Alcohol Consumption, Cardiovascular Risk Factors,
and Mortality in Two Chicago Epidemiologic Studies
ALAN R. DYER, PH.D., JEREMIAH STAMLER, M.D., OGLESBY PAUL, M.D.,
DAVID M. BERKSON, M.D., MARK H. LEPPER, M.D., HARLLEY MCKEAN, PH.D.,
RICHARD B. SHEKELLE, PH.D., HOWARD A. LINDBERG, M.D., AND DAN GARSIDE, B.S.
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
SUMMARY Multivariate analysis of the association at baseline
between problem drinking and cardiovascular risk factors among
1,233 white male employees of the Chicago Peoples Gas Company
age 40-59 showed the 38 problem drinkers with significantly higher
blood pressures and cigarette consumption and significantly lower
relative weights than the others. Similar analyses among 1,899 white
male employees of the Hawthorne Works of the Western Electric
Company in Chicago age 40-55 showed the 117 men consuming 5 or
more drinks per day with significantly higher blood pressures and
cigarette use than the others. No significant differences were recorded
between heavy drinkers and the others in serum cholesterol level.
The gas company problem drinkers had significantly higher 15year mortality rates from all causes, cardiovascular diseases, coronary heart disease, and sudden death. These differences could not be
entirely explained by their blood pressure, smoking, and relative
weight status. The Western Electric heavy drinkers had increased
10-year mortality rates both for all causes and noncardiovascular
causes.
HEAVY CONSUMPTION OF ALCOHOL has come under increased scrutiny in the last few years, both as a potential risk factor for the adult cardiovascular diseases, and for
possible associations with known cardiovascular risk factors
as well.
Data from two prospective epidemiological studies in industry in Chicago provided an opportunity to examine both
of these unresolved questions. This report thus focuses on
the relationship between alcohol consumption and cardiovascular risk factors, and on the association between
alcohol intake and mortality from cardiovascular and all
causes in 1,233 white male employees of the Chicago
Peoples Gas, Light, and Coke Company age 40-59 in 1958
and in 1,899 white male employees of the Hawthorne Works
of the Western Electric Company in Chicago, age 40-55 in
1957.
a red-label file for men classified as problem drinkers. A
man was so classified when unable to perform his job
properly due to chronic problems with alcohol. He had a
history of coming to work intoxicated and may have been
suspended from his job at some time for alcohol abuse. In
1958-59 there were 38 men in the cohort of 1,233 who had
been classified as problem drinkers by the Medical Department of the Gas Company. These 38 men constitute the high
alcohol consumption group in the subsequent analyses for
the Gas Company study.
The Chicago Western Electric Company study is a longterm investigation of coronary heart disease begun in the fall
of 1957 among men, whose ages at that time ranged between
40-55, employed by the Hawthorne Works of the Western
Electric Company in Chicago. The 5,397 men age 40-55 employed at the works for more than two years as of 1957 were
assigned an identification number by a randomizing process.
Members in the randomly selected group were invited to
Methods
The Chicago Peoples Gas, Light, and Coke Company
study is a longitudinal investigation of the natural history,
epidemiology, and etiology of the adult cardiovascular diseases in males whose ages fell between 40-59 years on 1
January 1958. Of the 1,594 male employees in this age range
on that date, 1,465 (91.9%) underwent complete physical examination in 1958. This examination and other details of
this study have been described at length elsewhere." 2 This
paper reports on 1,233 white males with complete baseline
data who were free of definite coronary heart disease on initial examination. These men have been followed long-term
without systematic intervention.
Data on alcohol consumption were not collected on these
men. However, the Gas Company Medical Department had
participate voluntarily by being called from the random list.
Those who accepted the invitation underwent complete
physical examination in late 1957 or in 1958. This examination and other details of this study have been described at
length elsewhere.3 The response rate among those who were
invited to participate was 67%. The group of 1,899 men considered in this report consists of white males free of definite
coronary heart disease on initial examination with complete
baseline data on all relevant variables. These men have been
followed long-term without systematic intervention.
Data on alcohol consumption were obtained yearly as
part of the annual physical examination and as part of a onetime nutrition survey taken during the first year of the study.
The assessment of the amount of alcohol consumed for each
individual is based on the data collected at the baseline
physical examination, the first anniversary examination, and
the nutrition survey. On each of these questionnaires the individual was asked for the number of drinks consumed per
month for each type of alcoholic beverage: beer, whiskey,
and wine. The size of the drink was monitored as much as
possible; each drink described provided about 0.7 ounces or
21 grams of absolute alcohol. The numbers of drinks for
each type of beverage were then added together to provide a
total number of drinks per month for that questionnaire.
From the Department of Community Health and Preventive Medicine,
Northwestern University Medical School, Chicago, Illinois.
Supported by grants from the Chicago Heart Association and by grant No.
2 RO 1 HL15174-05-07 from the National Heart and Lung Institute,
National Institutes of Health, U.S. Public Health Service.
Dr. Dyer was an Established Investigator of the American Heart Association while this research was being done.
Address for reprints: Alan R. Dyer, Ph.D., Department of Community
Health and Preventive Medicine, Northwestern University, 303 E. Chicago
Avenue, Chicago, Illinois 60611.
Received November 26, 1976; revision accepted June 27, 1977.
1067
CIRCULATION
1068
Based on a 30-day month, the means for the three surveys
29.3, 30.6, and 32.7, and the standard deviations were
39.0, 38.7, and 43.0. The correlations between the three surveys ranged between 0.61 and 0.75. The lowest correlation
was between the baseline physical exam and the nutrition
survey, while the highest was between the nutrition survey
and the first anniversary physical exam. A single alcohol
consumption code was obtained for each individual by taking the maximum number of drinks per month for the three
questionnaires. The maximum was selected instead of the
average of the three because the maximum makes comparisons with studies which take only a single survey easier,
and because it was felt that the maximum better minimizes
the probability of a heavy drinker being classified as not
heavy. In addition, the results are comparable whether one
uses the maximum or the average. To facilitate comparisons
with the Gas Company problem drinkers, each individual
was also dichotomized with respect to alcohol consumption
as heavy or not heavy, where heavy was defined as 5 or more
drinks per day on any one of the three questionnaires. With
this criterion, 117 of the 1,899 men were classified as heavy
drinkers. This group of 117 men constitutes the high alcohol
consumption group in the subsequent analyses for the
Western Electric Company study.
The high intake groups in each study are compared with
those consuming less alcohol for age, systolic blood
pressure, diastolic blood pressure, heart rate or pulse,
relative weight, cigarettes/day, and serum cholesterol.
(Relative weight is the ratio of actual weight to desirable
weight X 100, where desirable weight is obtained for a given
height from tables published by the Metropolitan Life Insurance Company.4) The studies differ on the blood pressure
coded for analysis. In the Gas Company the blood pressure
coded was the lowest of four (5th phase) diastolic readings,
along with its corresponding systolic, with the subject
seated. In Western Electric the blood pressure coded was
based on a single reading with the subject seated at the
beginning of the examination. In the Gas Company, heart
rate was calculated from the ECG. In the Western Electric
Company, pulse was taken. The variables are compared for
the two alcohol consumption groups within each study utilizing two sample t-tests for the univariate analyses and multiple logistic regression in the multivariate analyses.5 In addition, in the Western Electric study those variables which
show a significant association with alcohol consumption
based on the heavy-not heavy dichotomization are examined
over the full range of alcohol consumption based on the
maximum of the three codes.
The analysis of alcohol consumption and mortality is
based on 15 years of follow-up in the Gas Company and on
ten years of follow-up in Western Electric.
were
VOL 56, No 6, DECEMBER 1977
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
the two groups for each variable is also given. The univariate
analyses show that the problem drinkers had a systolic blood
pressure 8.5 mm Hg higher on the average than the rest of
the men, a diastolic pressure 4.7 mm Hg higher, a heart rate
5.2 beats/min faster, smoked 7.9 more cigarettes/day 86.8% of the problem drinkers were smokers compared to
67.5% among the rest of the men - and had a relative
weight 9.9 percentage points less than that of the rest of the
men. Each of these differences is statistically significant in
the univariate analyses. Although the problem drinkers had
a mean serum cholesterol concentration 10 mg/dl less than
that of the rest of the men, this difference is not statistically
significant.
Multivariate analysis using multiple logistic regression
with problem drinking coded 0-1 as the dependent variable
was also performed on these differences. Two multiple
logistic analyses were done - one with age, systolic blood
pressure, serum cholesterol, relative weight, heart rate, and
cigarettes/day as the independent variables, and a second
with diastolic blood pressure in place of systolic blood
pressure. The t-value for a test of the hypothesis that the
coefficient of the logistic regression model for a given
variable is equal to zero is also presented in table 1. The adjusted t's are taken from the first model for all variables except diastolic blood pressure, which is taken from the second. All of the significant associations from the univariate
analyses are retained in the multivariate analysis except
heart rate which just fails to reach significance.
Based on the higher mean levels of systolic and diastolic
blood pressure among the problem drinkers, one would expect this group to also have a higher prevalence of elevated
or high blood pressure. Using a systolic blood pressure of
160 mm Hg or greater or a diastolic blood pressure of 95
mm Hg or greater to define high blood pressure, 21.1% of
the problem drinkers had high blood pressure, while only
13.9% of the rest of the men had high blood pressure. If
analysis of covariance is used to adjust these percentages for
differences between the two groups in age, relative weight,
heart rate, serum cholesterol, and cigarettes/day, the
percentages become 22.8 and 13.9 respectively. In neither of
these analyses are the results statistically significant, but this
is probably a function of the small number of problem
drinkers.
Since high blood pressure and smoking are two of the
TABLE 1. Means of Risk Factors for Problem Drinkers and
Nonproblem Drinkers, Chicago Peoples Gas Company, (1233
white males free of definite CHD and age 40-59 in 1958)
Variable
Age
Results
Alcohol and Cardiovascular Risk Factors
Table 1 presents the baseline means for age, systolic blood
pressure, diastolic blood pressure, heart rate, serum
cholesterol, relative weight, and cigarettes/day for the 38
problem drinkers and the 1,195 nonproblem drinkers from
the Chicago Peoples Gas Company. The t-value for the test
of the hypothesis that there is no difference in the means for
Systolic
pressure
Diastolic
pressure
Heart rate
Serum
cholesterol
Relative weight
Problem
drinkers
(N 38)
Nonproblem
drnkers
(N 1195) t
Adjustedt
t
-
-0.74
49.5
49.7
141.9
133.4
2.62**
3.00**
86.0
76.8
81.3
71.6
2.81**
2.75**
3.40**
1.84
-0.21
236.3
-1.37
-1.40
121.4
-3.67** -.3.37**
3.97**
3.155**
13.5
Cigarettes/day
tAdjusted for other variables by multiple logistic regression.
**P
<0.01.
226.3
111.5
21.4
1069
ALCOHOL, RISK FACTORS, AND MORTALITY/Dyer et al.
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
three major risk factors for morbidity and mortality from
the adult cardiovascular diseases,' the higher levels of these
two variables in the problem drinkers place them at a higher
risk of morbidity and mortality than the other men.
(Previous findings in the Gas Company have shown the
association between relative weight and cardiovascular mortality to follow a U-shaped curve.6 Thus, the lower relative
weights of the problem drinkers also place them at higher
risk.)
Table 2 presents the baseline means for these same seven
variables for the 117 heavy drinkers and the 1,782 nonheavy
drinkers from the Chicago Western Electric Company. The
heavy drinkers had a systolic blood pressure at baseline
which was 9.7 mm Hg higher on the average than that of the
others, a diastolic pressure 5.9 mm Hg higher, and smoked
an average of 7.1 more cigarettes/day - 72.6% of the heavy
drinkers were smokers compared to 56.9% among the rest of
the men. Each of these differences is statistically significant
based on the simple univariate t's. These differences are
similar to those found among the problem drinkers in the
Gas Company. Unlike the heavy drinkers in the Gas Company, relative weight was higher on the average among the
heavy drinkers, but this difference is not statistically significant. Pulse was faster by about 2 beats/min among the
heavy drinkers, but this difference is not biologically or
statistically significant. For serum cholesterol there was no
significant difference noted between the heavy drinkers and
the rest of the men. When the differences between the two
groups are examined by multiple logistic regression, the
associations between blood pressure and alcohol consumption and between cigarettes and alcohol consumption remain
statistically significant, as evidenced by the adjusted t's.
Based on a systolic blood pressure of 160 mm Hg or
greater or a diastolic blood pressure of 95 mm Hg or
greater, the prevalence of high blood pressure was 43.6%
among the heavy drinkers and 22.9% among the rest of the
men. This difference is highly significant statistically. If
these percentages are adjusted by analysis of covariance for
differences between the two groups in age, relative weight,
pulse, cigarettes/day, and serum cholesterol, the percentages become 41.5 and 23.0, respectively, a difference that
continues to be statistically significant.
For the Western Electric cohort table 3 presents a more
TABLE 2. Means of Risk Factors for Heavy Drinkers (.6
drink8/day) and Nonheavy Drinkers (<5 drinks/day) Chicago
Western Electric Company, (1899 white male free of definite
CHD and age 40-55 in 1967)
Variable
Nonheavy
Heavy
drinkers
drinkers
- 117) (N - 1782)
(N
47.8
Adjustedt
t
t
0.50
Age
48.0
pressure
Diastolic
pressure
Pulse
Serum
cholesterol
Relative weight
Cigarettes/day
144.2
134.5
5.33**
92.6
80.7
86.7
78.8
5.50**
Systolic
250.1
117.0
17.3
247.4
115.2
1.85
0.53
1.24
6.64**
0.52
4.67**
5.00**
-0.28
-0.33
0.49
10.2
6.38**
tAdjusted for other variables by multiple logistic regression.
detailed analysis of the association between alcohol consumption and the three variables - systolic blood pressure,
diastolic blood pressure, cigarettes/day - that showed a
significant relationship with alcohol consumption when consumption was dichotomized into heavy and not heavy. For
this group of men data were also available on the long-term
history of alcohol use. It was thus possible to identify 67 men
who had stopped drinking prior to the start of the study.
Thus, using the maximum consumed, the 1,782 men consuming less than 5 drinks/day are subdivided into former
drinkers, teetotalers (zero consumption on each of the three
surveys), < 1 drink/day, 1- < 2 drinks/day, 2-< 3
drinks/day, 3- < 4 drinks/day, and 4- < 5 drinks/day.
The heavy drinkers are also divided into those consuming
5- < 6 per day and those consuming 6 or more per day.
Table 3 shows that mean systolic and diastolic blood
pressures both rise with the amount of alcohol consumed.
The lowest mean pressures occur among the teetotalers and
among those consuming less than one drink/day. The rise in
mean blood pressure is fairly linear until one reaches the
highest consumption group, where the rise in mean systolic
pressure over the next highest group is 5.9 mm Hg and the
rise in mean diastolic pressure is 4.8 mm Hg. The differences
between the means of the six and over group and the group
consuming less than one per day are 15.1 mm Hg for systolic
blood pressure and 9.7 mm Hg for diastolic blood pressure.
Adjusting these differences for differences between the two
groups in age and relative weight reduces the differences in
mean blood pressure to 13.0 mm Hg and 7.8 mm Hg,
respectively. The adjusted differences between the 5- < 6/
day group and the < 1/day group are 9.1 mm Hg for
systolic blood pressure and 5.0 mm Hg for diastolic blood
pressure. Thus, each of the two groups that make up the
heavy drinkers have substantially higher blood pressures
than those who drink little or not at all, and these differences
cannot be accounted for by differences in age and relative
weight between the two groups.
With respect to cigarette consumption, the teetotalers
have a very low mean consumption - only 22.1% are
smokers. The mean number of cigarettes smoked per day
rises slowly with alcohol consumption up to those classified
as heavy drinkers, where there is a larger rise in cigarette
use. The other variables which were examined in the
previous analysis - pulse, serum cholesterol, and relative
weight - were also considered in the more detailed analysis.
TABLE 3. Mean Systolic Blood Pressure, Diastolic Blood
Pressure, and Cigarettes/day by Level of Alcohol Consumption,
Chicago Western Electric Company (1899 white males free of
definite CHD and Age 40-65 in 1957)
Alcohol
consumption
Former
Teetotaler
< 1/day
1- <2/day
2- <3/day
3- <4/day
4- <5/day
5- <6/day
> 6/day
Mean
systolic
N
pressure
67
77
841
135.0
132.4
132.7
332
208
109
148
72
45
134.8
137.0
137.9
139.1
141.9
147.8
Mean
Mean
diastolic
pressure
cigarettes
smoked
84.9
84.9
85.8
86.8
13.1
3.1
9.5
88.8
88.9
89.1
90.7
95.5
10.5
10.9
13.5
12.2
16.2
18.9
CIRCULATION
1070
TABLE 4. 15-year Mortality Rates by Drinking Status Adjusted for Age, Chicago Peoples Gas Company (I 323 white males
free of definite CHD and age 40-59 in 1968)
Cause
of death
All
CVR
CHD
Sudden
Cancer
Other
1195 nonproblem
drinkers
Deaths Rate
38 problem
drinkers
Deaths Ratet
20
13
11
7
1
6
247
127
87
40
74
46
529.8
343.7
290.3
184.3
27.3
158.8
206.4
106.2
72.8
33.5
t*
Relative
risk
4.96 4.33
4.63 4.41
4.94 5.21
4.82 6.52
61.9
38.5
3.68
4.71
*AIli values are significant at P <0.01.
tPer 1,000 population.
CVR - cardiovascular-renal; CHD
-
coronary heart disease.
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
The results showed no association between alcohol consumption and serum cholesterol. Mean pulse began to show
a slight rise at four or more drinks per day. Below this level
of consumption, the means were the same for each of the
groups. For relative weight, the means were the same
through the group consuming 3- < 4 drinks/day. Beyond
this level, the results were not consistent. The group with the
highest mean relative weight was the group consuming 6 or
more per day, but the group with the lowest mean relative
weight was the group consuming 5- < 6/day, the means being 122.2 and 113.7, respectively.
Alcohol and Long-term Mortality
Table 4 presents 15 year mortality rates by drinking status
for the Chicago Peoples Gas Company cohort. The rates are
age-adjusted by analysis of covariance, utilizing the SPSS
ANOVA program, and the multiple classification analysis
option. Where appropriate, t-values from the analysis of
covariance are given for a test of the hypothesis that the
rates are the same for the two groups. The relative risk of
death among the problem drinkers is also presented. (In this
study relative risk is calculated by the formula,
[R,/R2].[(1000-R2)/(l000-R1)], where R1 and R, are the
mortality rates per 1,000 population in the two groups. For a
discussion of why this measure of relative risk is superior to
R,/R,, see reference 7.) The problem drinkers have a
significantly higher mortality rate for the several causes of
death, except cancer. The highest relative risk for the
problem drinkers is for sudden death. Sudden death in the
Gas Company was defined as death within one hour of onset
of symptoms. In the Western Electric study, the criterion for
sudden death was death within three hours of onset of symptoms.
VOL 56, No 6, DECEMBER 1977
For Western Electric table 5 presents the 10 year mortality rates age-adjusted by analysis of covariance for the 67
former drinkers, the 117 heavy drinkers, and the remaining
1,715 men. The highest mortality rate occurred among the
ex-drinkers. Their rate for death from all causes is more
than four times the rate among the 1,715. Those who stop
drinking, whether originally light or heavy, presumably do
so for reasons of health. Compared to the 1,715, the heavy
drinkers have significantly higher rates for all causes of mortality and for mortality from cancer and other causes.
Although the heavy drinkers have higher rates for both cardiovascular-renal diseases and coronary heart disease, these
differences are not statistically significant when compared to
rates of the 1,715.
The association between mortality from all causes and
alcohol consumption was also examined in greater detail in
Western Electric, analagous to the results presented in table
3. Table 6 presents the 10-year all cause mortality rates ageadjusted by analysis of covariance for the various levels of
alcohol consumption for the Western Electric cohort. The
results show essentially no increase in mortality with alcohol
consumption until one reaches 5 or more drinks/day, at
which point the mortality rate essentially doubles. For the
three individual surveys, the mortality rates were comparable for those drinking five or more drinks per day, the
age-adjusted rates being 153.4 for the 1958 physical exam
(56 men), 174.0 for the 1959 physical exam (46 men), and
144.4 for the nutrition survey (50 men). For each of the individual surveys, the rate in the highest group was also more
than double the rate in the next highest group, i.e., the group
drinking 4 or more per day but less than 5.
These univariate mortality findings and the indicated
associations between blood pressure and alcohol consumption, and between cigarettes and alcohol consumption pose
the important question: How much of the increased mortality among the problem drinkers in the Gas Company and
among the heavy drinkers in the Western Electric Company
is attributable to the higher levels of these and other risk factors? Table 7 presents the results of an analysis of this
matter for the problem drinkers in the Gas Company. The
first column in this table gives the expected number of deaths
among the problem drinkers based on the rates among the
cohort as a whole, ignoring drinking status. The second
column gives the expected number of deaths in the 38
problem drinkers based on their baseline values for age,
diastolic blood pressure, heart rate, serum cholesterol,
relative weight, and cigarettes/day. The coefficients for the
logistic regression model were estimated for the cohort ig-
TABLE 5. 10-year Mortality Rates by Drinking Status Adjusted for Age, Chicago Western Electric Company
(1899 white males free of definite CHD and age 40-65 in 1967)
Cause
of death
All
CVR
CHD
Sudden
Cancer
Other
*P <0.05.
Past only (67)
Deaths
Ratet
17
8
8
7
6
3
**P <0.01.
tHeavy vs nonheavy.
*Per 1,000 population.
245.2
114.8
115.2
101.7
87.1
43.4
Heavy (117)
Deaths
Rate
17
7
6
3
5
5
143.8
59.0
50.6
25.2
42.3
42.5
Nonheavy (1715)
Relative
Riskt
Deaths
Rate
tt
99
59
58.1
3.59**
1.32
2.72
34.7
53
35
16
24
31.1
20.6
9.4
14.1
1.11
1.66
2.94**
2.32*
4.27
3.10
1.74
ALCOHOL, RISK FACTORS, AND MORTALITY/Dyer et al.
TABLE 6. 10-year Age-Adjusted Mortality Rates for Death
from All Causes by Level of Alcohol Consumption, Chicago
Western Electric Company
Alcohol
consumption
Former
Teetotaler
< 1/day
1- <2/day
2- <3/day
3- <4/day
4- <5/day
5- <6/day
26/day
N
Deaths
67
77
841
332
208
109
148
72
45
17
6
49
18
11
9
6
10
7
Rate/1000
245.2
75.0
58.7
55.6
55.7
78.5
40.3
135.4
157.2
noring drinking status, and then these coefficients were used
a probability of death for each problem drinker
based on the values of his baseline variables. These individual probabilities were then summed to give the number
of expected deaths. (For cause-specific mortality the competing risk logistic discrimination model was used8.) Based
on the levels of their risk factors, the problem drinkers could
be expected to have mortality rates from all causes 50%
higher than the entire cohort, 45% higher for cardiovascular-renal causes, and 65% higher for coronary heart
disease. Over and above the excess mortality attributable to
these risk factors, there remains a substantial excess from all
causes, cardiovascular diseases, coronary heart disease, and
sudden death that cannot be explained. This excess mortality
ranges from 62% for death from all causes to 158% for
sudden death.
Table 8 presents a similar analysis for the heavy drinkers
in the Western Electric study. The first column of expected
deaths was calculated from the mortality rates in the heavy
and nonheavy drinkers combined. The second column of expected deaths was calculated from the logistic models fitted
to the entire cohort, with ex-drinking included as a 0-1
variable. A comparison of the numbers of expected deaths
for all cause mortality shows that the differences in the risk
factors at baseline raise the expected mortality rate among
the heavy drinkers by 47%. Although the risk factor
differences account for some of the excess mortality among
the heavy drinkers, there remains an additional excess of
56% that cannot be explained by the differences in age,
diastolic blood pressure, relative weight, pulse, serum
cholesterol, and cigarettes/day.
Among the problem drinkers in the Gas Company, the excess mortality is due primarily fo cardiovascular diseases,
whereas in the Western Electric study the excess mortality
among the heavy drinkers is due primarily to noncardiovascular causes. In each study a logistic analysis was also
performed with age, diastolic blood pressure, serum
cholesterol, cigarettes/day, relative weight, and heart rate or
pulse in the model in addition to a variable coded 0-1 for
heavy alcohol consumption. In the Gas Company, with the
other variables controlled, problem drinking was
significantly related to mortality from all causes, cardiovascular diseases, coronary heart disease, and sudden death.
In the Western Electric study, heavy drinking was
significantly related to mortality from all causes and noncardiovascular causes, but not to cardiovascular causes or coronary heart disease.
to calculate
1071
TABLE 7. Observed and Expectd 15-year Mortality among
38 Problem Drinkers, Chicago Peoples Ga Company
Type of death
Expected*
deaths
Expectedt
deaths
Observed
deaths
Ratio:
Obs/Expt
All causes
CVR
CHD
Sudden
Non-CVR
8.23
4.31
3.02
1.45
3.91
12.38
20
13
11
7
7
1.62
2.07
2.21
2.58
1.15
6.26
4.97
2.71
6.11
*Based on mortality rate in entire cohort of 1233.
tBased on mortality rate in 1233 controlling for age, diastolic blood
pressure, heart rate, serum cholesterol, relative weight, and cigarettes/day.
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
Discussion
The results of these two studies are consistent in showing
an association between heavy alcohol intake and increased
blood pressure. Analyses from Framingham" showed no
association between alcohol consumption and blood
pressure when alcohol consumption was broken down into
the groups: no intake, occasional, moderate, or heavy.
However, when the heavy group was further subdivided to
identify those consuming more than 100 ounces of alcohol
per month, the mean systolic pressure of this group was
about 7 mm Hg higher than that of the entire cohort. This
difference is similar to that found for the Gas Company
problem drinkers and the Western Electric heavy drinkers,
compared to all others in each of these two cohorts. In a
study of 922 men identified as problem drinkers working for
the Du Pont Company, Pell and D'Alonzo'0 found a
prevalence of hypertension among the problem drinkers 2.3
times as great as that of the controls, based on a systolic
pressure _ 160 or a diastolic pressure 2 95. Klatsky,
Siegelaub, Friedman, and Gerard"' found a significant
relationship between the amount of alcohol consumed and
blood pressure among men in the Kaiser-Permanente study.
Among Copenhagen men age 40-59,12 the differences in
systolic and diastolic blood pressure between those consuming six or more drinks per day and those consuming less
were 8.0 mm Hg and 4.5 mm Hg, respectively. These
differences are comparable to those reported here from the
Gas Company, but somewhat lower than those reported for
those consuming 6 or more drinks/day from the Western
Electric study. Thus, although studies in mice and dogs have
shown a decrease in blood pressure with chronic ingestion of
alcohol,"3 14 the findings documented in the present report
corroborate those from other epidemiological studies on an
association between heavy alcohol consumption and increased blood pressure and prevalence of hypertension in
man.
Whether the effect of alcohol on blood pressure is transient or long-term is not entirely clear. In the Framingham
TABLE 8. Observed and Expected 10-year Mortality among
1127 Heavy Drinkers, Chicago Westen Electric Company
Type of death
All causes
CVR
Non-CVR
Expected*
deaths
7.41
4.22
3.19
deaths
Observed
deaths
10.89
5.85
5.14
17
7
10
Expectedt
Ratio:
Obs/Expt
1.56
1.19
1.95
*Based on mortality rate in 1,832 heavy and nonheavy drinkers.
tBased on mortality rate in 1,899 controlling for age, diastolic blood
presure, serum cholesterol, pulse, relative weight, cigarettes/day, and
oxdine.
1072
CIRCULATION
Study,9 individuals reported having a drink just prior to the
physical examination, thus raising the question as to
whether or not the higher blood pressures in heavy drinkers
might be an acute effect. Grollman" noted that among non-
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
drinkers small amounts of alcohol caused transient
elevations in cardiac output, pulse, and blood pressure, and
larger doses further elevated blood pressure and increased
cardiac output. Among habitual drinkers the rises in blood
pressure and cardiac output acutely were smaller. The findings of Pell and D'Alonzol° suggest that the effect of alcohol
on blood pressure is long-term rather than transient, since
recovered alcoholics in that study had a higher prevalence
rate of hypertension than the controls, although the
differences were smaller than those between the known
alcoholics and controls. No data in the present studies were
available for the examination of this question.
With respect to the other risk factors considered in this
report, only cigarette smoking was significantly related to
heavy alcohol intake in both studies. An association between
heavy alcohol consumption and cigarette smoking was also
noted in Framingham'6 and by Schmidt and de Lint.17 The
problem drinkers in the Gas Company had significantly
lower relative weights than the rest of the men, while the
heavy drinkers in Western Electric had slightly higher
relative weights than the rest of that cohort. Mean heart rate
was significantly higher among the problem drinkers in the
univariate analyses, but when this difference was adjusted
for the other variables, particularly blood pressure, it was no
longer statistically significant. In the Du Pont Company
study, Pell and D'Alonzo"° found the prevalence of pulses
above 90 beats/min significantly higher among the problem
drinkers than the controls. The fact that the recovered
alcoholics in this study had only a slightly higher prevalence
of pulses above 90 than the controls suggests that if heavy
alcohol intake is associated with increased heart rate, this
association may be reversible with abstinence from alcohol.
Neither study showed an association between high alcohol
intake and serum cholesterol. This finding merits comment
from two aspects: first, it is reasonable to postulate that
autopsy studies reporting less atherosclerosis in derelict
alcoholics, based on material from three or more decades
ago'8, dealt with undernourished alcoholics whose serum
cholesterol levels were likely for nutritional reasons to be
sizably lower on the average than those of men of the same
age from the population at large. Apparently, even problem
drinkers today, at least those able to remain gainfully
employed, have nutritional patterns resulting in serum
cholesterol levels similar to American men generally. On the
other hand, evidence is available indicating that heavy
drinkers might be expected to develop higher levels of serum
cholesterol than those present in persons using alcohol
moderately or not at all. Thus, metabolic data indicate that
alcohol leads to increased synthesis of cholesterol.'9 Furthermore, an association has been repeatedly reported between
sizable alcohol intake and hyperprebetalipoproteinemia
(high levels of plasma very low density lipoproteins), with
associated high levels of serum triglycerides and
cholesterol.20' 21 Given this presumed phenomenon, it is unclear why higher mean serum cholesterols were not observed
in heavy drinkers in either of the cohorts reported on here,
unless this associatio'n is not really ethanol specific, but
VOL 56, No 6, DECEMBER 1977
rather related to caloric balance. In this regard, it is
noteworthy that the problem drinkers in the Gas Company
had a mean relative weight that was significantly lower than
that of the rest of the cohort.
Epidemiological studies in the United States and Canada
have been highly consistent in showing an association
between alcohol consumption and all cause mortality, and
between alcohol consumption and mortality from specific
types of cancer. The evidence for an association with certain
types of cancers is quite strong and has been reviewed
thoroughly."2 With respect to all cause mortality, Schmidt
and de Lint7'23 found that the mortality rate among
alcoholics admitted to addiction treatment facilities from
1951 to 1964 in Toronto was 2.03 times the rate for the
general adult population in Toronto. The California Division of Alcohol Rehabilitation"2 in a follow-up study of
treated alcoholics found an overall mortality rate among the
alcoholics 2.36 times as great as the rate in the general population. In the Du Pont Company study,'5 the problem
drinkers had a 5-year mortality rate 3.22 times that of the
controls. Data from Framingham22 showed that those consuming 50 or more ounces of alcohol per month and
abstainers each had the highest all cause mortality rates. In
contrast to these U.S. findings, Dahlgren26 found no excess
mortality among alcoholics in Sweden from 1939 to 1947.
Thus, the findings of an association between problem drinking and all cause mortality in the Gas Company, and
between heavy drinking and all cause mortality in Western
Electric corroborate the generally positive findings of other
U.S. and Canadian epidemiological studies. In addition,
these two studies showed that this excess mortality was not
entirely explained by differences in age, diastolic blood
pressure, heart rate or pulse, serum cholesterol, relative
weight, or cigarettes/day between the heavy drinkers and
the others in each of the cohorts.
Although the epidemiological evidence of an association
between all cause mortality and alcohol consumption is
quite strong, this is not the case with respect to the findings
on alcohol consumption and morbidity and mortality from
cardiovascular diseases and coronary heart disease. Among
studies on mortality of alcoholics who are part of rehabilitation and treatment programs, Schmidt and de Lint7'23
found the alcoholics with 1.74 times the mortality from
arteriosclerotic and degenerative heart disease as the general
adult population of Toronto. However, the alcoholics
followed by the California Division of Alcohol Rehabilitation"2 had a cardiovascular disease mortality rate which
was only 58% of the rate in the general population. In addition, Dahlgren26 found no excess cardiovascular mortality
among the alcoholics in his Swedish study.
Among studies relating reported alcohol consumption to
cardiovascular diseases, Framingham16 found no association
between alcohol consumption and coronary heart disease,
congestive heart failure, or intermittent claudication. In a
case-control analysis, the Kaiser-Permanente Epidemiologic
Study'7 showed a significant negative association between
alcohol consumption and myocardial infarction. Data from
the Tecumseh Health Study22 showed the former drinkers
with the highest rates of coronary heart disease. There was
little or no difference in CHD rates among nondrinkers,
light drinkers, and those consuming an average of four or
ALCOHOL, RISK FACTORS, AND MORTALITY/Dyer et al.
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
more ounces of absolute alcohol per week. The results of the
Los Angeles Heart Study22 also showed no association
between heart disease and alcohol consumption. In a study
of male twin pairs in Sweden and the United States,28 heavy
consumption of alcohol based on the occasional consumption of one pint of whiskey, two bottles of wine, or four
quarts of beer, was found to be related to an increased
prevalence of angina pectoris in both countries. Among U.S.
twin pairs discordant on heavy drinking, the relative risk of
angina with heavy alcohol consumption was 2.2 for
monozygous twin pairs and 1.7 for dizygotic twin pairs. In
Western Electric, although the heavy drinkers had higher
rates of mortality from the cardiovascular diseases and coronary heart disease, the differences were not statistically
significant.
Among those studies that examined cardiovascular disease among problem drinkers or the intemperate, Pell and
D'Alonzo20 found the Du Pont Company problem drinkers
with a 5-year cardiovascular mortality ratio of 2.28. This excess cardiovascular mortality could not be attributed to the
higher prevalence of hypertension at baseline among the
problem drinkers. For the problem drinkers in the Chicago
Peoples Gas Company study, the mortality ratios for cardiovascular-renal disease, coronary heart disease, and sudden
death were 3.24, 3.99, and 5.50, respectively. In a 10-year
longitudinal study of 50-year old men in Sweden,
Wilhelmsen, Wedel, and Tibblin29 found that heavy alcohol
consumption based on registration with the Swedish
Temperance Board was significantly - independent of
blood pressure and smoking - related to the development
of nonfatal acute myocardial infarction and sudden death
from CHD. Thirty-nine per cent of the men who developed
CHD during the study had at some time been registered with
the Temperance Board, while only 20% of the others had
ever been registered.
Thus, the findings of the epidemiological studies on the
association between heavy alcohol intake and cardiovascular
diseases range from the significantly negative results of the
Kaiser Study27 to the significantly positive results of the
study by Wilhelmsen, Wedel, and Tibblin29 and the Gas
Company and Du Pont Company studies. Why are the
results of these studies so inconsistent? Part of the answer to
this question may lie in the nature of the individuals studied
and in the way alcohol consumption has been assessed. The
cited studies fall generally into three categories - those that
relate CHD to the reported amount of alcohol consumed per
week or per month, those that relate CHD to problem drinking or alcoholic intemperance, and those that compare the
CHD mortality rates of alcoholics who are part of
rehabilitation and treatment programs with the CHD rates
in the general population. The individuals represented by
these categories may be quite distinct, e.g., the heavy drinker
in Western Electric or Framingham may be quite unlike the
problem drinker in the Gas Company or the Du Pont Company, who in turn may be quite dissimilar to an alcoholic in
a rehabilitation or treatment program. Given the marked
divergence of the results of the Toronto and California
studies on alcoholics in treatment programs, it is not even
clear whether or not the individuals studied are similar
within categories.
Among those studies which examined reported alcohol
1073
consumption and CHD, the only positive findings were those
of the twin study.28 All other studies in this category showed
either no association or a significant negative association.
Although the lack of positive findings in this category may
reflect the truth, this outcome may also be due to other
causes. For example, the nonsignificant results in Western
Electric may be due to a paucity of numbers rather than to
the lack of a real association. In addition, the definition of
heavy drinking has varied widely, e.g., in Western Electric
the cutpoint for heavy was about 100 ounces of alcohol per
month, in Framingham 50 ounces, and in Tecumseh, 16. The
positive results in the three studies on problem drinkers and
the intemperate offer the possibility that alcohol consumption may be related to increased risk only at extreme levels
of consumption, with moderate use showing no increase in
risk. The Western Electric data lend some support to this
possibility, since there was no increase in all cause mortality
with increasing alcohol consumption until one reached the
level of 5 or more drinks/day. Thus, the possibility exists
that studies which have utilized reported amount consumed
have not accurately isolated the increased risk heavy
drinker, either through the use of too low a cutpoint or
through inaccuracies in the way the amount consumed was
reported. In Western Electric three different questionnaires
were used for the characterization of heavy and we were able
to show an association with mortality from all causes above
5 drinks/day.
The two studies on problem drinking in industry, i.e., the
Du Pont Company study and the Gas Company study
reported here, as well as the study of Wilhelmsen, Wedel,
and Tibblin,29 were consistent in showing a positive association between problem drinking and cardiovascular disease.
However, even the positive findings of these studies must be
viewed with some caution. Are the findings on coronary
death real, or are they artifactual, based on an incorrect
diagnosis of death? Heavy alcohol consumption might be
associated with coronary like clinical syndromes and sudden
death from causes other than severe atherosclerosis. For the
Gas Company, the cause of death coding was based
primarily on the death certificate, and on autopsy data and
hospital records only when available. Thus, the Gas Company results could be due to an incorrect diagnosis of death.
Although the epidemiological evidence on the role of
alcohol in the etiology and epidemiology of the atherosclerotic diseases is inconclusive, most clinicians seem to
now agree that excessive use of alcohol has deleterious
effects on the heart.30 31 Before bread was enriched with B
vitamins, alcoholics tended to develop heart disease due to
nutritional deficiencies, e.g., beri-beri.30 However, beri-beri
is now very uncommon in the United States and occurs only
in alcoholics with almost no intake of solid food. Today, the
clinician is often confronted by an individual with heart muscle damage who has consumed large amounts of alcohol
over a long period of time, has a good job, and eats amply.20
As a result of such encounters, there is a growing body of
literature on so-called alcoholic cardiomyopathy, or
alcoholic heart muscle disease, the symptoms of which have
been described.30 3 In addition to these clinical findings on
the effects of alcohol on the heart, animal experimental
studies on the effects of alcohol ingestion on mice and
dogs34 35have shown that excessive alcohol intake - even in
1074
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
the presence of an adequate diet - can damage the myocardium, and that, in fact, alcohol is a cardiotoxin. Furthermore, these results35 suggest that the toxic effect of alcohol
on the heart is cumulative.
While the clinical and animal experimental evidence indicate that excessive use of alcohol can damage the heart, it
is not clear whether or not there is any association between
such damage to the heart muscle and coronary artery
atherosclerosis. Autopsy studies have tended to show
chronic alcoholics with less atherosclerosis than nonalcoholics,36 and these results have been viewed as suggesting
that chronic alcoholism or cirrhosis of the liver might be
protective against the development of atherosclerosis.
However, Wilens'l and others37 have pointed out that the
lower prevalence of atherosclerosis among alcoholics at
necropsy was not attributable to alcoholism perse, but was
due to associated differences in such variables as age and
blood pressure.
The precise nature of the role, if indeed there is one,
between excessive use of alcohol and the process of atherogenesis remains to be elucidated. It is, however, clear that
heavy alcohol intake is related to both increased blood
pressure and the prevalence of high blood pressure.
Acknowledgments
It is a pleasure to express appreciation to the Officers and executive
leaderships of both the Peoples Gas Company and the Western Electric Company in Chicago; their continuous cooperation and support over the years
since the late 1950s made these two studies possible. Thanks are also due to
the research staff of the Peoples Gas Company Study: Elizabeth Stevens,
R.N.; research assistant, Wanda Drake and Celene Epstein; also to Patricia
Collette, M.A.; and to the staff of the Company Medical Department,
Patricia Benson, R.N., Carl Checchia, M.D., Luis Evangelista, M.D., John
Finch, M.D., Mary Ann Guzik, R.N., Gerald Holtz, M.D., Ruth Jaffke,
Aaron Kerlow, M.D., Robert Matthews, M.D., Hermine Mizelle, R.N., H.
Bates Noble, M.D., Harry Petrakos, M.D., Henry Ruder, M.D., Grace
Smith, Elsie Traina, R.N. and David Trish.
The following physicians were active in the Western Electric Company
Study and their invaluable assistance is gratefully acknowledged: Maurice
Albala, M.D., Harry Bliss, M.D., Herschel Browns, M.D., Marvin Colbert,
M.D., Henry De Young, M.D., Peter Economou, M.D., Sanford Franzblau,
M.D., Robert Felix, M.D., John Graettinger, M.D., Buford Hall, M.D.,
Wallace Kirkland, M.D., Joseph Muenster, M.D., Hyman Mackler, M.D.,
Adrian Ostfeld, M.D., Robert Parsons, M.D., Charles Perlia, M.D., William
Phelan, M.D., Norman Roberg, M.D., Marvin Rosenberg, M.D., George
Saxton, M.D., Armin Schick, M.D., John Sharp, M.D., Jay Silverman,
M.D., Donald Tarun, M.D. and Walter Wood, M.D.
The biochemical analyses for the Peoples Gas Company studies were done
in the research laboratories of the Chicago Board of Health and Chicago
Health Research Foundation, under the direction of Morton B. Epstein,
Ph.D., and Howard Adler, Ph.D.; we are also pleased to acknowledge the fine
contribution of the two chief technicians, Dana King and Ika Tomaschewsky.
It is also a pleasureto express appreciation to Tom Tokich who helped in
the computer programming.
2.
3.
4.
5.
Berkson D, Lindberg H: Risk factors:
6. Dyer A, Stamler J, Berkson D, Lindberg H: Relationship of relative
weight and body mass index to 14-year mortality in the Chicago Peoples
Gas Company study. Chron Dis 28: 109, 1975
7. Ipsen J, Feigl P: Bancroft's introduction to biostatistics. New York,
Harper & Row, 1970, pp 149-151
8. Dyer A: An analysis of the relationship of systolic blood pressure, serum
cholesterol, and smoking to 14-year mortality in the Chicago Peoples Gas
Company study. II: Coronary and cardiovascular-renal mortality in two
3
competing risk models. J Chron Dis 28: 571, 1975
9. Dawber T, Kannel W, Kagan A, Donabedian R, McNamara P, Pearson
G: Environmental factors in hypertension. In The Epidemiology of
Hypertension, edited by Stamler J, Stamler R, Pullman T. New York,
Grune & Stratton, 1967, p 255
10. D'Alonzo C, Pell S: Cardiovascular disease among problem drinkers. J
Occup Med 10: 344, 1968
11. Klatsky A, Siegelaub A, Friedman G, Gerard M: Alcohol consumption
and blood pressure. Paper presented to Council on Epidemiology of the
American Heart Association, New Orleans, Louisiana, February 28,
1976
12. Gyntelberg F, Meyer J: Relationship between blood pressure and
physical fitness, smoking and alcohol consumption in Copenhagen males
aged 40-59. Acta Med Scand 195: 375, 1974
13. Maines J, Aldinger E: Myocardial depression accompanying chronic
consumption of alcohol. Am Heart J 73: 55, 1967
14. Webb, W, DegertiI: Ethyl alcohol and the cardiovascular system: Effects
on coronary blood flow. JAMA 191: 1055, 1965
15. Grollman A: Metabolic observations in essential hypertension. Postgrad
Med 33: 532, 1963
16. Kannel W, Woolsey P: Alcohol and cardiovascular risk. (abstr) Circulation 52 (supplII): 11-200, 1975
17. Schmidt W, de LintJ: Causes of death in alcoholics. Quart J Stud Alc 33:
171, 1972
18. WilensS: The relationship of chronic alcoholism to atherosclerosis.
JAMA 135: 1136, 1947
19. Nestel PJ, Simons LA, Homma Y: Effects of ethanol on bile acid and
cholesterol metabolism. Am J Clin Nutrit 29: 1007, 1976
20. Fredrickson DS, Levy RI, Lees RS: Fat transport in lipoproteins An
integrated approach to mechanisms and disorders. N Engl J Med 276: 34,
94, 148, 215, 273, 1967
21. Baraona E, Liever CS: Hyperlipemia and ethanol. In Nutrition and Cardiovascular Disease, edited by Feldman E. New York, AppletonCentury-Crofts, 1976, p 158
22. U.S. Department of Health, Education and Welfare: Second special
report to the U.S. Congress on alcohol and health from the Secretary of
Health, Education and Welfare. Preprint Edition, 1974.
23. Schmidt W, de Lint J: Mortality experience of male and female alcoholic
patients. Quart J Stud Alc 30: 112, 1969
24. California State Department of Public Health Alcoholic Rehabilitation
Division: Followup studies of treated alcoholics. Pub. No. 6, Alcoholism
and California, Berkeley, 1961
25. PellS, D'Alonzo C: A five year mortality study of alcoholics. Occup
Med 15: 120, 1973
26. Dahlgren KG: On death rates and causes of death in alcohol addicts.
Acta Psychiat Neur Scand 26: 297, 1951
27. Klatsky A, Friedman G, Siegelaub A: Alcohol consumption before
myocardial infarction. Results from the Kaiser-Permanente
Epidemiological Study of myocardial infarction. Ann Intern Med 81:
294, 1974
28. Hrubec Z, Cederlof R, Friberg L: Background of angina pectoris: Social
and environmental factors in relation to smoking. Am Epidemiol 103:
16, 1976
29. Wilhelmsen L, Wedel H, Tibblin G: Multivariate analysis of risk factors
for coronary heart disease. Circulation 48: 950. 1973
30. Brandfonbrenner M: The effects of alcohol on the heart. Heartbeat 1:
No. 8, 1976
31.
G, Giles T: Alcoholic cardiomyopathy, concept of the disease and
its treatment. Am J Med 50: 141, 1971
32. Evans W: Alcoholic cardiomyopathy. Am Heart 3 61: 556, 1961
33. Demakis JG, Proskey A, Rahimtoola SH, Jamil M, Sutton GC, Rosen
KM, Gunnar JR, Tobin JR: Natural course of alcoholic cardiomyopathy. Ann Intern Med 80: 293, 1974
34. Bureh G, Colcolough H, Harb 3, Tsui C: The effect of ingestion of ethyl
alcohol, wine and beer on the myocardium of nice. Am 3 Cardiol 27:522,
1971
35. Ettinger P, Lyons M, Oldewurtel H, Regan T: Cardiac conduction abnormalities produced by chronic alcoholism. Am Heart J 91: 66, 1976
36. Hirst AE, Hadley GG, Gore I: The effect of chronic alcoholism and
cirrhosis of the liver on atherosclerosis. Am J Med Sci 249: 143, 1965
37. Parrish H, Eherly A: Negative
association of coronary atherosclerosis
with liver cirrhosis and chronic alcoholism
a statistical fallacy.
J
Ind
Med Assoc 54: 341, 1961
J
J
Burch
References
1. Stamler J,
VOL 56, No 6, DECEMBER 1977
CIRCULATION
Their role in the
etiology and pathogenesis of the atherosclerotic diseases. In The
Pathogenesis of Atherosclerosis, edited by Wissler R, Geer J. Baltimore,
Williams & Wilkins, 1971, p 41
Dyer A, Stamler J, Berkson D, Lindberg H, Stevens E: High blood
pressure: A risk factor for cancer mortality? Lancet 1: 1051, May 1975
Paul 0, Lepper M, Phelan W, Dupertuis G, MacMillan A,McKean H,
Park H: A longitudinal study of coronary heart disease. Circulation 28:
20, 1963
Metropolitan Life Insurance Company: New weight standards for men
and women. Stat Bull Metropol Life Ins Co 40: 1. Nov-Dec 1959
Walker S, Duncan D: Estimation of the probability of event
function of several independent variables. Biometrika 54: 167, 1967
an
as a
-
Alcohol consumption, cardiovascular risk factors, and mortality in two Chicago
epidemiologic studies.
A R Dyer, J Stamler, O Paul, D M Berkson, M H Lepper, H McKean, R B Shekelle, H A
Lindberg and D Garside
Downloaded from http://circ.ahajournals.org/ by guest on June 15, 2017
Circulation. 1977;56:1067-1074
doi: 10.1161/01.CIR.56.6.1067
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1977 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on
the World Wide Web at:
http://circ.ahajournals.org/content/56/6/1067
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally
published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Editorial Office. Once the online version of the published article for which permission is being requested is
located, click Request Permissions in the middle column of the Web page under Services. Further
information about this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation is online at:
http://circ.ahajournals.org//subscriptions/