Difference in Carotid Artery Wall Structure Between
Swedish and French Men at Low and High Coronary Risk
Nicolas Denarié, MD; Alain Simon, MD; Gilles Chironi, MD; Jérôme Gariepy, MD;
Lars Kumlin, MD; Marc Massonneau, MD; Catherine Lanoiselée, MD;
Lennart Dimberg, MD, PhD; Jaime Levenson, MD
Background and Purpose—We attempted to detect a group-specific north-south difference in carotid artery intima-media
thickness (IMT), a marker of subsequent cardiovascular complication, by means of a case (high risk)– control (low risk)
study in French and Swedish men.
Methods—The selection of high-risk and low-risk subjects was performed within the lower and upper percentiles of the
Framingham risk distribution of 2 samples of 1000 white, male auto workers (45 to 50 years of age) in France (Renault)
and Sweden (Volvo). In total, 299 men at low risk (79 French, 76 Swedish) and high risk (61 French, 83 Swedish), free
from sustained hypertension, definite hypercholesterolemia, and cardiovascular disease, were included. Both common
carotid arteries, by ultrasonography and central off-line computerized analysis, provided measurements of far wall
media thickness, lumen diameter, and cross-sectional area IMT (CSA-IMT).
Results—As compared with low-risk status, high-risk status was associated with higher IMT (P,0.001), diameter
(P,0.01), and CSA-IMT (P,0.001) in French men and higher CSA-IMT (P,0.05) in Swedish men. IMT, diameter,
and CSA-IMT were higher in Swedish than in French men in the low-risk group (P,0.001) and in the high-risk group
(P,0.01, P,0.001, P,0.001). The multivariate analysis of the whole population showed that IMT, diameter, and
CSA-IMT were associated with risk status (P,0.01, P,0.01, P,0.001) and geographic status (P,0.001).
Conclusions—These findings show that the geographic status influences carotid artery structure independent of traditional
cardiovascular risk factors and that this may affect the mortality and morbidity gradient between Northern and Southern
Europe. (Stroke. 2001;32:1775-1779.)
Downloaded from http://ahajournals.org by on January 11, 2022
Key Words: arterial wall n carotid arteries n geography
D
ata to explain higher cardiovascular mortality and morbidity rates in Northern than in Southern Europe1 are
scarce. An explanation may be related to geographic difference, either in the burden of traditional cardiovascular risk
factors or in the reactivity of arteries to risk factors.2 To
explore these hypotheses, we have designed the Renault
Volvo Cœur Study with the objective to compare the cardiovascular risk profile between similar types of auto workers in
France (Renault) and in Sweden (Volvo).2 In a previous
cross-sectional analysis, we have shown that the multifactorial risk burden, as estimated with the use of the Framingham
index, was not different between French and Swedish groups,
therefore excluding a geographic difference in the burden of
traditional risk factors.2 In the present work, we tested the
alternative hypothesis of a geographic difference in the
arterial reactivity to traditional risk factors by comparing
ultrasonographically assessed carotid artery structure, especially intima-media thickness (IMT),3 at the same level of
cardiovascular risk, between French and Swedish samples of
the Renault Volvo-Cœur Study.
Subjects and Methods
Two hundred ninety-nine white men, 45 to 50 years, at low and high
risk, were selected among auto workers in France (Renault) and in
Sweden (Volvo) (Renault Volvo Cœur study),2 exclusive of those
with blood pressure .170/100 mm Hg or antihypertensive treatment
for $3 months, total cholesterol .6.5 mmol/L, or lipid-lowering
drug treatment for $3 months and cardiovascular disease such
coronary artery disease, stroke, or arteriopathy of the lower limbs.
Subjects at low risk were chosen within the lower percentile of the
distribution of the Framingham risk index (see below for estimation)
of 1000 randomly selected subjects in each country to obtain an
expected number of 90 individuals by country (see statistical issue).
The so-chosen subjects were proposed to undergo ultrasound arterial
investigation at the hospital (Broussais Hospital, Paris, France, or
Östra Hospital, Goteborg, Sweden). They were classified by type of
work into manual or clerical worker. Of these, 79 subjects in France
(48% manual workers) and 76 subjects in Sweden (25% manual
workers) agreed to undergo hospital investigation after being given
detailed information on the protocol (Table 1).
Received December 28, 2000; final revision received April 4, 2001;accepted May 3, 2001.
From Centre de Médecine Préventive Cardiovasculaire, Hôpital Broussais, Paris, France (N.D., A.S., G.C., J.G., M.M.); Renault Automobiles,
Boulogne, France (C.L.); and Volvo Aero Corporation, Trollhättan, Sweden (L.K., L.D.).
Correspondence to Pr Alain Simon, Hôpital Broussais, Centre de Médecine Préventive Cardiovasculaire, 96, Rue Didot, 75014 Paris, France. E-mail
alain.simon@brs.ap-hop-paris.fr
© 2001 American Heart Association, Inc.
Stroke is available at http://www.strokeaha.org
1775
1776
August 2001
Stroke
TABLE 1.
Cardiovascular Risk Factors by Risk and Geographic Status
Low Risk
Age, y
P*
High Risk
France
(n579)
Sweden
(n576)
France
(n561)
Sweden
(n583)
France
Sweden
50.7 (1.7)
49.3 (1.6)‡
51.0 (1.7)
49.7 (1.1)‡
NS
,0.05
38 (48)
19 (25)†
27 (44)
36 (43)
NS
0.01
Work type (n, %)
Manual
41 (52)
57 (75)
34 (56)
47 (57)
Body mass index, kg/m2
Clerical
25.3 (2.6)
23.8 (2.6)‡
26.5 (3.1)
27.2 (3.7)
0.01
,0.001
Systolic pressure, mm Hg
125 (8.2)
121 (10.1)†
133 (15.1)
136 (16.7)
,0.001
,0.001
Total–to–HDL cholesterol ratio
3.41 (0.62)
3.34 (0.73)
5.19 (0.97)
5.72 (1.28)†
,0.001
,0.001
Blood glucose, mmol/L
5.61 (0.59)
4.35 (0.47)‡
5.91 (1.11)
4.64 (1.66)‡
,0.05
NS
Current smoking, n (%)
13 (16)
1 (1)†
40 (66)
42 (51)
,0.001
,0.001
Framingham risk (% at 10 y)
5.5 (2.1)
4.1 (1.5)†
13.7 (5.0)
14.3 (5.3)
,0.001
,0.001
Values are mean (SD) or number (n) with percentages (%). Between-group comparisons were done by unpaired
t test for continuous variables and by x2 test for qualitative variables.
*Comparison between low and high risk.
†,‡Mean P,0.01, P,0.001 for comparing France with Sweden.
Subjects at high risk were chosen within the upper percentile of
the distribution of the Framingham risk index of the 1000 previously
mentioned subjects to obtain an expected number of 90 individuals
(see statistical issue). Among these, 61 subjects in France (44%
manual workers) and 83 subjects in Sweden (43% manual workers)
agreed to undergo arterial investigation at the hospital after being
given detailed information (Table 1). The study was approved by
local ethics committees in both countries.
Downloaded from http://ahajournals.org by on January 11, 2022
blood pressure (sphygmomanometry); fasting values of total cholesterol, HDL cholesterol after LDL and VLDL precipitation, and blood
glucose (enzymatic methods); and current smoking (regular smoking
for the previous 3 months regardless of the amount smoked). The
Framingham risk index (percent of coronary event at 10 years) was
estimated by entering into the Framingham model equations the
following variables: age, male sex, systolic pressure, total to HDL
cholesterol ratio, current smoking, and diabetes defined as glycemia
$7.8 mmol/L, the use of antidiabetic drugs, or both.6
Ultrasound Investigation
In each country, all subjects were investigated by the same sonographers trained to the measurements required by the protocol.
Carotid ultrasonography was performed with a real-time, B-mode
ultrasound imager (Ultramark 4 in Sweden and Ultramark 9 in
France, Advanced Technologies Laboratories), with a 7.5-MHz
probe. Longitudinal images of lumen diameter and far-wall IMT of
the right and left common carotid arteries, 2 to 3 cm distal to the
bifurcation, were performed according to a standardized procedure.3
Once the lumen diameter and the two parallel echogenic interfaces
(lumen-intimal and medial-adventitial) defining IMT were adequately visualized, the carotid image was frozen in end diastole by
ECG R-triggering, transferred to a computer (Apple Macintosh),
digitized, and stored on optical disk for off-line central reading.3
Central Reading
The same reader, blinded to the risk and geographic status of the
patient, performed the analyses of all ultrasound images obtained in
both countries by means of an automated computerized program (Iô,
IôDP).4 The program automatically located the two interfaces
defining far-wall IMT and calculated their average distance along at
least 1 cm of longitudinal length.4 The program also located the two
leading edges of lumen-intimal interfaces of the near and far walls
defining the lumen diameter and calculated their distance along the
same length as the IMT.4 The IMT cross-sectional area (CSA-IMT)
was calculated from IMT and lumen diameter (D) as
p3IMT3(IMT1D).5 The final values of IMT, diameter, and CSAIMT considered in the study were the average of left and right
measurements.
Risk Factors
Risk factors were evaluated concomitantly with the ultrasound
investigation, including body mass index (weight/height2); resting
Statistical Issues
We targeted a group of 4390 subjects to reach 4375 subjects,
which, based on our power estimate, would give a 90% chance to
show a 10% difference for IMT.
Univariate comparisons of cardiovascular risk factors by risk and
geographic status in Table 1 were done with an unpaired t test for
continuous variables and x2 test for qualitative variables. In Table 2,
carotid arterial parameters were adjusted for age, work type, and
body mass index in the 4 study groups (defined according to risk and
geographic status) by using a multivariate linear regression with the
arterial parameter as the dependent variable and the type of group
(0,1, 2,3), age, body mass index, and work type (0,1) as independent
variables. For comparing carotid artery parameters between groups 2
by 2 in Table 2, a multivariate linear regression analysis was used
with the arterial parameter as the dependent variable and the type of
group (0,1), age, body mass index and work type (0,1) as independent variables. A separate analysis of the whole study population in
Table 3 used multivariate linear regressions to assess associations
between carotid arterial parameters (dependent variable) and age,
the type of work, 0,1, body mass index, the risk status (0,1) and
the geographic status (0,1) as independent variables. The distributions of the 3 independent variables (IMT, diameter, CSAIMT) were analyzed, and the nonstatistical significance of the test
for normality (Shapiro-Wilk W test) showed that the distributions
were normal.
Results
Table 1 shows that, as expected, high-risk subjects had a
higher Framingham risk index than did low-risk subjects in
both countries (P,0.001). As compared with low-risk status,
high-risk status was associated with (1) higher smoking
Denairé et al
TABLE 2.
Geographic Status and Intima-Media Thickness
1777
Carotid Artery Parameters by Risk and Geographic Status
Low Risk
IMT, mm
Lumen diameter, mm
CSA-IMT, mm2
P*
High Risk
France
Sweden
France
Sweden
France
Sweden
0.54 (0.01)
0.60 (0.01)‡
0.58 (0.01)
0.62 (0.01)†
,0.001
NS
5.76 (0.06)
6.27 (0.06)‡
6.01 (0.07)
6.40 (0.06)‡
,0.01
NS
10.72 (0.26)
12.96 (0.30)‡
12.02 (0.31)
13.87 (0.26)‡
,0.001
,0.05
Values are mean (SE) adjusted for age, type of work, and body mass index.
*Comparison between low and high risk.
†,‡Mean P,0.01, P,0.001 for comparing France with Sweden.
Adjustments and group comparisons were done by multivariate linear regression models (see Methods).
Downloaded from http://ahajournals.org by on January 11, 2022
frequency and higher systolic pressure and total to HDL
cholesterol ratio in both countries (P,0.001); (2) higher body
mass index in France (P50.01) and in Sweden (P,0.001);
(3) higher blood glucose in France (P,0.05); and (4) older
age (P,0.05) and higher manual worker frequency (P50.01)
in Sweden. In the high-risk group, the Framingham index and
risk factors did not differ between countries, except lower age
and blood glucose (P,0.001) in Sweden and higher total to
HDL cholesterol ratio (P,0.01) in France. In the low-risk
group, lower values of age (P,0.001), manual worker
frequency (P,0.01), body mass index (P,0.001), systolic
pressure (P,0.01), blood glucose (P,0.001), smoking frequency (P,0.01) and Framingham index (P,0.01) were
found in Sweden as compared with France.
Table 2 shows carotid artery parameters by risk and
geographic status after adjustment for age, type of work, and
body mass index. As compared with low-risk status, high-risk
status was associated with higher carotid IMT (P,0.001),
higher lumen diameter (P,0.01), and higher CSA-IMT
(P,0.001) in France and with higher CSA-IMT (P,0.05) in
Sweden. IMT, lumen diameter, and CSA-IMT were higher in
Sweden than in France in the low-risk group (P,0.001) and
in the high-risk group (P,0.01, P,0.001, P,0.001).
Table 3 shows that in the multivariate analysis of the whole
study population, (1) IMT was independently associated with
age (P,0.05), body mass index (P,0.05), risk status
(P,0.01), and geographic status (P,0.001); (2) lumen diameter was independently associated with body mass index
(P,0.001), risk status (P,0.01), and geographic status
(P,0.001); (3) CSA-IMT was independently associated with
age (P,0.05), body mass index (P,0.001), risk status
(P,0.001), and geographic status (P,0.001).
Discussion
The two main findings are that the carotid artery structure is
influenced independently by (1) the burden of traditional
cardiovascular risk factors, as estimated with the Framingham
index and (2) the geographic status of the subject, which
thereby may be considered as a potential risk marker independent of the Framingham risk factors.
Before these findings can be interpreted, some methodologic aspects need to be considered. First, the objective to
show a geographic difference needs to control as far as
possible for ethnic and genetic factors. We have tried to
control for ethnicity by restricting our study to Caucasian
subjects, so excluding immigrants from other ethnic origin
who are frequent among auto workers. However, the usual
method consisting in looking at immigrant populations within
the same population to assess the effects of geography would
have answered the geography question to a greater degree
than the approach in the present work. A second problem is
that our subject samples are not representative of the general
population in both countries. However, our selection processes were well characterized and so have provided individuals at high and low risk for cardiovascular disease. We have
also minimized influences of age and sex by studying men
with a narrow range of age from 45 to 50 years. We have
taken into account potential bias regarding socioeconomic
status assessed by defined the type of work in a dichotomous
TABLE 3. Multivariate Analysis of Carotid Artery Parameters on Demographic
Risk and Geographic Status in the Whole Study Population
Independent Variables
IMT
Lumen Diameter
Age (continuous)
0.0074 (0.0030)*
0.0224 (0.0189)
Type of work (0, manual; 1, clerical)
0.0021 (0.0047) 20.0047 (0.0297)
CSA-IMT
0.212 (0.089)*
0.022 (0.139)
Body mass index (continuous)
0.0037 (0.0014)*
0.0430 (0.0093)‡ 0.165 (0.043)‡
Risk status (0, high; 1, low)
0.0156 (0.0049)†
0.0931 (0.0310)† 0.552 (0.144)‡
Geographic status (0, Sweden; 1, France)
0.0255 (0.0050)‡
0.2280 (0.0316)‡ 1.024 (0.148)‡
R2
0.17
0.27
Values are slopes (SE).
*,†,‡Mean P,0.05, P,0.01, P,0.001.
Multivariate linear regression models were used (see Methods).
Distributions of IMT, diameter, and CSA-IMT were not skewed (see Methods).
0.26
1778
Stroke
August 2001
Downloaded from http://ahajournals.org by on January 11, 2022
way (manual or clerical worker). As the percentage of these
two categories of workers differed between certain groups,
we have performed 2 by 2 comparison of groups by adjusting
for type of work. We have also introduced the type of work
as an independent variable in the multivariate analyses of
carotid artery parameters in the whole population. Unfortunately, we did not have additional information on the socioeconomic status, such as family income, about which questioning of subjects was not possible because of the problem of
confidentiality. A third issue is the physiological meaning of
the changes measured in the structure of the distal common
carotid artery free from atherosclerosis. These changes are
not measures of atherosclerotic plaque7 and represent rather
adaptive changes of the arterial wall to shear and/or tensile
stresses.8 Moreover, previous studies have shown that IMT is
a picture of integrated effects of traditional risk factors such
as hypertension3,9 but is not necessarily related to embolic
disease. Therefore, the relation of IMT to clinical complications such as stroke is not clear, although several prospective
studies have shown that increased carotid IMT was a powerful predictor of subsequent clinical events including myocardial infarction and stroke.10,11 A last methodologic aspect is
the reliability of our arterial measurements. The reliability
was enhanced by the fact that ultrasound images obtained by
sonographers were sent after storage on optical disk to a
blinded central reader who performed off-line measurements.
However, one limitation may be that the reproducibility
between sonographers was not tested. Therefore this source
of measurement error could not be evaluated, but sonographers in both countries were extensively trained in a common
protocol before the beginning of the study. Moreover, on the
basis of the work of Riley et al,12 it has been shown that the
standardized ultrasonic technique yields highly reproducible
measures of carotid IMT. In addition, our IMT measurement
was completely automated, thanks to a computerized edgetracking program allowing optimal precision and reproducibility rates to be obtained.4
The finding that high-risk status was associated with
increased carotid IMT, increased lumen diameter, and
increased CSA-IMT is not surprising. Previous studies
have shown that IMT correlated positively with all risk
factors used in the Framingham risk estimation,3,9 supporting the view that IMT is a comprehensive picture of the
integrated effects of traditional risk factors. The association between carotid artery lumen diameter and high-risk
status is newer and more in line with a previous study in
middle-aged subjects showing that the common carotid
artery diameter was correlated with risk factors for cardiovascular disease.13 The mechanisms of the carotid artery
dilation found in high-risk subjects may be related to early
compensatory enlargement to preserve lumen area. However, the influence of high-risk status on carotid artery
IMT and lumen diameter was more marked in the French
than in the Swedish subjects, suggesting that effects of
traditional risk factors on carotid artery structure is blunted
in the latter, perhaps by other nontraditional risk factors
exclusive of geographic difference. Such factors may be
new or emerging risk factors, of environmental and/or
genetic nature, not analyzed in the present work. The
strong and independent influence of the geographic status
on carotid artery IMT, diameter, and CSA-IMT constitutes
the main finding of our work. As compared with French
subjects, Swedish subjects have hypertrophy of the carotid
wall as shown by CSA-IMT, which is an estimation of the
arterial mass5 and a dilation of the carotid lumen. This
Swedish-French difference in carotid artery structure exists in subjects at low risk and at high risk. It is therefore
independent of the level of cardiovascular risk of the
subjects, as confirmed by the multivariate analysis of
arterial parameters on the geographic status in the whole
population. The multivariate analysis shows also that the
association of geographic status with carotid artery structure is independent of the two determinant factors of IMT
and lumen diameter, namely, age and body mass index.3,9
However, the mechanism relating the geographic status to
the carotid artery structure has not been elucidated and
may be caused by factors unmeasured in the present work.
In conclusion, this study shows for the first time, to our
knowledge, an independent association between the geographic status (ie, Swedish or French status) of white
middle-aged men and a well-recognized marker of early
arterial wall change as carotid IMT. Because increased
carotid IMT predicts the occurrence of a subsequent
cardiovascular event, its geographic dependence should be
used in further investigations for more in-depth analysis of
the factors contributing to the cardiovascular mortality and
morbidity gradient between Northern and Southern Europe. However, the small number of subjects of our study
reduces its statistical power, and larger studies are needed
to confirm our findings.
Appendix
The Renault Volvo Coeur Projet Group
Renault: Catherine Lanoiselée, MD; Guillemette Latscha, MD;
Christine Morvan, MD; Madeleine Leroy, MD; Dominique Roussel,
MD; Olivier Galamand, MD; Sylvie Selosse, MD; and Jacques
Sissler, MD. Volvo: Lennart Dimberg, MD, PhD; Carl-Erik Hedström, MD; Lars Kumlin, MD; Carlgunnar Lidström, MD; Häkan
Sterlind, MD; Gisela Rose, MD; and Irma Wright, MD. Östra,
Sahlgrensk, and Uppsala University Hospitals: Sverker Jern, MD,
PhD; Per Björntorp, MD, PhD; Per Morin, MD, PhD; Lennart
Hansson, MD, PhD; and Björn Dahlöf, MD, PhD. Broussais
Hospital: Alain Simon, MD; Jaime Levenson, MD; Jérôme Gariépy,
MD; Gilles Chironi, MD; Nicolas Denarié, MD. IôDP, Paris: Marc
Massonneau, MD. Nordic School of Public Health: Bo Eriksson,
PhD.
Acknowledgments
We thank the Renault and Volvo Companies for sponsoring the
Study; the “group PCV METRA” for invaluable advice in the
conception of the study, and the Renault Volvo Cœur Project group
(see Appendix). We also are grateful to the nurse, Margareta Leijon,
for performing the ultrasound investigation under the direction of
Sverker Jern, MD (Östra Hospital, Göteborg, Sweden). We thank
Maria Ortega for excellent secretarial assistance.
References
1. WHO MONICA Project. Prepared by Tunstall-Pedoe H, Kuulasmaa K,
Amoyel P, Arveiler D, Rajakangos AM, Pajak A. Myocardial infarction
Denairé et al
2.
3.
4.
5.
6.
and coronary deaths in the WHO-MONICA project: registration procedures, event rates, and case-fatality rates in 38 populations from 21
countries in 4 continents. Circulation. 1991;90:583– 612.
Simon A, Dimberg L, Levenson J, Lanoiselee C, Massonneau M, Erikson
B, Jern S, Kumlin L, Marin P, Dahlof B, Hansson L, Bjorntorp P, for the
Cœur Project Group. Comparison of cardiovascular risk profile between
male employees of two automative companies in France and Sweden. Eur
J Epidemiol. 1997;13:885– 891.
Gariepy J, Salomon J, Denarie N, Laskri F, Megnien JL, Levenson J,
Simon A. Sex and topographic differences in associations between largeartery wall thickness and coronary risk profile in a French working
cohort. Arterioscler Thromb Vasc Biol. 1998;18:584 –590.
Graf S, Gariepy J, Massonneau M, Armentano RL, Mansour S, Barra
JG, Simon A, Levenson J. Experimental and clinical validation of
arterial diameter waveform and intimal media thickness obtained from
B-mode ultrasound image processing. Ultrasound Med Biol. 1999;9:
1353–1363.
Linhart A, Gariepy J, Giral P, Levenson J, Simon A. Carotid artery and
left ventricular structural relationship in asymptomatic men at risk for
cardiovascular disease. Atherosclerosis. 1996;127:103–113.
Grundy SM, Pasterncek R, Greenland P, Smith S, Fuster V. Assessment
of cardiovascular risk by use of multiple-risk-factor assessment equations: a statement for healthcare professionals from the American Heart
Association and the American College of Cardiology. Circulation. 1999;
100:1481–1492.
Geographic Status and Intima-Media Thickness
1779
7. Gariepy J, Massonneau M, Levenson J, Heudes D, Simon A, and the PCV
METRA Group. Evidence for in vivo carotid and femoral wall thickening
in human hypertension. Hypertension. 1993;22:111–118.
8. Stary HC, Blankenhorn DH, Chandler B, Glagov S, Insull W Jr,
Richardson M, Rosenfeld ME, Schaeffer SA, Schwartz CJ, Wagner WD.
A definition of the intima of human arteries and of its atherogenic-prone
regions. Arterioscler Thromb. 1992;12:120 –134.
9. Crouse JR, Goldbourt U, Evans G, Pinsky J, Sharrett AR, Sorlie P, Riley
W, Heiss G, for the ARIC Investigators. Risk factors and segmentspecific carotid arterial enlargement in the Atherosclerosis Risk in Communities (ARIC) cohort. Stroke. 1996;27:69 –75.
10. Bots ML, Hoes AW, Koudstaal PJ, Hofman A, Grobbee DE. Common
carotid intima-media thickness and risk of stroke and myocardial
infarction: the Rotterdam Study. Circulation. 1997;96:1435–1437.
11. O’Leary DH, Polak JF, Kronmal RA, Manolio TA, Burke GL,
Wolfson SK, for Cardiovascular Health Study Collaborative Research
Group. Carotid-artery intima and media thickness as a risk factor for
myocardial infarction and stroke in older adults. N Engl J Med.
1999;340:14 –22.
12. Riley WA, Barnes RW, Applegate WB, Dempsey R, Hartwell T, Davis
VG, Bond G, Furberg CD. Reproducibility of noninvasive ultrasonic
measurement of carotid atherosclerosis: the Asymptomatic Carotid Artery
Plaque Study. Stroke. 1992;23:1062– 68.
13. Jensen-Urstad K, Jensen-Urstad M, Joahnsson J. Carotid artery diameter
correlates with risk factors for cardiovascular disease in a population of
55-year-old subjects. Stroke. 1999;30:1572–1576.
Downloaded from http://ahajournals.org by on January 11, 2022