Document 6544

.Original Investigations
Coffee Consumption and Mortality
<:'•
Total Mortality, Stroke Mortality, .
and Coronary Heart Disease Mortality
Siegfried Heyden, MD;irferman A. TyroleJV&D; Gerardo Heiss, MD; Curtis G. Haroes, MO; Alan Bartel, MD
• Total mortality showed no association with coffM uaaga In
tha four raca-aax yrotip* of Evans County, Georgia. Deaths ot
coronary haart cUaaaaa (CHD) in white man and woman and
bfack man ehowed no statistically significant difference batwaan
high and low coffee consumer*. In an araa thai haa baan
daalgnatad aa tha "Stroke Bait," nalthar CHD nor earabrovaacular daath rates aaam related to coffee-drinking habits. However,
to rafuta or donflrm tha allegations of a detrimental Influanca of
high coffaa Intake, larger aamplaa ara needed. Nevertheless, our
finding that mortality from ail cauaaa ia not incraaaad In tha high
coffee-consummg group maana that a finding of incraaaad CHO
mortality with high coffaa consumption would hava to ba
compensated by a protactiva lowar rata for othar cauaaa of
daath. '
medical literature during the past 20 years. However, the
pertinent literature is listed for the interested reader. The
purpose of this communication is to present evidence from
a 4'£-year prospective mortality study—based on the
S— also editorial p 1 4 7 1 .
surveillance of a total community—that neither total
mortality nor stroke mortality or coronary heart disease
(CHD) mortality differs between high and low coffee
consumers.
METHOD
A prevalence survey in Evans County, Georgia, was conducted
during 1960 to 19-52,1* and the study population was reexamined
between 1967 end VJS9. During this second study of 2,530 adults
(60% white, 40% black), the two examining- physicians {A3, and
3.H.) asked each person a few standardized questions concerning
What twist is it in man's devious make-up that make* him
round on the seemingly more wholesome and pleasurable aspects of coffee consumption. Persons who gave a history of drinking five
cups of coffee or more per day regularly, i«, during summer and
his environment and suspect them of being causes of his misfortunes? Whatever it is, stimulants of all kinds (and especially coffee winter, were placed in the high coffee-consuming group. All others
were classified as low consumers or as nonconsumers. The cohort
and caffeine) maintain a position high on the list of suspicion
was followed up annually for 4fc years with questionnaires.
despite a continuing lack of real evidence of any hazard to
Between July 31, 1969, and Jan 1. 1974, a total of 339 deaths
health.
occurred. Of these 339 deaths, 130 (38%) were confidently attributed to cardiovascular and cerebrovascular causes. They were
BRITISH MEDICAL JOURNAL
confirmed
by autopsy reports, hospital records, and reviews by a
(May i t 1976; page 1031)
neurologist and a cardiologist of all available information, including interviews of family and/or co-workers in cases of sudden
death.
"Possible" cardiovascular and cerebrovascular deaths were
* p h e multitude of ailments of civilization and risk factors
classified
with all other causes of death, eg, accidents, postopera•^ of chronic degenerative diseases that have been
tive
complications,
pneumonia, or cancer. Age adjustment within
alleged to be related to the regular drinking of coffee
each race-sex group was used to avoid possible confounding in the
(Table 1) leaves the practicing physician confused and
analysis stemming from the well-known observation that older
skeptical. This is not the place or the xime to discuss ail
people tend to drink less coffee than middle-aged people.
accusations or suspicions that have been found in the
The choice of mortality as the endpoint for this study was made
because our cross-sectional study from 1967 to 1969, which
Accepted for publication Jan 6, 1978.
analyzed coffee arinking in relation to CHD and stroke, bad not
From ihe Department of Community and Family Medicine, Duke Univer- disclosed any difference in high or low coffee-consuming groups.'1
sity Medical Center, Durham. KC (Drs Heyden and Bartei). the Department At that time, we pointed out the disadvantage of the study, which
of Epidemiology. School of Public Health, University of North Carolina,
is the one implicit in a prevalence survey: one automatically limits
Chapel HiH fDrs Tyroler and Heiss). and the Health Department, Evans
the study to survivors of the diseases under consideration. Two
Count;; Heart Study. Claxton. Ga (Dr Hamesj.
manifestations
of ischemic heart disease, death of myocardial
Reprint requests to Department of Community and Family Medicine.
infarction, and sudden death, as well as stroke deaths, thus escape
Duke University Medical Center. Durham, NC 2i*10 (Dr Heyden).
(Arch Intern Mad 138:1472-1475, 1978)
1472
Arch Intern Med-Voi 138, Oct 1978
Coffee and Mortality—Heyden et al
* _»--E-*^ '* iw*ffj ** *•• - *• _ "«** »c"-"»««* ~ « a-**.;'-
Tl 09140030
&-croft-sectional study. If it were assumed that the heavy coffee
drinkers all had died of ischemic heart disease or stroke prior to
our 1967-19© survey, we would indeed have missed these important manifestations of CHD and cerebrovascular disease (CVD).
RESULTS
Mortality from all causes in this total community was
not significantly different for white men, white women, or
black men, regardless of high coffee consumption (> 5
cups/day) or little or no coffee consumption. There was a
suggestion of lower total mortality among black women
who drank five cups or more. However, despite statistical
significance (P < .02), this result must be considered
tentative because of the small number observed.
Therefore, total mortality was not influenced by the
coffee-drinking habit. If anything, the group of high
coffee consumers appeared slightly favored by a lower
Table 1 .—Most Frequent Associations
Between Coffee Consumption and Diseases*
Type*
Abnormality
Metabolic
Gastrointestinal
Disease
Hyperlipoproteinemia
Diabetes meilltua
Hyperuricemia/gout
Cirrhosis ot liver
Peptic ulcer
Hyperacidity and
esophageal gastric acid reflux
Drop in lower
esophageal
sphincter pressure
Gastric cancer
Cardiovascular and
cerebrovascular
Urogenital
1977 Assessment
Under debate"
Disproved1
Disproved4
Disproved*-1
Disproved*
No difference between coffee and
decaffeinated coffee'
Opposite findings'**
Anticarcinogenlc
effect of
caffeine'*
No association"
Disproved"
Colonic cancer
Hypertension
Myocardial
infarct! onf
Stroke
Premature, ventricular contractions
Renal cancer
Bladder cancer
Prostatic cancer
Disproved'-'5-'4-*
Disproved*
Under debate'*- f
Disproved"
Disproved'*-"
Disproved'*
•Twenty-year review of the medical literature.
tTabie S lists the presently available seven prospective and five retrospective studies of this particular subject
JPresent study.
5According to H. Blackburn, MD (oral communication. March 13,
1976).
mortality in comparison -with low consumers or with
nonconsiimers (Table 2);
Mortality from CVD showed differences between the
two coffee-drinking groups. While stroke deaths were
found more often in white and black men who had reported
low lifetime coffee intake or no coffee intake, white and
black women who were heavy consumers of coffee had a
somewhat higher age-adjusted stroke mortality than their
counterparts who were low consumers or nonconsumers.
Attention once more, however, is called to the small sample
sizes {Table 3).
Mortality from CHD did not show any consistent differences between the heavy coffee drinkers and the low coffee
drinkers or the nondrinking persons. White men who
drank five cups of coffee or more had a slightly higher
CHD mortality. White women in the five-cup-a-day group
had a marginally lower CHD mortality than white women
who did not drink coffee. The CHD mortality for blacks,
though higher for those with low coffee consumption,
cannot be seriously considered because of the small number
of blacks who were high coffee consumers (Table 4). Since
there were no major differences in CHD rates among the
coffee-consuming groups, no adjustment for cigarette
smoking was necessary.
Lack of systematic differences in vascular mortality
among the four race-sex groups led us to the conclusion
that there was no evidence of an association between
coffee-drinking habits and mortality either from all causes
or from specific vascular diseases. If one were to assume
there was a higher CHD death rate among heavy coffee
drinkers one would have to explain a "protective" effect of
coffee for other causes of death, since all causes of death
(total mortality) were equally distributed between high
coffee consumers and low consumers or nonconsumers
(Table 2).
Our strict criteria for the diagnosis of death due to CHD
or stroke (see "Method") may have favored the category
"other causes, of death" somewhat Thus, the theoretical
possibility exists that we may have misplaced a CHD death
into the "mortality from other causes" category. In this
eventuality, the chances are that this rare instance would
have occurred among the lower socioeconomic group with
less documentation by either ECG, hospital records, or
autopsy reports.
COMMENT
A review of seven prospective and five retrospective
epidemiological studies on the association of coffee
consumption and CHD demonstrated a lack of association.
None of the prospective jind only two retrospective
inves54 2
tigations, which were reported by the same group,
- " and
one "total adult population" study from Finland2" showed
an appreciable effect of heavy coffee consumption on the
prevalence of myocardial infarction and CHD death. The
Table 2. -Risk of Mortality From AM Causes Adjusted for Age and Smoking Habits*
Race
and
Sex
White
Men
Women
Black
Men
Women
Coffee Consumption < 5 Cue*/Day
Coffee Consumption > 5 Cups/Day
PAA/Casee
Adjusted Mortality, %
SMR
PAR/Cases
Adjusted Mortality, %
SMR
556/72
639/53
12.9
8.3
1.0
1.0
9</lO
134/9
11.3
8.9
0.9
1.1
323/53
427/52
16.4
12.2
1.0
1.0
18/3
32/1
19.8
3.6
1.2
0.3
indirect method. Abbreviations are as follows; PAR, population at risk; SMR, standardizes mortality ratio.
Arch Intern Med—Vol 138. Oct 1978
Coffee and Mortality—Heyden et al
1473
TI09140031
Framingham study' and the Evans County study
presented herein have, in addition, revealed no association
between heavy coffee drinking and total mortality. The
Framingham study had provided convincing evidence that
cholesterol level was not related to coffee drinking and had
shown also that hypertension was unrelated to the use of
coffee; pharmacologists proved that there was no correla-
tion between uric acid metabolism and coffee intake, and
endocrinologists demonstrated diabetes to be unrelated to
caffeine. I t was therefore difficult to imagine how, ie,
through which biological mechanism, one could have
explained an independent association if it had been shown
to exist. While the commonly accepted risk factors were
unrelated to coffee use, it is well documented that cigarette
T a b l e 3.—Risk of M o r t a l i t y F r o m S t r o k e A d j u s t e d for A g e *
Coffee Consumption > S C u p s / O a y
Coffee C o n s u m p t i o n < 5 C u p s / O a y
Race
and
Sex
White
Men
Women
Black
Men
Women
Adjusted
Mortality. %
Adjusted
Mortality, %
SMR
608/13
2.1
:.o
673/4
0.6
1.0
334/9
2.7
1.0
18/0
0.0
0.0
1.6
1.0
32/1
8.4
5.4
PAR/Cases
446/7
PAR/Cases
102/0
136/4
SMR
0.0
0.0
3.9
8.5
'Indirect method. Abbreviations are as follows: PAR, population at risk; SMR. standardized mortality ratio.
T a b l e 4. —Risk of Mortality F r o m C o r o n a r y H e a r t D i s e a s e A d j u s t e d for A g e *
Race
and
Sex
Coffee Consumption > 5 C u p s / O a y
Coffee Consumption < 5 C u p s / D a y
PAR/Cases
Adjusted Mortality, %
SMR
575/13
2.3
1.0
96/4
4.5
2.0
646/10
!.5
1.0
138/1
1.0
0.6
337/4
1.2
1.0
19/0
0.0
0.0
441/4
0.9
1.0
34/0
0.0
0.0
White
Men
Women
Black
Men
Women
PAR/Cases
Adjusted Mortality, %
SMR
"Indirect method. Abbreviations are as follows: PAR. population at risk; SMR. standardized mortality ratio.
T a D l e 5. — C o f f e e C o n s u m p t i o n a n d C o r o n a r y H e a r t D i s e a s e
Source
Year
Population Study
Endpoini Disease
Results*
Prospective Studies
Paul"
1968
Western Electric study
Wilhemsen et a l 1 1 "
1973
1977
Klatsky et al7*
1973
men
All manifestations
Negative after c o n trolling for cigarette smoking
Gothenburg, S w e d e n . Tien
Myocardial infarction
Negative after controlling for cigarette smoking
Kaiser-Permanente. men and
women
Myocardial infarction
Negative after c o n trolling for cigarette smoking
Kiatsky et a l "
1974
Kaiser-Permanente. men
Sudden death
Negative
Oawber et at'
1974
Framingham. Mass. men and
women
All manifestations
Negative after controlling for cigarette smoking
Present study
1978
Evans County. Georgia, n e n 3nd
women blacks and whites
General mortality daath of
coronary heart disease
and stroke
Negative
Yano et at"
1977
Honolulu. Jaoanese men
All manifestations
Negative after controlling for cigarette smoking
Myocardial infarction
Positive
Retrospective Studies
Boston Collaborative
Drug Surveillance
Program**
1972
Hospital patients, men and w o m e n
Jick et a l "
1973
Hospital oatrents. men and w o m e n
Myocardial infarction
Positive
Hrubec*
1973
National Research Council, male
twins
Angina pectoris
Negat've after c o n trolling for cigarette smoking
Hennekens et a l "
1976
Florida, men
Death of coronary heart
disease
Negattve
Hemmtnski and
Pesonen*
• 97-7
Total adult population of F-niand
Death of coronary heart
cisease
Positive
-Most frequently used cutort point was five or su cups of coffee.
1474
A r c h intern Med—Vol ':38. Oct - 9 7 8
Coffee and Monattty—Heyden et al
smoking is highly significantly related to coffee drinking.1-"*" Interestingly, only the Boston. Collaborative Drug
Surveillance Program" did not find this strong correlationThere was an unusually small difference in proportions of
cigarette smokers between the myocardial infarction
group and the control group. The authors suggested that
the controls may have had a high proportion of smokers
because many have had tobacco-associated diseases. For
this reason alone, the results from community studies, as
opposed to hospital studies, are to be preferred.
The 1967 to 1969 cross-sectional study in Evans County"
did not demonstrate an increase in any of the common risk
factors predisposing to ischemic heart disease among
heavy coffee drinkers, with the exception of cigarette
smoking. Cigarette smoking was strongly correlated with
heavy coffee consumption. If there were a high CHD
incidence among heavy coffee drinkers compared with
nonconsumers or low consumers, it could be explained on
the basis of the strong correlation between the two habits,
as consistently shown by three earlier studies: the Chicago
electrical workers study," the Gothenburg study, 1 "' and
the National Research Council study." The Framingham
study, as well as the Chicago Western Electric study, the
Kaiser-Permanente study, and the Florida community
(death certificate) study,-7 refuted an association between
heavy coffee consumption per se and the - incidence of
myocardial infarction or death of ischemic heart disease.
To "prove" the absence of an association is one of the most
challenging tasks in nonexperimental research and is one
that places costly demands on study design and sample
size. The 4^-year mortality follow-up study in Evans
County showed no differences in CHD deaths among heavy
consumers and nondrinkers or low-consumption coffee
drinkers, and adds findings of a prospective nature to the
reported observations that consistently point to a lack of
association between coffee consumption and cardiovascular mortality.
-.
i:
I
i .
I'
. f-
•:
r.
References
1. Dawber TR, Kannel WB, Gordon T: Coffee and cardiovascular disease:
Observations from the Framingham study- .V Engl J Med 291:871-874,
1974.
2. Rhoads CG, Kagan A. Yano K: Association between dietary factors
and plasma lipoproteins, abstracted. Circulation, suppl 2, 1976, p 53,
3. Studlar M, Pichler 0: Metabolic effects of coffee and caffeine in normal
subjects, diabetics and patients with iiver affections. Z Emaehrungsunss
15:80-9!. 197G.
4. Goodman LS. Oilman A: The Pharmacological Basts of Therapeutics,
ed 4. New York, The Macmillan Co Publishers, 1970.
5. Ortmans H, Eisenberg K: Influence of caffeine on serum lipid concentrations of patients with liver diseases. Z Ernaekrungsvnss 13:43-49,
1974.
6. Friedman GD, Siegelaub AB, Seltzer CC: Cigarettes, alcohol, coffee
and peptic ulcer. A' Engl J Med 290:46M73. 1974.
7. Cohen 3, Booth GH J n Gastric acid secretion and tower-esophageulsphincter pressure in response to coffee and caffeine. *V Engl J Med 293:897,
1975.
8. Coffee drinking and peptic ulcer disease, editorial. Nutr Rev 34:167,
1976.
9. Pope CE II: Cuppa coffee for the cardia? or Sanka soothes the
sphincter? S Engl J Med 293:931-932, 1975.
10. Caffeine, coffee, and cancer, editorial. Br Med J, May 1, 1976,
p 1031-1032.
11. Higginson J: Etiological factors in gastrointestinal cancer in man.
J Natl Cancer Inst 37:527-532. 1966.
12. Dawber TR, Kannel WB, Kagan A, et al: Environmental factors in
hypertension, in Stamler J, Stamler R, Pullman TN* (edsj: Epidemiology of
Hypertf7i.-'ion. New York. Grune & Stratton Inc. 1967.
13. Wilhelmsen L, Tibblin G. Elmfeldt D, et al: Coffee consumption and
coronarv heart disease in middle-aged Swedish men. Acta Med Scand
201:5-17-552. 1977.
14. Yano K. Rhoads CG, Kagan A: Coffee, alcohol and risk of coronary
heart disease among Japanese men living in Hawaii. -V Engl J Med
297:405-409, 1977.
15. Heyden S. Bartel A, Cassel JC f et al: Coffee consumption, vascular
diseases and risk factors in the Evans County, Georgia study. Z Entaehrungsvriss. suppl 14, 1972, pp 1-10.
16. Armstrong B, Garrod A, Doll R: A retrospective study of renal cancer
with special reference to coffee and animal protein consumption. Br
J Cancer :J3:127-136, 1976.
17. Bross IDJ, Tidings J: Another look at coffee drinking and cancer of
the urinary bladder. Prcv Med 2:445*451, 1973.
18. Morgan RW, Jain MG: Bladder cancer Smoking, beverages and
artificial sweeteners. Can Med Assoc J 111:1067-1071, 1974.
19. Heyden S: Coffee consumption and carcinogenesis: No etiological
relation. Z Ernachrungswiss, suppl 14. 1972, pp 11-18.
20. Paul 0: Stimulants and coronaries. Postgrad Med 44:196-199, 1968.
21. Wilhelmsen h. Wedel H, Tibblin G: Multivariate analysis of risk
factors for coronary heart disease. Circulation 48:950-958. 1973.
22. Klatsky AL, Friedman GD, Siegelaub AB: Coffee drinking prior to
myocardial infarction: Results from Kaiser-Permanente epidemiological
study of myocardial infarction. JAMA 226:540-543, 1973.
23. Klatsky AL, Friedman EG, Siegelaub AB: Habits and sudden cardiac
death, abstracted. Circulation, suppl 3, 1974, pp 99-100.
24. Coffee drinking and acute myocardial infarction. Report from the
Boston Collaborative Drug Surveillance Program. Lancet 2:1278-1281,
1972.
25. Jick H. Miettinen 0, Neff RK, et al: Coffee and myocardial infarction.
N Engl J Med 289:63-67,1973.
26. Hrubec Z: Coffee drinking and ischaemic heart-disease. Lancet 1:548,
1973.
27. Hennekens CH, Drolette ME. Jesse MJ. et al: Coffee drinking and
death due to coronary heart disease. .V Engl J Med 294:633-636. 1976.
2S. Hemminski E, Pesonen T: Regional coffee consumption and mortality
from ischemic heart disease in Finland. Acta Med Scand 201:127-130,
1977.
n
fa
Hypertension Update—One-Day Course.—The Johns Hopkins Medical Institutions
Department of Medicine. Baltimore, will offer a one-day course, "Hypertension Update,"
Oct 26,1978. The registration fee is $25. The course is approved for 7Vi hours of category
1 credit. For information, contact Course Coordinator, Office of Continuing Education,
720 Rutland Ave, Baltimore. MD 21205 (301-955-5880).
Arcn Intern M e d - V o l 138. Oct 1978
Coffee a n d Mortality—Heyden et af
1475
TI09140033