions: 

http://www.lww.com/reprints Information about reprints can be found online at: Reprints:

  document. Permissions and Rights Question and Answer process is available in the

Request Permissions in the middle column of the Web page under Services. Further information about thisOnce the online version of the published article for which permission is being requested is located, click

can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office.Strokein Requests for permissions to reproduce figures, tables, or portions of articles originally publishedPermissions:

by guest on July 12, 2018http://stroke.ahajournals.org/

Dow

nloaded from

1

SUPPLEMENTAL MATERIALS

Supplementary Table 1. Multivariate logistic model results for baseline factors associated with

high FCRS*

Model I† Model II‡ Model III§

Factor OR (95% CI) p value OR (95% CI) p value OR (95% CI) p value

History of carotid

endarterectomy 1.51 (1.05-2.17) 0.03 1.34 (0.91-1.97) 0.14 -- ns

NIHSS: 1-4 vs. 0 1.34 (1.11-1.61) 0.002 1.30 (1.07-1.59) 0.008 1.26 (1.00-1.58) 0.04

NIHSS: ≥5 vs. 0 1.30 (0.93-1.81) 0.13 1.38 (0.96-1.98) 0.08 1.15 (0.76-1.73) 0.51

BMI: overweight vs.

normal weight 1.29 (1.05-1.58) 0.02 1.51 (1.21-1.88) <0.001 1.28 (0.99-1.66) 0.06

BMI: obese vs. normal

weight 1.10 (0.88-1.38) 0.39 1.49 (1.17-1.91) 0.002 1.66 (1.25-2.20) <0.001

Triglycerides >150

mg/dL 1.64 (1.38-1.94) <0.001 2.02 (1.67-2.43) <0.001 2.85 (2.28-3.56) <0.001

LDL >100 mg/dL 1.68 (1.37-2.06) <0.001 1.80 (1.45-2.24) <0.001 2.64 (2.06-3.40) <0.001

Antithrombotic use 1.13 (0.96-1.34) 0.15 1.15 (0.96-1.38) 0.12 1.14 (0.93-1.41) 0.21

Alcohol use in prior

year -- ns 1.23 (1.02-1.48) 0.03 0.73 (0.59-0.92) 0.006

Race: black vs. white -- ns 1.69 (1.14-2.50) 0.009 1.73 (1.09-2.74) 0.02

Race: other vs. white -- ns 1.35 (1.03-1.76) 0.03 1.78 (1.31-2.43) <0.001

* Antidyslipidemic medication use, history of congestive heart failure and history of prior stroke were not

significant at p<0.25 in any model.

†Adjusted for race, stroke severity, carotid endarterectomy, body mass index (BMI), alcohol intake, antithrombotic

use, low density lipoprotein cholesterol (LDL) >100 mg/dL and serum triglycerides >150 mg/dL

‡Adjusted for above factors and age

§Adjusted for above factors, age and sex

2

Supplementary Table 2. Association between high FCRS and individual FCRS components

with risk of MI in VISP subjects (n=3384 persons with complete data)*

Unadjusted Model I† Model II‡ Model III§

Factor HR (95% CI) p value HR (95% CI) p value HR (95% CI) p value HR (95% CI) p value

Age, per year 1.04 (1.02-1.06) <0.001 1.04 (1.02-1.06) <0.001 -- -- 1.03 (1.01-1.05) 0.008

Female 0.63 (0.43-0.94) 0.02 0.53 (0.34-0.83) 0.01 -- -- 0.68 (0.42-1.12) 0.13

SBP, per mm Hg 1.01 (1.00-1.02) 0.23 -- ns -- -- -- ns

DBP, per mm Hg 0.98 (0.97-1.00) 0.06 -- ns -- -- -- ns

Tot chol, per mg/dl 1.002 (0.998-1.005) 0.36 1.004 (1.002-1.007) 0.002 -- -- 1.004 (1.001-1.007) 0.01

HDL, per mg/dl 0.98 (0.97-1.00) 0.04 -- ns -- -- -- ns

Diabetes 2.37 (1.67-3.36) <0.001 2.26 (1.51-3.40) <0.001 -- -- 2.03 (1.35-3.05) 0.001

Smoker 1.11 (0.71-1.74) 0.66 1.64 (1.00-2.68) 0.05 -- -- 1.44 (0.89-2.32) 0.14

High FCRS (vs.

low/intermediate

FCRS)

3.45 (2.04-5.84) <0.001 -- -- 3.70 (2.14-6.38) <0.001 2.17 (1.16-4.09) 0.02

AIC 1727.34 1768.67 1722.97

Cross-validated

AUC (SE) 0.740 (0.056) 0.701 (0.061) 0.744 (0.050)

* All models were adjusted for six additional factors using backwards stepwise search (prior stroke, CHF, alcohol intake,

CEA, BMI, and antidyslipidemic use)

† Model I contains the subset of individual FCRS components most strongly associated with the outcome, not containing

FCRS predictor

‡ Model II contains the FCRS predictor, but does not contain the individual FCRS components

§ Model III contains the FCRS predictor and the subset of individual FCRS components most strongly associated with

outcome

NS: not significant at p<0.15