PATHOPHYSIOLOGY AND NATURAL HISTORY
HYPERTENSION
The relationships between alcoholic beverage use
and other traits to blood pressure: a new Kaiser
Permanente study
ARTHUR L. KLATSKY. M.D., GARY D. FRIEDMAN, M.D.,
AND
MARY ANNE ARMSTRONG, M.A.
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
ABSTRACT In a new study controlled for many factors, we reconfirmed the relationship of higher
blood pressure to alcohol use. This relationship was slightly stronger in men, whites, and persons 55
years of age or older. A slight increase in blood pressure appeared in men who drank one to two drinks
daily, and a continuous increase occurred at all higher drinking levels among white men who had
constant drinking habits. Among women, an increase occurred only at three or more drinks daily. The
data suggest complete regression, beginning within days, of alcohol-associated hypertension upon
abstinence. Blood pressure showed minor differences with beverage preference: those who preferred
liquor had higher adjusted mean blood pressure than those preferring wine or beer. The results of this
study contribute to the likelihood that the alcohol-blood pressure association is causal. Smoking,
coffee use, and tea use showed no association with higher blood pressure. Systolic pressure showed a
positive relationship to total serum calcium and an inverse relationship to serum potassium, but
diastolic pressure showed little relationship to these blood constituents; the explanations include a
possible direct effect on regulation of blood pressure.
Circulation 73, No. 4, 628-636, 1986.
A SUBSTANTIAL number of population studies have
almost unanimously shown an empiric link between
regular use of alcoholic beverages and hypertension.
Several recent reviews have been published.'-1 Two
intervention clinical studies have shown a pressor
effect of alcohol, operative within days of starting ingestion.9 1' Nevertheless, the absence of a proven
mechanism for the alcohol-hypertension relationship
prevents general acceptance of a causal association.
Other problems and inconsistencies in data reported
from population studies include disparity in age, sex,
race, type of alcoholic beverage, shape of the apparent
dose-response curve, and systolic compared with diastolic pressure. We present data from a new study
From the Department of Medicine, Division of Cardiology, and the
Department of Medical Methods Research, Kaiser Permanente Medical
Care Program, Oakland, CA.
The study protocol was approved by the regional Institutional Review
Board. The research was supported in part by The Alcoholic Beverage
Medical Research Foundation, Inc.? and in part by the Community
Service Program of the Kaiser Foundation Hospitals.
Address for correspondence: Arthur L. Klatsky, M.D., Department
of Medicine, Kaiser Permanente Medical Center, 280 W. MacArthur
Blvd., Oakland, CA 94611.
Received Aug. 5, 1985; revision accepted Dec. 19, 1985.
Presented in part at the 17th Annual Meeting of the Society for
Epidemiological Research, Houston, June 1984, and in part at the 34th
Annual Meeting of the American College of Cardiology, Anaheim,
March 1985.
628
designed to explore some of these problems. We also
examined the relationships between several other habits, health measurements, and demographic traits to
blood pressure. Because much of the current interest in
hypertension risk factors is focused on possible nutritional determinants, we include in this article data
concerning the relationship of blood pressure to the
following: adiposity, coffee or tea use, and concentrations of calcium, potassium, glucose, cholesterol, and
uric acid in the blood.
Methods
Study population and data collection. We collected data
from approximately 80,000 persons of several races who had
multiphasic health examinations" at the Oakland or San Francisco Kaiser Permanente Medical Centers from January 1978 to
December 1981. Included were all white and black men and
women who answered "no" to the questionnaire item, "Are you
now under treatment for hypertension (high blood pressure),"
and who provided the following: (1) a completed health history
questionnaire, (2) a special research alcoholic beverage history
questionnaire, (3) supine blood pressure measurement, (4)
height and weight measurement, and (5) blood test data. The
66,510 persons who met these requirements represented a wide
range of ages and socioeconomic characteristics. Most had the
procedure as a routine health examination.
Persons classified themselves racially by checking a labeled
box in a multiple-choice question. Similarly, information about
smoking, coffee or tea use, marital status, and education was
obtained on the questionnaires.
CIRCULATION
PATHOPHYSIOLOGY AND NATURAL HISTORY-HYPERTENSION
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
Alcoholic beverage information. The two sides of a sheet of
paper presented different sets of queries for nondrinkers and
drinkers. The nondrinkers' side was headed, "If you have taken
no alcoholic beverages in the past year, please answer the questions below. If you have taken any alcoholic beverages in the
past year, please turn this sheet over." The drinkers' side had the
converse heading.
Among the drinkers' questions was this item: "How many
alcoholic drinks (wine, beer, whiskey, cocktails) did you usually have during the past year?" The options offered were, in this
order, "nine or more per day," "six to eight per day," "three to
five per day," "one or two per day," "less than one per day but
more than one per month," "less than one per month (special
occasions only)." Drinkers were asked separate multiple-choice
questions about drinking in the 24 hr and week before examination. They were also asked separate items about use of "wine,"
"hard liquor," and "beer." Definitions of beverage preference
were derived from the wine, liquor, and beer questions.
Drinkers were asked also about drinking in the past year
compared with the previous 10 years and about weekend compared with weekday drinking. On the basis of responses to these
two questions plus the items about drinking in the past 24 hr,
past week, and past year, a drinking variability score was derived for each white man. Variability points were not assigned
to a man who reported drinking the same amount as usual in the
past year as during the past 10 years, the same amount as usual
in the past week, and the same amount as usual or only one
category different on our six-category scale for weekends,
weekdays, and the past 24 hr. Such persons were classified as
having low drinking variability; points were assigned for progressive variations on each question, yielding a range of variability scores from 1 to 12.
Nondrinkers were asked, "Did you drink alcohol in the past?"
Their options were "yes" or "never or almost never." Those who
answered "yes" were asked to indicate the largest amount of
daily drinking in the past for any period of 1 year.
Data showing racial patterns and other correlates of drinking
in this population have been published. 12 We have discussed in
another publication8 the general problems raised by possible
inaccurate reporting of alcohol intake in studies of this type. The
evidence suggests that in a nonjudgmental situation, such as
completing a questionnaire at our examination procedure, people report as much alcohol intake as at personal interviews.
Furthermore, the presumed tendency to underreport drinking is
far more likely to reduce the apparent strength of any association with alcohol than to artificially produce such an association.
Blood pressure and laboratory tests. Examinees were
asked to take no food or drink for at least 4 hr before their
appointment time, except for water, black coffee, or tea with no
sugar or cream. No specific instructions about smoking were
given. At a specific stage in the examination, patients were
asked to lie down on their backs and to relax. Blood pressure
was then determined with the patient in the supine position by a
Roche Arteriosonde Blood Pressure Monitor, Model 1216.
Blood chemistry tests were performed with a Technicon SMA
12-6 Multichannel Analyzer from January through May 1978;
thereafter, the tests were performed with an American Monitor
KDA Automated Blood Chemistry Analyzer.
Data analysis and presentation. Multiple regression analyses, using the SAS General Linear Models (GLM) program,
were performed for each race-sex group with systolic and diastolic pressure as dependent variables. Covariables (continuous
except where stated) in all analyses were age, Quetelet's adiposity index, highest educational attainment (five categories),
marital status (four categories), current cigarette smoking
("yes" or "no"), coffee and tea use (number of cups per day),
Vol. 73, No. 4, April 1986
hemoglobin level, leukocyte count, and six blood chemistry
tests (total cholesterol, uric acid, glucose, potassium, calcium,
and creatinine). Either adjusted mean blood pressure, computed
by the least-squares means option of GLM, or regression coefficients are reported for each analysis. Separate regression analyses were performed on the subsets of persons who fit the definition of those preferring wine, liquor, or beer. Other subsets used
for additional analyses included white men who took three or
more drinks daily, white male nondrinkers, white men with
specific scores of drinking variability, and white men and women whose blood calcium and potassium determinations fell into
designated ranges.
Results
Alcohol and blood pressure by race, sex, and age.
Among white men (table 1, figure 1, top left) who
drank alcoholic beverages less often than daily, there
was no significant alcohol-blood pressure relationship; among daily drinkers, a progressive rise in systolic and diastolic pressure was found that peaked at six
to eight drinks per day. White men taking nine or more
drinks per day had blood pressure similar to that of
men reporting three to five drinks - significantly
higher than in nondrinkers but lower (p < .001 for both
systolic and diastolic pressure) than in men reporting
six to eight drinks per day. White women (table 1,
figure 1, top right) showed similar but less marked
alcohol-blood pressure relationships to those of white
men; the substantially lower blood pressure of the 26
white women reporting nine or more drinks daily must
be interpreted in the context of the small number of
women in this drinking category.
Black men (table 2, figure 1, lower left) showed a
continuous relationship of more drinking to higher diastolic pressure at all drinking levels, but the elevations
in systolic pressure related to drinking leveled off at
three to five drinks per day. Black women (table 2,
figure 1, lower right) showed the least consistent relationship of blood pressure to intake of alcoholic beverages. Like black men, the black women showed a
closer association of diastolic than systolic pressure to
amount of drinking.
Breakdown of the data by age groups (figure 1)
showed that the alcohol-blood pressure relationship
was independent of the age subsets chosen, but a
slightly stronger relationship was found in the oldest
(-60 years of age) persons. Higher adjusted mean
blood pressure was found among heavier (three or
more drinks per day) drinkers in all race-sex-age subgroups.
Nondrinkers: role of past drinking. There was no racesex group in which those who drank in the past had
significantly higher blood pressure than lifelong abstainers. Among women of both races, in fact, the past
drinkers had slightly lower blood pressure than life629
KLATSKY et al.
TABLE 1
Alcohol, selected factors,A and blood pressure in 20,171 white men and 23,191 white women not on high blood pressure
treatment
Change in blood pressure (mm Hg) associated with factorB
White women
White men
Systolic
19)c
(R'
Factor
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
Alcohol (drinks/day)
None, lifelong (reference category)
None, past drinker
<1/mo
>l/mo, <liday
1-2/day
3-5/day
6-8/day
-9/day
Age (10 yr)
Quetelet index (1.0)
Cigarette smoker (yes/no)
Coffee (cups/day)
Tea (cups/day)
Total cholesterol (100 mg/dl)
Uric acid (mg/dl)
Glucose (100 mg/dl)
Potassium (meq/l)
Total calcium (mg/dl)
Creatinine (mg/dl)
Hemoglobin (g/dl)
0.0
0.4
1.4
1.2
2 .6E
5.6E
9. 1
4.8E
3.4E
5 8E
- 0.2
-0.4E
-0 3D
0.6
5.2 E
-1.7
21E
1.2
0.7E
Diastolic
(R`
=
.19)'
0.0
0.2
0.9
0.7
3.5E
5.6E
8D
1 E
3.5E
- 0.7E
0.0
-
1 7E
0 1
o. o.
-0.3
- 0.2
0.2
0.7
Systolic
= .35)(
(R'
0.0
2.4D
-0.8
-0.6
0.5
29E
4.6DF
-
Diastolic
(R'
=
.24)`
0.0
0.9
-0.1
0.1
1.4
-
31 E
2DF
6.2F_ 3 4F
l 9E
5.OE
-
5 5E
3.5EF
0.8[)
- 0. E
E
0.4 D
1 IE
0.4E
54E
0.9E
21E
0.3
00
0.0
1.9E
0.
1.3oEl.0O
-I,0E
0.3
0.1
0.9
l.5E
AAlso controlled for marital status and education.
BRegression coefficients; represents change in blood pressure independent of all other factors.
CR2 coefficient of multiple determination. which estimates proportion of total variation in blood pressure explained by these
factors.
=
Dp <
Ep <
Fn
=
.01.
.001.
118 for 6-8/day, 26 for -9/day.
long abstainers. Because the white men had the largest
proportion of past drinkers (651 of 1506 nondrinkers,
43.2%), we chose these for a more detailed analysis.
Of the 651 white men who were past drinkers, 342
(52.8%) stated that their maximal daily drinking had
been three or more drinks per day for at least 1 year.
Controlling for all variables, as in the other analyses,
the adjusted mean blood pressure was similar for these
subsets: lifelong abstainers, 124.8/73.5 mm Hg; past
drinkers of fewer than three drinks per day, 124.8/73.5
mm Hg; past drinkers of three drinks or more per day,
124.0/73.5 mm Hg. Because the number of past drinkers who reported previous daily drinking of three or
more drinks was quite small in other race-sex subgroups, we did not perform a similar analysis in these
subgroups.
Beverage preference. Small differences were seen in
adjusted mean blood pressure for persons whose alcoholic beverage histories indicated a clear preference
630
for wine, liquor, or beer (table 3). Those who preferred
wine had the lowest systolic pressure, those who preferred beer had the lowest diastolic pressure, and those
who preferred liquor had the highest systolic and diastolic pressure. We did not study size of drink as a
possible explanation for these differences.
Blood pressure and drinking in the day or week before
examination. In this analysis, we again studied white
men who reported their usual intake to be three or more
drinks per day in the past year. Among these heavier
drinkers, those who reported no intake in the past week
had substantially lower adjusted mean blood pressure
than men who reported any drinking in the past week
(table 4). Those who reported nine or more drinks in
the past 24 hr had higher adjusted mean blood pressure
than those who reported consuming lesser amounts of
alcoholic beverages.
Alcohol and blood pressure among white men with low
drinking variability. About 25% of white men reported
CIRCULATION
PATHOPHYSIOLOGY AND NATURAL HISTORY-HYPERTENSION
to that of nonsmokers. Coffee and tea use had a slight
inverse relationship to systolic pressure in all race-sex
subsets, but coffee and tea use had no consistent relationship to diastolic pressure.
I
White men
150
\%
>59
Laboratory tests
-40-59 _
130
_ .
<407
m) 110 P
E
90
E
>59 _
_
40-59
CO,
U)
<40
a)
a 60
<40
0
1F
0
Black women
Black men
c
a)
_
>59
-
>59
ao
a)
40 59
< 130
40-59
-*_
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
<40
._._.
110
90
>59
F
_
.
>59
=-40-59:..g
<40
.<40
=:'59
0
.<40
-
60~~~~~~~~~~~~~~4
60
Never Past <llmo <1
1-2 3-5 6-8 >9
Never Past dl/mo <1
1-2 3-5 6-8 >9
Alcohol consumption (usual number of drinks per day)
FIGURE 1. Adjusted mean systolic and diastolic blood pressures (mm
Hg) according to alcohol consumption by three age groups (top left,
white men; lower left, black men; top right, white women; lower right,
black women). Dashed lines and open circles indicate 10 < n < 25.
Data omitted from figure for categories with n < 11 (white women age
40 to 59 years, nine or more drinks/day and age > 59 years, six to eight
and nine or more drinks/day; black men age > 59 years, six to eight and
nine or more drinks/day; black women age 40 to 59 years, nine or more
drinks/day and age > 59 years, three to five, six to eight, and nine or
more drinks/day).
low variability in usual drinking (similar in past 24 hr,
week, year, 10 years; similar weekend and weekday
amounts) (table 5). These persons with no important
variability in drinking behavior exhibited an uninterrupted positive relationship between usual number of
drinks and systolic pressure from less than one drink
per day up through the range of drinking studied. This
trend was not reversed in the subgroup reporting nine
or more drinks per day. Among white women, black
men, and black women, there were insufficient numbers of persons with low drinking variability to allow
performance of a similar analysis of these subgroups.
Age, adiposity, smoking, and coffee and tea use. Age
and Quetelet's adiposity index showed the expected
strong positive relationship to blood pressure in both
races and both sexes (tables 1 and 2). White cigarette
smokers had slightly lower blood pressure than nonsmokers, but black smokers had blood pressure similar
Vol. 73, No. 4, April 1986
Calcium. Total serum calcium showed a positive relationship to systolic pressure in all race-sex groups, but,
for diastolic pressure, the calcium-blood pressure relationship was significant only in black men (tables 1
and 2). Figure 2 shows the relationship of total calcium
to blood pressure in white men (top left) and white
women (top right) in a different way, depicting adjusted mean blood pressure at various calcium levels.
Among white men, a clear progressive association of
higher adjusted mean systolic pressure to the categories chosen was seen throughout the normal calcium
range. Among white women, a similar progressive
association was present up to the highest calcium category. This calcium-blood pressure progression was
weak for diastolic pressure.
Potassium. This blood test showed an inverse relationship to systolic pressure in all race-sex groups and
was the only laboratory test to show a consistent inverse association with blood pressure. Analysis of
specified potassium categories in relation to adjusted
mean blood pressure (figure 2) showed a clear inverse
progression in white men (lower left) at all potassium
levels and in white women (lower right) up to the
highest potassium level. This potassium-blood pressure inverse progression was weak for diastolic pressure.
Other tests. Total cholesterol was positively related to
blood pressure in all race-sex groups; in general, this
relationship was stronger for systolic pressure. Uric
acid was positively related to systolic and diastolic
pressure, except in white men, in whom uric acid was
inversely related to systolic pressure. Glucose showed
a relationship to systolic pressure in all groups, but,
except for an inverse relationship in white women,
glucose showed little association with diastolic pressure. Creatinine showed no consistent relationship to
blood pressure. Hemoglobin showed a definite relationship to systolic and diastolic pressure in white men
and women, but, in black men and women, hemoglobin was related to diastolic pressure only.
Discussion
A direct positive relationship between regular use of
alcoholic beverages and higher blood pressure, independent of many potential confounders, has been confirmed by these data. The alcohol-blood pressure relationship in this study was generally similar for systolic
631
KLATSKY
et
at.
TABLE 2
Alcohol, selected factors,A and blood pressure in 6999 black men and 10,930 black women not on high blood pressure
treatment
Change in blood pressure (mm Hg) associated with factorB
Black women
Black men
Diastolic
Systolic
Systolic
(R.20)'(c = 1) (RR
.29)c
Factor
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
Alcohol (drinks/day)
None, lifelong (reference category)
None, past drinker
<1/mo
>1/mo, <1/day
I-2/day
3-5/day
6-8/day
>9/day
Age (10 yr)
Quetelet index (1.0)
Cigarette smoker (yes/no)
Coffee (cups/day)
Tea (cups/day)
Total cholesterol (100 mg/dl)
Uric acid (mg/dl)
Glucose (100 mg/dl)
Potassium (meq/l)
Total calcium (mg/dl)
Creatinine (mg/dl)
Hemoglobin (g/dl)
AAlso controlled for marital
0.0
0.4
0.6
1.5
2. 1 D
4.4E
4.4D
4D
4 oE
5.2E
- 0.5
- 0.4 D
- 0.5
0.7
0.4
3.8E
- 2.6'E
2.9E
- 0.7
-0. 1
0.0
- 0.4
0.3
0.9
l.2D
2.3E
3.4E
3.6E
2. 6E
3.'E
0.0
0.1
0.4
1.3'
0.5E
0.5
0.3
0.6
0.7
0.5E
Diastolic
(R
.23)C
0.0
-0.5
-0.4
0.2
0.0
2.0
- 1.5E
-0.7
0.8
0.6
-
-
1.6E
1.1
1.3F
2.6F
1.5F
2.4E
2.7E
1 8F
5 1E
4.6E
- 0.3
0.5
0. 1
0.1
1.7'
0.7E
- 0.3
0.5
-0. 1
- 0. 1
0. E
-0.4D
2.2E
0.8E
4. 1 E
- 0.6
0.9D
- 1.4
0.2
-
0.9E
status and education.
BRegression coefficients; represents change in blood pressure independent of all other factors.
CR2 = coefficient of multiple determination, which estimates proportion of total variation in blood pressure explained by these
factors.
Dp
< .01.
Ep <
Fn
=
.001.
51 for
6-8/day,
37 for >9/day.
TABLE 3
Beverage preferenceA and adjusted meanB blood pressure in white
menc and womenD
Beverage
Systolic (mm Hg)
Men
Women
Men
Women
Wine
Beer
133.93
135.08
136.86
130.66
133.46
80.70
80.40
81.80
76.16
75.95
76.96
133.87
ASee text for definition of preference.
BControlled for all traits in tables and 2, plus education, marital
status, and leukocyte count.
1447.
CWine n = 2070, beer n
1702, liquor n
1218.
DWine n = 3400, beer n
451, liquor n
and diastolic pressure. As in some previous studies,'3-15 the association was clearer and more continuous in men than in women. White men and women
show a more clear-cut and consistent alcohol-blood
pressure association than blacks of both sexes. Thus
632
Systolic
n
(mm Hg)
Diastolic
(mm Hg)
212
423
1990
170
133.85
134.20
134.14
127.82
79.53
79.60
79.70
76.18
67
307
1302
623
496
136.90
131.15
131.61
79.75
78.79
78.53
78.01
78.68
Diastolic (mm Hg)
preferred
Liquor
TABLE 4
Recent drinking and adjusted meanA blood pressure among white
men who reported usual intake of three or more drinks daily
Past week (vs usual)
MoreB
LessB
Same
None
No. drinks previous 24 hr
'9
6-8
3-5
1-2
None
131.51
131.35
AControlled for all traits in tables 1 and 2 except tea, plus education,
marital status, and leukocyte count.
B"MoreC and "less" were presumably answered with reference to
usual intake and have no necessary relationship to absolute amount
consumed in previous week.
CIRCULATION
PATHOPHYSIOLOGY AND NATURAL HISTORY-HYPERTENSION
TABLE 5
Alcohol and systolic blood pressureA among white men according to drinking variabilityB
Change in blood pressure (mm Hg) associated with alcoholc
Variability score
LowD (n
Usual No.
drinks
29/day
6-8/day
3-5/day
1-2/day
>1/mo, <1/day
< 1/mo
=
5063)
Change (n)
9.5
7.4
5.7
1.3
-0.2
0.0E
(16)
(79)
(525)
(1610)
(1972)
(861)
MediumD (n = 6957)
HighD (n = 8185)
p value
Change (n)
p value
Change (n)
p value
.01
<.001
<.001
.05
.68
4.1
12.4
4.3
1.5
0.0
0OE
(18)
(112)
(693)
(1805)
(3279)
(1050)
.24
<.001
<.001
.01
.95
3.0
6.3
3.9
1.7
0.3
0OE
(88)
(273)
(1290)
(2775)
(3541)
(218)
.09
<.001
<.001
.08
.73
AAlso controlled for age, Quetelet's index, smoking, coffee, tea, education, marital status, and eight blood tests.
BSee text for definitions of variability.
CRegression coefficients; represents change in blood pressure independent of all other factors.
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
DLow, variability score
EReference.
130
CD
cn
Rn
ouv
0; medium, variability score
[
120 r_---
=
White men
-
1
I
E
UI)
'O
708.50- 9.00- 9.50- 10.00- 10.508.99 9.49 9.99 10.49 10.99
a)
E0
8.50- 9.00- 9.50- 10.00- 10.508.99 9.49 9.99 10.49 10.99
Calcium categories (mg/dI)
0
0
cl
co
130-
E
-0
a)
120
White men
White women
3.51- 4.01- 4.34- 4.68- 5.014.00 4.33 4.67 5.00 5.50
3.51- 4.01- 4.34- 4.68- 5.014.00 4.33 4.67 5.00 5.50
80
70
Potassium categories (meq/liter)
FIGURE 2. Adjusted mean systolic and diastolic blood pressures (mm
Hg) at various total serum calcium levels (meq/liter) (top left, white
men; top right, white women) and at various serum potassium levels
(meq/liter) (lower left, white men; lower right, white women).
black women showed only a minimal alcohol-blood
pressure relationship. We cannot explain the race-sex
disparities.
A large-scale epidemiologic survey such as this one
Vol. 73, No. 4, April 1986
1 or 2; high, variability score -3.
cannot resolve the question of a possible threshold for
apparent risk of hypertension, i.e., the maximal safe
amount of drinking. The threshold question is controversial. Several studies have suggested little or no effect of up to one to two drinks per day on blood pressure.' 1, 16-20 Others have shown a progressive linear
association.2' 26 Even if the alcohol-blood pressure association proves to be causal, much individual variation in response will probably be found. A slight
increase in blood pressure was found in men reporting
as few as one to two drinks daily in this survey. In our
previous study,'3 we found an apparent threshold at
three to five drinks per day, but our two studies may
not be discrepant since all drinking below three drinks
per day was included in a single questionnaire category
in the earlier survey. As in our earlier survey and
several other reports,'5 19 20 women showed a slightly
curvilinear alcohol-blood pressure relationship; that
is, the lighter drinkers had slightly lower blood pressure than abstainers. In both our studies, the strongest
alcohol-blood pressure association was found at reported drinking levels of three to five and six to eight
drinks daily. Plateauing or a downturn in the curve for
some race-sex subgroups was present at higher drinking levels in both studies. However, in this study,
analysis of data about drinking during the day and
week before examination showed that white men who
took nine or more drinks up to the day of examination
showed no plateauing. Thus apparent plateauing of the
alcohol-blood pressure dose-response curve may be
largely caused by very recent lessening of alcohol intake among the very heaviest drinkers. Persons who
reported daily intake of nine or more drinks also reported substantially higher variability than persons
633
KLATSKY et al.
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
with lower alcohol intake. These data also suggest that
elevated blood pressure regresses rapidly (within several days) upon reduction of drinking. These findings
are compatible with the intervention study of Potter
and Beevers,9 who found an apparent pressor effect of
four pints of beer daily in 2 to 4 days and regression of
this effect within a similar period.
The data in this study show that reported past drinking has no relationship to blood pressure in current
abstainers. In all race-sex subgroups, former drinkers
had blood pressure similar to that of lifelong abstainers. Among white men, this was true even among
those who reported past intake of three or more drinks
for a year, an amount of drinking clearly associated
with higher blood pressure among current drinkers.
Studies with relevant data3 21 2'`4
26-8 have all been
consistent with partial or total reversibility of alcoholassociated hypertension. The importance of this apparent regression of alcohol-associated high blood pressure is obvious for practitioners who advise
hypertensive patients ingesting substantial amounts of
alcohol. These data provide compelling evidence for a
trial of reduction of alcohol intake in hypertensive persons who drink daily.
The slightly higher blood pressure of those who
prefer liquor and the lower blood pressure of those
preferring wine and beer are possibly related to the
amount of ethanol per drink. Although this assumption
of a causal relationship is the simplest explanation,
other possibilities exist. Preference for a beverage type
probably has dietary and other lifestyle correlates that
might be related to blood pressure. Taking an alcoholic
beverage with food (presumably the habit of many who
prefer wine and beer) might tend to blunt the fluctuations of alcohol level in the blood, and this might
lessen a possible effect on blood pressure. Yet another
factor could be the nutrient content of the various types
of beverages. The substantial potassium content of
wine, for example, might modulate a possible blood
pressure effect. In any case, the small variations associated with beverage preference may be of little biological importance.
The Lipid Research Clinics Prevalence Study'5
found an independent association of both beer and
liquor. Studies of populations consuming mainly one
beverage type have demonstrated an alcohol-blood
pressure relationship for use of beer,22 23 25.29. 30
wine,24 sake,26 or liquor and beer combined.3' The apparent independence from beverage type of the alcohol-hypertension link is compatible with the hypothesis that the relationship is an effect of ethanol.
Our data do not support the studies'4 32 that found an
634
alcohol-blood pressure relationship among older persons only. However, the association we found was
somewhat stronger among persons over 60 years of
age. Whether this represents a more sustained blood
pressure effect, a slightly different biological response, less variability of drinking behavior in older
persons, different confounders related to age, or some
other factor is speculative.
We shall comment selectively and only briefly on
the other factors studied. The strong consistent relationship of age and adiposity to blood pressure is well
known. The data suggest that use of tobacco, coffee,
and tea are not associated with elevated blood pressure. The hint of a slight inverse relationship of smoking to blood pressure has been reported by others.28 3
The tendency we observed for slightly lower blood
pressure associated with coffee use is compatible with
some previous data, 34 but another recent study35
showed a slightly positive relationship. The increase in
blood pressure associated with short-term administration of caffeine36 3 does not appear to translate into
sustained blood pressure increases. Alternatively, any
caffeine-associated effect on blood pressure may be
counterbalanced by some other lifestyle trait of coffee
drinkers that has a reducing effect on blood pressure.
A positive relationship of total serum calcium to
blood pressure has been reported previously,38X0 but
the subject is complicated by reports of an inverse
blood pressure relationship to both serum ionized calcium4`42 and calcium intake.43 The ionized calcium
level among hypertensive patients is related to renin
plasma levels,42 but the applicability of this finding to
normotensive persons is not clear. The progressive
relationship of total serum calcium to blood pressure in
our data raises the possibility that calcium either is a
partial determinant of systolic pressure or is related to
such a determinant. The inverse relationship of serum
potassium to blood pressure might be a manifestation
of mineralocorticoid activity. Mineralocorticoids
could produce both an increase in blood pressure and a
lowering of serum potassium. Serum potassium may
also be related to oral potassium intake, which, in turn,
has been described as inversely related to blood pressure.4`6 Finally, serum potassium might be a factor
directly or indirectly determining blood pressure.
A number of the traits studied (including calcium
and potassium levels and use of coffee or tea) showed a
more consistent relationship to systolic pressure than
to diastolic pressure. Traditionally, this type of disparity has been thought to represent an effect more on
cardiac stroke volume than on peripheral resistance.
The concept has also been widely accepted that systol-
CIRCULATION
PATHOPHYSIOLOGY AND NATURAL HISTORY-HYPERTENSION
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
ic pressure is affected earlier in the course of hypertension, whereas both systolic and diastolic pressure are
affected later. But the mechanisms are probably much
more complex than these statements imply.
This type of study cannot prove causality or establish a mechanism for the confirmed, empiric alcohol-blood pressure association. However, we believe
that these features of our data contribute to the likelihood that the relationship is a direct causal one: (1) the
continuous dose-response relationship in white men
whose reported drinking behavior was fairly constant
from the less than one through nine or more drinks daily
subgroups, (2) the independence from a great many
potential confounders, (3) the limited importance of
type of alcoholic beverage, (4) the apparent complete
regression of the alcohol-blood pressure association
upon cessation of drinking, and (5) the lower blood
pressure in heavier drinkers who have abstained recently. We previously estimated' that about 5% of
hypertension in our prepaid health plan population
may be associated with alcohol. Others have reported
higher estimates (up to 24%),23 but even our small
estimated proportion may equal that of all other secondary causes of hypertension combined. Because alcohol-associated hypertension might be controlled by
omitting the offending agent instead of adding antihypertensive drugs, the public health implications are
substantial.
We thank Cynthia Landy and Rita J. Coston for data collection, Della J. Mundy, M.L.S., for editorial assistance, and Lyn
D. Wender for technical assistance.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
References
1. Friedman GD, Klatsky AL, Siegelaub AB: Alcohol intake and
hypertension. Ann Intern Med 98: 846, 1983
2. Mathews JD: Is alcohol use a preventable cause of hypertension?
Med J Aust 2: 53, 1983
3. Ramsay LE: Alcohol use and hypertension. Pract Cardiol 5: 27,
1979
4. Kaysen G, Noth RH: The effects of alcohol on blood pressure and
electrolytes. Med Clin North Am 68: 221, 1984
5. Wallace RB, Lynch CF, Pomrehn PR, Criqui MH, Heiss G: Alcohol and hypertension: epidemiologic and experimental considerations. The Lipid Research Clinics. Circulation 64(suppl III): III41, 1981
6. Luscher T, Siegenthaler W, Vetter W: Alcohol and hypertension.
Schweiz Rundsch Med Prax 73: 591, 1984
7. Beilin LJ, Puddey IB: Alcohol and essential hypertension. Alcohol
Alcohol 19: 191, 1984 (editorial)
8. Klatsky AL: Blood pressure and alcohol consumption. In Bulpitt
CJ, editor: Handbook of hypertension. Vol. 6: Epidemiology of
hypertension. Amsterdam, 1985, Elsevier, pp 159-174
9. Potter JF, Beevers DG: Pressor effect of alcohol in hypertension.
Lancet 1: 119, 1984
10. Puddey IB, Beilin LJ: The clinical importance of alcohol consumption in patients with hypertension. Pract Cardiol 10: 59, 1984
11. Collen MF, Davis LF: The multitest laboratory in health care. J
Occup Med 11: 355, 1969
12. Klatsky AL, Siegelaub AB, Landy C, Friedman GD: Racial patterns of alcoholic beverage use. Alcoholism (NY) 7: 372, 1983
13. Klatsky AL, Friedman GD, Siegelaub AB, Gdrard MJ: Alcohol
Vol. 73, No. 4,
14.
April
1986
29.
30.
31.
32.
33.
34.
35.
36.
consumption and blood pressure: Kaiser-Permanente multiphasic
health examination data. N Engl J Med 296: 1194, 1977
Fortmann SP, Haskell WL, Vranizan K, Brown BW, FarquharJW:
The association of blood pressure and dietary alcohol: differences
by age, sex, and estrogen use. Am J Epidemiol 118: 497, 1983
Criqui MH, Wallace RB, Mishkel M, Barrett-Connor E, Heiss G:
Alcohol consumption and blood pressure: the Lipid Research Clinics Prevalence Study. Hypertension 3: 557, 1981
Clark VA, Chapman JM, Coulson AH: Effects of various factors
on systolic and diastolic blood pressure in the Los Angeles Heart
Study. J Chronic Dis 20: 571, 1967
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
Kannel WB, Sorlie P: Hypertension in Framingham. In Paul 0,
editor: Epidemiology and control of hypertension. New York,
1975, Stratton Intercontinental, pp 553-592
Wallace RB, Barrett-Connor E, Criqui M, Wahl P, Hoover J,
Hunninghake D, Heiss G: Alteration in blood pressures associated
with combined alcohol and oral contraceptive use: the Lipid Research Clinics Prevalence Study. J Chronic Dis 35: 251. 1982
Harburg E, Ozgoren F, Hawthorne VM, Schork MA: Community
norms of alcohol usage and blood pressure: Tecumseh, Michigan.
Am J Public Health 70: 813, 1980
Arkwright PD, Beilin LJ, Rouse I, Armstrong BK, Vandongen R:
Effects of alcohol use and other aspects of lifestyle on blood pressure levels and prevalence of hypertension in a working population.
Circulation 66: 60, 1982
Arkwright PD, Beilin LJ, Vandongen R, Armstrong BK, Masarei
JRL: Alcohol and hypertension. Aust NZ J Med 14: 463, 1984
Cooke KM, Frost GW, Thornell IR, Stokes GS: Alcohol consumption and blood pressure: survey of the relationship at a healthscreening clinic. Med J Aust 1: 65, 1982
Milon H, Froment A, Gaspard P, Guidollet J, Ripoll JP: Alcohol
consumption and blood pressure in a French epidemiological study.
Eur Heart J 3(suppl C): 59, 1982
Mitchell PI, Morgan MJ, Boadle DJ. Batt JE, Marstrand JL,
McNeil HP, Middleton C, Rayner K, Lickiss JN: Role of alcohol in
the aetiology of hypertension. Med J Aust 2: 198, 1980
Ueshima H, Shimamoto T, lida M, Konishi M, Tanigaki M, Doi
M, Tsujioka K, Nagano E, Tsuda C, Ozawa H, Kojima S, Komachi Y: Alcohol intake and hypertension among urban and rural
Japanese populations. J Chronic Dis 37: 585, 1984
D'Alonzo CA, Pell S: Cardiovascular disease among problem
drinkers. J Occup Med 10: 344, 1968
Gordon T, Kannel WB: Drinking and its relation to smoking, BP,
blood lipids, and uric acid: the Framingham Study. Arch Intern
Med 143: 1366, 1983
MacMahon SW, Blacket RB, Macdonald GJ. Hall W: Obesity,
alcohol consumption and blood pressure in Australian men and
women: The National Heart Foundation of Australia Risk Factor
Prevalence Study. J Hypertension 2: 85, 1984
Cairns V, Keil U, Kleinbaum D, Doering A, Steiber J: Alcohol
consumption as a risk factor for high blood pressure: Munich Blood
Pressure Study. Hypertension 6: 124, 1984
Salonen JT, Tuomilehto J, Tanskanen A: Relation of blood pressure to reported intake of salt, saturated fats, and alcohol in healthy
middle-aged population. J Epidemiol Community Health 37: 32,
1983
Barboriak PN, Anderson AJ, Hoffman RG, Barboriak JJ: Blood
pressure and alcohol intake in heart patients. Alcoholism (NY) 6:
234. 1982
Kesteloot H, Van Houte 0: An epidemiologic survey of arterial
blood pressure in a large male population group. Am J Epidemiol
99: 14, 1974
Tanskanen A, Tuomilehto J, Salonen JT, Nissinen A, Puska P:
Coffee, drinking and blood pressure a link or not'? Presented at
International Symposium on Diet and Primary Prevention of Hypertension, Kuopio, Finland, June 1984
Lang T, Degoulet P, Aime F, Fouriaud C, Jacquinet-Salord M-C,
Laprugne J, Main J, Oeconomos J, Phalente J, Prades A: Relation
between coffee drinking and blood pressure: analysis of 6,321
subjects in the Paris region. Am J Cardiol 52: 1238, 1983
Robertson D, Frolich JC, Carr RK, Watson JT, Hollifield JW,
Shand DG, Oates JA: Effects of caffeine on plasma renin activity,
635
KLATSKY et al.
catecholamines and blood pressure. N Engi J Med 298: 181. 1978
37. Whitsett TL, Manion CV. Christensen HD: Cardiovascular effects
of coffee and caffeine. Am J Cardiol 53: 918. 1984
38. Bulpitt CJ, Hodes C, Everitt MG: The relationship between blood
pressure and biochemical risk factors in a general population. Br J
Prev Soc Med 30: 158, 1976
39. Kesteloot H, Geboers J: Calcium and blood pressure. Lancet 1:
813, 1982
40. Harlan WR, Hull AL, Schmouder RL. Landis JR, Thompson FE,
Larkin FA: Blood pressure and nutrition in adults: the National
Health and Nutrition Examination Survey. Am J Epidemiol 120:
17, 1984
41. McCarron DA: Low serum concentrations of ionized calcium in
patients with hypertension. N EngI J Med 307: 226, 1982
42. Resnick LM, Laragh JH, Sealey JE, Alderman MH: Divalent cations in essential hypertension: relations between serum ionized
calcium, magnesium, and plasma renin activity. N Engl J Med 309:
888, 1983
43. Belizan JM, Villar J, Pineda 0, Gonzalez AE, Sainz E, Garrera G,
Sibrian R: Reduction of blood pressure with calcium supplementation in young adults. JAMA 249: 1161, 1983
44. Langford HG: Dietary potassium and hypertension: epidemiologic
data. Ann Intern Med 98: 770, 1983
45. Gruchow HW, Sobocinski KA, Barboriak JJ: Alcohol, nutrient
intake, and hypertension in US adults. JAMA 253: 1567, 1985
46. Tannen RL: Effects of potassium on blood pressure control. Ann
Intern Med 98: 773, 1983
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
636
CIRCULATION
The relationships between alcoholic beverage use and other traits to blood pressure: a
new Kaiser Permanente study.
A L Klatsky, G D Friedman and M A Armstrong
Downloaded from http://circ.ahajournals.org/ by guest on July 31, 2017
Circulation. 1986;73:628-636
doi: 10.1161/01.CIR.73.4.628
Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1986 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/73/4/628
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/