- Email: [email protected]
Coffee and serum lipids: Findings from the Olivetti heart study
Coffee and Serum Lipids: Findings from the Olivetti Heart Study FABRIZIO JOSSA, MD, VITTORIO KROGH, MD, MS, EDUARDO FARINARO, MD, SALVATORE PANICO, MD, MS, DANTE GIUMETTI, MD, ROCCO GALASSO, MD, EGIDIO CELENTANO, MD, MARIO MANCINI, MD, AND MAURIZIO TREVISAN, MD,
MS
The relationship between coffee consumption and blood lipids was analyzed in a sample of 900 male workers of southern Italy participating in the Olivetti Heart Study. In the univariate analysis, coffee drinkers (n = 856) had higher values for body mass index (P 5 0.05) and number of cigarettes smoked per day (P I O.OOl),and lower levels of serum high-density-lipoprotein cholesterol @’5 0.0.5), compared to noncoffee drinkers (n = 44). In addition, coffee consumption (cups/d) was positively related to serum triglyceride levels (r = 0.105, P I 0.01) and cigarette smoking (r = 0.491, P 5 0.01), and was inversely related to age (r = -0.122, P I 0.01). After multivariate adjustment, coffee consumption remained significantly related to age, cigarette smoking, and body mass index (data not shown). After stratification for smoking status, a significant positive linear trend between coffee consumption and serum total cholesterol was observed only in smokers. No significant trend was observed for serum triglycerides and high-density-lipoprotein cholesterol with coffee intake according to smoking status. This finding suggests that the relationship between coffee consumption and serum total cholesterol may change with the smoking status. Ann Epidemiol 1993; 3:250-255. KEY WORDS:
Coffee
consumption,
serum total cholesterol, serum triglycerides, cigarette
smoking.
This study analyzed the relationship between coffee con-
INTRODUCTION Epidemiologic
studies have suggested a relation
coffee consumption cently,
investigators
relationship
and coronary
between
heart disease (1, 2). Re-
have focused their attention
between coffee and a number
on the
of risk factors
sumption and blood lipids in a sample of healthy workers of southern Italy consuming “espresso” coffee, which differs from boiled, filtered, or percolated coffee in caffeine content and in brewing methods.
for coronary heart disease, (e.g., serum lipids (3) and blood pressure (4, 5)), in order to define a possible mechanism this association.
The relationship
tion and serum total cholesterol cross-sectional
for
between coffee consumphas been evaluated in 22
studies carried out in eight different coun-
tries (see the article by Thelle and coworkers (3) for review); the findings were somewhat
conflicting.
It has been sug-
gested that this association may be related to the clustering of atherogenic
behaviors
in coffee drinkers
noncoffee drinkers (6- 1 I), components
compared
to
of coffee other than
caffeine (since tea or cola drinks do not have the same effect) (9, 12, 13), or particular
methods
of brewing
(14-17).
MATERIALS
AND
METHODS
The Olivetti Heart Study is a longitudinal risk factors for atherosclerosis,
Factory in Naples, Italy (18). The baseline examination
on was
conducted in 1975 to 1976; the 5-year and 12-year follow-up examinations
were carried out in 1980 to 1981 and 1987 to
1988, respectively. At each examination, invited to participate, previous examinations.
all employees were
regardless of their participation
in
At the 1987 to 1988 examination
a total of 990 workers (942 males and 48 females) agreed to participate in the study, a participation
From the Department of Social and Preventive Medicine, State University of New York at Buffalo, Buffalo, NY (F. J., V. K., M. J.), and the Institute of Internal Medicine and Metabolic Diseases, University of Naples, Medical School, Naples, Italy (F. J., V. K., E. F., S. P., D. G., R. G., E. C., M. M., M. T.). Address reprint requests to: Fabrizio Jossa, MD, Department of Social and Preventive Medicine, State University of New York at Buffalo, 270 Farber Hall, Buffalo, NY 14214. Received January 28, 1992; revised August 26, 1992.
investigation
carried out at the Olivetti
rate of 80%. Because
of the small number, females were excluded from the present analysis. The data reported here were collected in the 1987 survey. The participants
were seen in the morning,
in the
medical facilities of the factory, after an overnight fast. The medical examination included (a) systolic and diastolic blood pressure measurements, electrocardiogram,
(b) a 12-lead standard resting
(c) anthropometric
measurements,
and
1047.2797/93/$06.00
AEP Vol. 3, No. 3 Mav 1993: 250-255
COFFEE
TABLE 1. Baseline characteristics
of 900 male workers in southern
Italy (mean f standard
Age (Y)
46.9
Body mass index serum
(kg/m2)
cholesterol
(mg/dL)
Low-density-lipoprotein Serum
cholesterol
triglycerid&
High-density-lipoprotein Alcohol
(mL/wk)
consumption
Cigarette
smoking
Prevalence
+ 7.5
(mg/dL)
46.9
? 7.4
47.7 f
9.8
26.4 2 3.1
25.3 f
2.7
148.0
+ 31.7
168.4 2 31.8
139.9 -r- 30.0
149.8 2 89.2
140.8 f
264.7
f
89.6 11.5
? 332.1
(cigarettes/d)
10.7 2 11.3
(96)
21
to gather
information
quality
balance
scale with participants
cilip) for HDL-C.
clothing.
Body mass index
shoeless (kg/m’)
and wearing
was calculated.
were performed protocol
by trained
described
Methods
indoor
by the
of the World
Manual Health
for Orga-
Coffee
consumption
was
the habitual
assessed
number
by
one
consumption
to evaluate
previous
variable
because
was ascertained
the amount
week and was also analyzed
was calculated
according
= 11% ethanol
volume, concerning
= 30% ethanol
the number
of cigarettes
described
cular ResearchSurvey Smoking
beer
was measured
tion of serum cholesterol
were both determined
2 1). High-density-lipoprotein acid
All biochemical
for Cardiovas-
Organization
variable
by enzymatic
- HDL-C
measurements
(19).
Serum
total
(TG) concen-
statistics matrix.
out to calculate cholesterol Multiple N-l
of the relationship
the adjusted
and
among
variables
standard
of covariance mean
to coffee
data were obtained
dummy
heart disease risk
(mean Analysis
according
comparisons
values
of
consumption
the different
cate-
using regression
(23). Multiple
linear
regression analysis was performed to evaluate linear between LDL-C and coffee consumption (cups/d).
trend
(20,
Table
1 summarizes
whole
sample
chloride-phosphomethods
separately. (22).
was calculated
the
and coffee
These
southern
baseline drinkers
for the drinkers
men were characterized
by a high weekly alcohol consumption and a high prevalence of both smokers (56%) and coffee drinkers (95%). The
noncoffee under
characteristics and noncoffee
Italian
coffee drinkers had significantly index and number of cigarettes
- (TG/5) performed
RESULTS
was ana-
procedures
(HDL-C)
enzymatic
were
after
(cigarettes/d).
variables.
triglycerides
and
descriptive
serum
with
was collected
and coronary
was carried
analysis
for
intake),
for determina-
Low-density-lipoprotein cholesterol (LDL-C) according to the following formula: CHOL
per day accord-
were available
for the assessment
and correlation
total
data
available
values for alcohol
on this parameter
methods
included
deviation)
this report
of missing
not being
consumption
coffee consumption
factors
analysis,
in any of
of the study.
gories for the adjusted
of magnesium
precipitation
Statistical between
for 939 of
data
The majority
(5 1 were missing
categories.
cholesterol
lyzed by a combination
by
in the present
on alcohol
and 5.9 (Pre-
available
of missing
lipid determinations
information
of variation for cholesterol,
for TG,
were
Because
considered
by a set of questions
by venipuncture
and serum
table:
1.65 (Precipat)
consumption
by volume. smoked
lipid levels and other
(CHOL)
of ethanol
conversion
of the World Health
was drawn
vari-
= 3.5% ethanol
by the Manual
was used as a continuous
Blood sample
dein the
as a continuous
to the following
smoking
ing to the protocol
consumed
consumption
by volume,
and liquor
Cigarette
by questions
of alcohol
Coefficients
1.65 (Precipat)
on 900 participants.
(n = 30). Data
The weekly absolute
tungstic
focuses
849 participants
able (mL/wk).
trations
the variables
the beginning
Alcohol
wine
on coffee
question
as a continuous
and
the 942 male workers.
of cups of coffee consumed
(cups/d).
procedure:
and
was due to serum
per day and was used in the analysis
signed
control
2.15 (Precinorm) Data
physicians
(19).
concerning
199.3 + 275.7
1.6 3.0 + 6.9d
1.84 (Precinorm)
nization
* 334.7
(SI) = mg/dL x 0.02586. (si) = mg/dL x 0.01129. vs. coffee drinkers. vs. coffee drinkers.
questionnaire
Survey
97.6
47.4 & 12.2’
58
strict
Cardiovascular
+ 36.7
11.1 + 11.4
were
to the
211.3
11.5
3.0 ?
56
(d) a self-administered
according
?
268.3
on life-style characteristics and personal and family history. In detail, weight and height were measured on a beam
All measurements
+ 41.3
43.4
3.0 + 1.6
’ Standard International Units ’ Standard International Units ’P < 0.05, noncoffee drinkers ’ P < 0.001, noncoffee drinkers
drinkers
(n = 44)
221.8
(cups/d)
of smokers
Noncoffee
+ 41.2
43.6
(n = 849)
drinkers
221.3
149.3 t
cholesterol
consumption
Coffee
(mg/dL)”
(mg/dL)
the Olivetti Heart Study
(n = 856)
26.3 ? 3.1
251
Jossa et al. SERUM LIPIDS
AND
deviation):
Coffee
All (N = 900)
Variables
Total
CONSUMPTION
drinkers,
in the coffee drinkers
while
higher levels of body mass smoked per day than did
HDL-C
compared
was significantly
to the noncoffee
lower drinkers.
252
Jossaet al. COFFEE CONSUMFTON
AEP Vol. 3, No. 3 May 1993: 250-255
AND SERUM LIF’IDS
TABLE 2. Correlation matrix: coffee consumption, lioids: the Olivetti Heart StudP
body mass index, age, cigarette smoking, alcohol intake, and serum
Body coffee Coffee consumption
Age
mass index
Smoking
0.12lb -0.050 0.043 0.12Ob 0.052 0.011 0.143b
-O.lOlb -0.021 0.142b 0.239b -0.162b 0.092‘
0.047 0.019 0.107b -0.226b 0.048
Alcohol (n = 859)
Cholesterol
High-densitylipoprotein cholesterol
Triglycerides
(cups/d)
Age (Y) Body mass index (kg/m’) Cigarette smoking (cigarettes/d) Alcohol intake (mL/wk) (n = 849) Serum total cholesterol (mg/dL) Serum triglyceridesb (mg/dL) High-density-lipoprotein cholesterol (mg/dL) Low-density-lipoprotein cholesterol (mg/dL)
-0.122b 0.031 0.491b 0.032 0.060 0.105b -0.152b 0.073
0.024 - 0.020 0.141b
0.007
0.40FJb 0.063 0.910b
-0.344b 0.158”
-0.066
”N = 900 unless otherwise indicated. b P I 0.01. r P _c 0.05.
Table 2 reports the correlation matrix for coffee consumption (cups/d) and other coronary heart disease risk factors. Coffee intake (cups/d) was significantly and positively related with number of cigarettes smoked per day and TG level, while it was inversely associated with age and HDL-C level. Serum total cholesterol and TG concentrations were both positively related to body mass index, while TG correlated positively with cigarette smoking and negatively with HDL-C; HDL-C was inversely related to cigarette smoking and body mass index and positively related with alcohol intake. In addition, LDL-C was positively related with age, body mass index, cholesterol, and TG. A significant positive interaction was detected between coffee use (yes/no) and smoking (yes/no) with regard to total serum cholesterol but not TG or HDL-C (data not shown). Therefore, the association between total serum cholesterol and coffee consumption was analyzed separately in smokers and nonsmokers. Table 3 reports the outcome of the multiple regression analysis for LDL-C after stratification for smoking status: Coffee consumption was an independent positive correlate of LDL-C only in smokers. Age was positively and significantly related to LDL-C both in smokers and nonsmokers. As indicated by the regression coefficient in Table 3, one additional cup of coffee was associtaed with an increase in LDL-C levels of 2.01 mg/dL in smokers and of 1.17 mg/ dL in nonsmokers. In the sample as a whole, in a multiple regression analysis including age, body mass index, and cigarette smoking, the relationship between coffee and either serum TG or HDL-C failed to reach statistical significance (data not shown). Tables 4,5, and 6 show the adjusted (age and body mass index) mean values for cholesterol, HDLC, and TG in the different categories of coffee consumption for the whole sample and in smokers and nonsmokers separately. The analysis of covariance indicated an overall significant difference for the average LDL-C level in the various catego-
ries of coffee intake (cups/d) in only the smokers. After participants who smoke but do not drink coffee (n = 9) were excluded, the observed difference was still significant (data not shown). No significant association was detected for coffee and LDL-C in nonsmokers and in the sample as a whole. A decrease in the serum concentration of HDL-C (see Table 5) and an increase in TG (see Table 6) were detectable across coffee categories in smokers, and in the group as a whole, even though in both cases the results of analysis of covariance did not indicate statistical significance. The inclusion of alcohol consumption as a covariate in the model did not change the results (data not shown).
DISCUSSION The relationship between coffee consumption and serum lipids has been investigated in a number of cross-sectional, longitudinal, and experimental studies, with conflicting results (3, 24-32). In a review of 22 cross-sectional studies (3), ten showed a significant positive relationship between coffee and serum total cholesterol in both sexes, while in the remaining 12 TABLE 3. Multiple linear regression coefficients for LDL-C in smokers and nonsmokers: the Olivetti Heart Study Dependent
variable
LDL-C
(mg/dL) Smokers (n = 504) Independent
variables
Coffee consumption
(cups/d)
Age (~1 Body mass index (kg/m*) RI
Nonsmokers (n = 396)
B
SE
B
SE
2.012” 0.4574 0.687”
1.068 0.233 0.667
0.175 0.632” 0.679
1.115 0.190 0.498
0.02
0.04
n P I 0.05. SE = standard error of B; LDL-C = low-density-lipoprotein cholesterol.
AEP Vol. 3, No. 3 May 1993: 250-255
COFFEE CONSUMPnON
Coffee categories (cups/d) 0 l-2 3-4 5+
LDL-C (mg/dL)
n 44 296 417 143
F (overall)
140.0 146.7 148.5 153.4
Smokers (n = 504)
Nonsmokers (n = 396) LDL-C (mg/dL)
n 35 195 143 23
2.414
143.3 146.8 148.5 147.2
LDL-C n
(mg/dL)
9 101 274 120
0.311
124.6 145.3b 148.21‘ 153.0 2.983’
studies the evidence for a link was less consistent. For example, in the Tromso Heart Study (24) coffee consumption was positively associated with cholesterol and TG levels in both sexes and was inversely associated with HDL-C in females, while in the Tecumseh Study (25) an association was reported only in women. In a cross-sectional study of nearly 5000 Australians, Shirlow and Mathers (26) reported that caffeine consumption was significantly associated with cholesterol levels only in females, while use of caffeinated coffee, but not total caffeine consumption, was positively related to cholesterol levels in males. Kark and colleagues (27) reported a positive association of coffee drinking (but not tea) with cholesterol in both sexes in Jerusalem. A recent report from the Health Professional Follow-up Study detected an increase in cholesterol levels with increasing consumption of decaffeinated coffee but not regular coffee (28). Finally, no significant relationship between coffee consumption and cholesterol has been found in the Framingham Study (29), the National Health and Nutrition Examination Survey (30), and the Western Electric Study (31). Lack of association in the present study, after multivariate adjustment, between coffee and TG and HDL-C is consistent with most of the findings of previous reports (see the article by Thelle and coauthors (3) for review). The TABLE 5. Adjuste& mean values of HDL-C by coffee consumption categories: the Olivetti Heart Study
Coffee categories (cups/d) 0 l-2 3-4 5+ F (overall)
n 44 296 417 143
HDL-C (mg/dL) 47.3 46.0 46.4 45.6 1.527
Nonsmokers (n = 396)
n 35 195 143 23
’Adjusted for age and body mass index. HDLC = high-density-lipoprotein cholesterol.
All (N = 900) Coffee categories (cups/d) 0 l-2 3-4 5+ F (overall)
n 44 296 417 143
TG (mg/dL) 140.8 139.1 151.2 156.0 1.677
Nonsmokers (n = 396)
n 35 195 143 23
TG (mg/dL) 138.5 134.3 142.2 131.3 1.591
Smokers (n = 504) TG n 9 101 274 120
(mg/dL) 124.8 139.2 158.0 158.6 2.456
’Adjusted for age and body mass index TG = triglycerides.
’Adjusted for age and body mass index. b P < 0.01. c P < 0.05. LDL-C = low-density-lipoprotein cholesterol.
All (N = 900)
253
TABLE 6. Adjusted” mean values of serum TG by coffee consumption categories: the Olivetti Heart Study
TABLE 4. Adjusteh mean values of LDL-C by coffee consumption categories: the Olivetti Heart Study All (N = 900)
Jossaet al. AND SERUM LIPIDS
HDL-C (mg/dL) 48.3 47.5 45.9 50.0 1.277
Smokers (n = 504) HDL-C n 9 101 274 120
(mg/dL) 43.7 44.0 41.0 40.9 2.254
results of the studies on the relationship of coffee consumption with coronary heart disease are contradictory as well. Several studies suggested that excessive coffee drinking is associated with an increased risk of coronary heart disease, but many others failed to show such an association (see earlier publications (1, 2) for reviews). Case-control and cross-sectional studies more often showed a positive association, while longitudinal studies generally did not support this finding (1, 2). The discrepancies on the association between coffee and cholesterol among studies have been postulated to be due to (a) the inaccurate measurement of coffee and caffeine intake (32), (b) differences in brewing methods (14-17), and/or (c) the possible role of confounding variables (33). With regard to the brewing methods, some investigators speculated that boiled or percolated coffee contains a lipid-raising factor which is removed by filtering (14-17). Therefore consumption of filtered coffee should not be associated with an increase in cholesterol. However, in contrast with this hypothesis, Pietinen and coworkers (34) reported in a Belgian study with approximately 18,000 participants, that consumption of filtered coffee was associated with increased cholesterol levels. The present study has a number of differences compared to previous studies that need to be emphasized. First, Italians drink “espresso” coffee, the grinding procedure and brewing method for which differs from those for boiled, filtered, or percolated coffee. Espresso is brewed in machines in which hot water under forced steam passes in a short time through darkly roasted, finely ground coffee beans. Compared to boiled and percolated coffee, brewed by a prolonged contact between coffee beans and hot water, espresso is characterized by a very short period of contact between water and the coffee grounds. Second, a demitasse of espresso (approximately 30 to 50 mL) contains on average 50 to 70 mg of caffeine (35), while the amount of caffeine@ a cup of boiled, filtered, or percolated coffee (approximately 150 to 200 mL) is much higher (100 to 150 mg) (36). Third, Italians drink their coffee black so that cream or milk consumption with coffee, hypothesized to be responsi-
254
Jossaet al. COFFEE
CONSUMPTION
AEP Vol. 3, No. 3 May 1993: 250-255
AND SERUM LIPIDS
ble for the increased cholesterol levels among coffee drinkers in other countries (37), does not apply to our population. Finally, diet and health behaviors
are different in our
sample compared to those observed in studies from northern European
and North
clude differences sumption,
American
countries
(38). These in-
in diet, cigarette smoking,
alcohol con-
and their intercorrelations.
Since both cigarette smoking and coffee consumption are life habits strongly related in the general population, their possible interactive effect on cholesterol may have important public health implications in the prevention of coronary heart disease. Further
studies are needed to confirm
this finding and to detect the responsible
components
in-
volved.
Despite these differences, we detected a positive relationship between coffee consumption terol and LDL-C
and serum total choles-
but only in smokers.
Dr. Trevisan
Our results support the previous findings of Heyden and associates (39), who found that LDL-C cantly higher and HDL-C significantly) County
levels were signifi-
levels were lower (although
in 361 smoking
participants
Study who consumed
not
of the Evans
five or more cups of coffee
per day compared to nonsmokers
tween smoking and coffee among 2033 Australian
found a significant positive
between coffee and cholesterol
not in nonsmokers,
after controlling
in smokers but
for a number of con-
founders (age, body mass index, alcohol consumption,
and
dietary fats). The observed interactive tween coffee consumption
effect suggests a synergism be-
and cigarette smoking on choles-
terol levels, although neither the pharmacologically agents in cigarette mechanisms
have been elucidated. As experimental
have shown (41, 42), the only established
metabolic
of cigarette smoking and coffee consumption in free fatty acid and TG concentrations
studies effect
is an increase through
release. (In addition,
gested that coffee contains the excretion
active
smoke or coffee nor their physiologic
crease in catecholamine
an in-
Bjelke sug-
components
that may reduce
of bile acids and neutral
sterols, therefore
increasing serum total cholesterol
levels and reducing the
risk factor for colon cancer (43).) It has also been questioned the components
responsible
whether caffeine is one of
for the observed
association
between coffee and cholesterol (12-14, 27, 28,41,44). Two trials found that decaffeinated coffee increased cholesterol levels (28,45)
and a number of studies reported no associa-
tion of cholesterol with caffeinated beverages different from coffee (12, 13, 27). A number of studies reported that the positive association between coffee and cholesterol could be related to the clustering of atherogenic behaviors in coffee drinkers compared to noncoffee drinkers (9, 11, 13): for instance, drinkers may have a more atherogenic
coffee
diet compared
to
noncoffee drinkers. Dietary data are not available in the present study, but a cross-sectional observation carried out in Italy (40) detected a significant association between coffee and cholesterol
after controlling
of a Research
was funded
Heart,
Career
Lung,
in part by a grant from NATO
Development
and Blood
Award
Institute.
for International
(K04
The study
Collaboration
in Research. We would Company,
like to acknowledge
with special thanks
and Ms. Ida Bartolomei
(head
the support
to Dr. Antonio nurse)
received
from the Olivetti
Scottoni
(medical
of the Olivetti
Factory
director)
in Naples.
be-
females
but not in 2724 males. Panic0 and coworkers (40), in the Italian Nine Community Study, a multicenter investigation association
is a recipient
from the National
who did not drink coffee.
Shirlow and Mathers (26) detected a similar interaction
of risk factors for atherosclerosis,
HL02189)
for dietary fats.
REFERENCES 1. Rosmarin Cardiovasc 2. Christiansen consumption 15:391-408.
PC. Coffee and coronary Dis. 1989;32:239-45.
heart
disease:
L, Murray T. A review of the relationship and coronary heart disease, J Community
A review,
Prog
between Health.
coffee 1990;
3. Thelle DS, Heyden S, Fodor JG. Coffee and cholesterol in epidemiolog ical and experimental studies, Atherosclerosis. 1987;67:97-103. 4. Stensvold 1, Tverdal A, Per Foss 0. The effect of coffee on blood lipids and blood pressure. Results from a Norweigian cross-sectional study, men and women, 40-42 years, J Clin Epidemiol. 1989;42:877-84. 5. Salvaggio A, Periti M, Miano L, Zambelli C. Association between habitual coffee consumption and blood pressure levels, J Hypertens. 1990;8:585-90. 6. Klatsky AL, Friedman GD, Siegelaub AB. Coffee drinking prior to acute myocardial infarction. Results from the Kaiser Permanente epidemiologic study of myocardial infarction, JAMA. 1973;226:540-3. 7. Kannel WB. Coffee, Med. 1977;297:443-4. 8. Anonymous. 1283-4.
Coffee
cocktails, and
and
cholesterol
coronary
candidates,
(editorial),
N Engl J
Lancet.
9. Haffner SM, Knapp JA, Stern I’, Hazuda HP, Rosenthal LJ. Coffee consumption, diet, and lipids, Am J Epidemiol. 1-12.
1985;2:
M, France 1985;122:
10. Jacobsen BK, Thelle DS. The Tromso Heart Study: Is coffee drinking an indicator of a life-style with high risk for ischemic heart disease?, Acta Med Stand. 1987;222:215-21. J, Barrett-Connor E, Ganiats TG. Clustering 11. Puccio EM, McPhyllips of atherogenic behaviors in coffee drinkers, Am J Public Health. 1990; 80:1310-3. 12. Curb JD, Reed DM, Kantz JA, Yano K. Coffee, caffeine cholesterol in Japanese men in Hawaii, Am J Epidemiol.
and serum 1986;123:
648-6. 13. Klatsky AL, Petitti DB, Armstrong MA, Friedman and cholesterol, Am J Cardiol. 1985;55:577-8.
GD.
Coffee,
tea
M, Harryvan JL. 14. Zoeck PL, Katan MB, Merkus MP, Van Dusseldorp Effect of lipid rich-fraction from boiled coffee on serum cholesterol, Lancet. 1990;335:1235-7. Grobbee ED. The effect on serum cholesterol levels of 15. Bak AAA, coffee brewed by filtering or boiling, N Engl J Med 1989;321:1432-7. J, Kostiainen 16. Aro A, Tuomilehto coffee increases serum low density lism. 1987;36:1027-30.
E, Uusitalo U, Pietinen lipoprotein concentration,
P. Boiled Metabo-
Jossa et al.
AEP Vol. 3, No. 3 May 1993: 250-255
COFFEE
17. Thelle DS. Coffee, cholesterol is in the brewing,
and coronary
heart disease. The secret
Br Med J. 1991;302:804.
CONSUMPTION
AND SERUM
255
LIPIDS
32. Shreiber GB, Maffeo CE, Robins M, Masters MN, Bond AI’. Measurement of coffee and caffeine intake: Implications for epidemiologic research,
Prev Med. 1988;17:280-94.
18. Farinaro E, Panico S, Oriente P, Paggi E, Mancini M. Prevalence of risk factors in an urban working population of southern Italy. In: Carlson LA, Paoletti R, Sirtori CR, Weber G, eds. International Conference on Atherosclerosis. New York: Raven Press,l978:375-8.
33. Shreiber GB, Robins M, Maffeo CE, Masters MN, Bond AP, Morganstein D. Confounders contributing to the reported associations of coffee or caffeine with disease, Prev Med. 1988; 17:295-309.
19. Rose GA, Blackburn H, Gillum RF, Prineas RJ. Cardiovascular Methods. Geneva: WHO; 1982:936-942.
34. Pietinen P, Geboers J, Kesteloot cholesterol: An epidemiological 1988;17:98-104.
Survey
20. Siedel J, Hagele EO, Ziegenhorn J, Wahlefeld AW. Reagent for the enzymatic determination of serum total cholesterol with improved lipolytic efficiency, Clin Chem. 1983;29:1075-80. 21
Nagele V, Hagele EO, Lauer G, et al. Reagent for the enzymatic determination of serum total triglycerides with improved lipolytic efficiency, J Clin Chem Biochem. 1984;22:165-74.
22
Draeger G, Wahlefeld AW, Ziegenhorn J. A practical HDL-cholesterol quantitation, Lab Med. 1982;6:198-202.
method
23
Kleinbaum DC, K upper LL, Muller KE. Dummy variables in regression. In: Applied Regression Analysis and Other Multivariable Methods. 2nd ed. 1988:260-97.
24. Thelle DS, Arnesen E, Forde OH. The Tromso Heart Study. coffee raise serum cholesterol?, N Engl J Med. 1983;300:1454-7.
for
Does
25. Nichols AB, Ravencroft C, Lamphiear DE, et al. Independence of serum lipid levels and dietary habits: The Tecumseh Study, JAMA. 1976;236:1948-53, 26. Shirlow MJ, Mathers CD. Caffeine consumption terol levels, Int J Epidemiol. 1984;13:422-7.
and serum choles-
27. Kark JD, Friedlander Y, Kaufmann NA, Stein Y. Coffee, tea, and plasma cholesterol: The Jerusalem Lipid Research Clinic prevalence study, Br Med J. 1985;291:699-704. 28. Grobbee DE, Rimm EB, Giovannucci E, et al. Coffee, caffeine and cardiovascular disease in men, N Engl J Med. 1990;323:1026-32. 29. Dawber TR, Kannel WB, Gordon T. Coffee and cardiovascular disease: Observations from the Framingham Study, N Engl J Med. 1974; 291:871-4. 30. Kovan MC, Fulwood R, Feinleib N Engl J Med. 1983;309:1249.
M. Coffee and cholesterol
(letter),
31. Shekelle RB, Gale M, Paul 0, Stamler J. Coffee and cholesterol N Engl J Med. 1983;309:1249-50.
(letter),
H. Coffee consumption and serum study in Belgium, Int J Epidemiol.
35. Finocchiaro G, Pezzana A, Pernigotti L, Bo M. The influence of coffee on plasma lipids. In: Thelle DC, van der Stegen, eds. Coffee and Coronary Heart Disease. The Nordic School of Public Health. 1990: 101-5. 36. Gilbert RM. Caffeine as a drug abuse. In: Gibbins RJ, Israel Y, Kalant H, Popham RE, Schmidt W, Smart RG, eds. Research Advances in Alcohol and Drug Problems. v. 3. John Wiley and Sons; 1976:49178. 37. Roeckel I. Coffee 1248-178.
and cholesterol
(letter).
N Engl J Med.
1983;309:
38. Keys A. Seven Countries. A Multivariate Analysis of Death and Coronary Heart Disease. Cambridge, MA: Harvard University Press; 1980. 39. Heyden S, Heiss G, Manegold C, et al. The combined effect of smoking and coffee drinking on LDL and HDL-cholesterol, Circulation. 1979; 60:22-5. 40. Panic0 S, Celentano E, Krogh V. Coffee and blood lipids in Italy. In: Thelle DC, van der Stegen, eds. Coffee and Coronary Heart Disease. The Nordic School of Public Health. 1990:63-70. 41. Whitsett TL, Manion CV, Christensen HD. Cardiovascular coffee and caffeine, Am J Cardiol. 1984;53:918-22.
effect of
42. Robertson D, Froelich JC, Carr RK, et al. Effect of caffeine on plasma renin activity, catecholamine and blood pressure, N Engl J Med. 1978; 298:181. 43. Bjelke E. Colon
cancer
and blood cholesterol,
Lancet.
1974;1:1116.
44. Brown CR, Benowitz NL. Caffeine and cigarette smoking: Behavioral, cardiovascular and metabolic interactions, Pharmacol Biochem Behav. 1989;34:565-70. 45. Naismith DJ, Akinyanju PA, Szanto S, Yudkin J. The effect in volunteers of coffee and decaffeinated coffee on blood glucose, insulin, plasma lipids, and some factors involved in blood clotting, Nutr Metab. 1970;12:144-51.
MS
The relationship between coffee consumption and blood lipids was analyzed in a sample of 900 male workers of southern Italy participating in the Olivetti Heart Study. In the univariate analysis, coffee drinkers (n = 856) had higher values for body mass index (P 5 0.05) and number of cigarettes smoked per day (P I O.OOl),and lower levels of serum high-density-lipoprotein cholesterol @’5 0.0.5), compared to noncoffee drinkers (n = 44). In addition, coffee consumption (cups/d) was positively related to serum triglyceride levels (r = 0.105, P I 0.01) and cigarette smoking (r = 0.491, P 5 0.01), and was inversely related to age (r = -0.122, P I 0.01). After multivariate adjustment, coffee consumption remained significantly related to age, cigarette smoking, and body mass index (data not shown). After stratification for smoking status, a significant positive linear trend between coffee consumption and serum total cholesterol was observed only in smokers. No significant trend was observed for serum triglycerides and high-density-lipoprotein cholesterol with coffee intake according to smoking status. This finding suggests that the relationship between coffee consumption and serum total cholesterol may change with the smoking status. Ann Epidemiol 1993; 3:250-255. KEY WORDS:
Coffee
consumption,
serum total cholesterol, serum triglycerides, cigarette
smoking.
This study analyzed the relationship between coffee con-
INTRODUCTION Epidemiologic
studies have suggested a relation
coffee consumption cently,
investigators
relationship
and coronary
between
heart disease (1, 2). Re-
have focused their attention
between coffee and a number
on the
of risk factors
sumption and blood lipids in a sample of healthy workers of southern Italy consuming “espresso” coffee, which differs from boiled, filtered, or percolated coffee in caffeine content and in brewing methods.
for coronary heart disease, (e.g., serum lipids (3) and blood pressure (4, 5)), in order to define a possible mechanism this association.
The relationship
tion and serum total cholesterol cross-sectional
for
between coffee consumphas been evaluated in 22
studies carried out in eight different coun-
tries (see the article by Thelle and coworkers (3) for review); the findings were somewhat
conflicting.
It has been sug-
gested that this association may be related to the clustering of atherogenic
behaviors
in coffee drinkers
noncoffee drinkers (6- 1 I), components
compared
to
of coffee other than
caffeine (since tea or cola drinks do not have the same effect) (9, 12, 13), or particular
methods
of brewing
(14-17).
MATERIALS
AND
METHODS
The Olivetti Heart Study is a longitudinal risk factors for atherosclerosis,
Factory in Naples, Italy (18). The baseline examination
on was
conducted in 1975 to 1976; the 5-year and 12-year follow-up examinations
were carried out in 1980 to 1981 and 1987 to
1988, respectively. At each examination, invited to participate, previous examinations.
all employees were
regardless of their participation
in
At the 1987 to 1988 examination
a total of 990 workers (942 males and 48 females) agreed to participate in the study, a participation
From the Department of Social and Preventive Medicine, State University of New York at Buffalo, Buffalo, NY (F. J., V. K., M. J.), and the Institute of Internal Medicine and Metabolic Diseases, University of Naples, Medical School, Naples, Italy (F. J., V. K., E. F., S. P., D. G., R. G., E. C., M. M., M. T.). Address reprint requests to: Fabrizio Jossa, MD, Department of Social and Preventive Medicine, State University of New York at Buffalo, 270 Farber Hall, Buffalo, NY 14214. Received January 28, 1992; revised August 26, 1992.
investigation
carried out at the Olivetti
rate of 80%. Because
of the small number, females were excluded from the present analysis. The data reported here were collected in the 1987 survey. The participants
were seen in the morning,
in the
medical facilities of the factory, after an overnight fast. The medical examination included (a) systolic and diastolic blood pressure measurements, electrocardiogram,
(b) a 12-lead standard resting
(c) anthropometric
measurements,
and
1047.2797/93/$06.00
AEP Vol. 3, No. 3 Mav 1993: 250-255
COFFEE
TABLE 1. Baseline characteristics
of 900 male workers in southern
Italy (mean f standard
Age (Y)
46.9
Body mass index serum
(kg/m2)
cholesterol
(mg/dL)
Low-density-lipoprotein Serum
cholesterol
triglycerid&
High-density-lipoprotein Alcohol
(mL/wk)
consumption
Cigarette
smoking
Prevalence
+ 7.5
(mg/dL)
46.9
? 7.4
47.7 f
9.8
26.4 2 3.1
25.3 f
2.7
148.0
+ 31.7
168.4 2 31.8
139.9 -r- 30.0
149.8 2 89.2
140.8 f
264.7
f
89.6 11.5
? 332.1
(cigarettes/d)
10.7 2 11.3
(96)
21
to gather
information
quality
balance
scale with participants
cilip) for HDL-C.
clothing.
Body mass index
shoeless (kg/m’)
and wearing
was calculated.
were performed protocol
by trained
described
Methods
indoor
by the
of the World
Manual Health
for Orga-
Coffee
consumption
was
the habitual
assessed
number
by
one
consumption
to evaluate
previous
variable
because
was ascertained
the amount
week and was also analyzed
was calculated
according
= 11% ethanol
volume, concerning
= 30% ethanol
the number
of cigarettes
described
cular ResearchSurvey Smoking
beer
was measured
tion of serum cholesterol
were both determined
2 1). High-density-lipoprotein acid
All biochemical
for Cardiovas-
Organization
variable
by enzymatic
- HDL-C
measurements
(19).
Serum
total
(TG) concen-
statistics matrix.
out to calculate cholesterol Multiple N-l
of the relationship
the adjusted
and
among
variables
standard
of covariance mean
to coffee
data were obtained
dummy
heart disease risk
(mean Analysis
according
comparisons
values
of
consumption
the different
cate-
using regression
(23). Multiple
linear
regression analysis was performed to evaluate linear between LDL-C and coffee consumption (cups/d).
trend
(20,
Table
1 summarizes
whole
sample
chloride-phosphomethods
separately. (22).
was calculated
the
and coffee
These
southern
baseline drinkers
for the drinkers
men were characterized
by a high weekly alcohol consumption and a high prevalence of both smokers (56%) and coffee drinkers (95%). The
noncoffee under
characteristics and noncoffee
Italian
coffee drinkers had significantly index and number of cigarettes
- (TG/5) performed
RESULTS
was ana-
procedures
(HDL-C)
enzymatic
were
after
(cigarettes/d).
variables.
triglycerides
and
descriptive
serum
with
was collected
and coronary
was carried
analysis
for
intake),
for determina-
Low-density-lipoprotein cholesterol (LDL-C) according to the following formula: CHOL
per day accord-
were available
for the assessment
and correlation
total
data
available
values for alcohol
on this parameter
methods
included
deviation)
this report
of missing
not being
consumption
coffee consumption
factors
analysis,
in any of
of the study.
gories for the adjusted
of magnesium
precipitation
Statistical between
for 939 of
data
The majority
(5 1 were missing
categories.
cholesterol
lyzed by a combination
by
in the present
on alcohol
and 5.9 (Pre-
available
of missing
lipid determinations
information
of variation for cholesterol,
for TG,
were
Because
considered
by a set of questions
by venipuncture
and serum
table:
1.65 (Precipat)
consumption
by volume. smoked
lipid levels and other
(CHOL)
of ethanol
conversion
of the World Health
was drawn
vari-
= 3.5% ethanol
by the Manual
was used as a continuous
Blood sample
dein the
as a continuous
to the following
smoking
ing to the protocol
consumed
consumption
by volume,
and liquor
Cigarette
by questions
of alcohol
Coefficients
1.65 (Precipat)
on 900 participants.
(n = 30). Data
The weekly absolute
tungstic
focuses
849 participants
able (mL/wk).
trations
the variables
the beginning
Alcohol
wine
on coffee
question
as a continuous
and
the 942 male workers.
of cups of coffee consumed
(cups/d).
procedure:
and
was due to serum
per day and was used in the analysis
signed
control
2.15 (Precinorm) Data
physicians
(19).
concerning
199.3 + 275.7
1.6 3.0 + 6.9d
1.84 (Precinorm)
nization
* 334.7
(SI) = mg/dL x 0.02586. (si) = mg/dL x 0.01129. vs. coffee drinkers. vs. coffee drinkers.
questionnaire
Survey
97.6
47.4 & 12.2’
58
strict
Cardiovascular
+ 36.7
11.1 + 11.4
were
to the
211.3
11.5
3.0 ?
56
(d) a self-administered
according
?
268.3
on life-style characteristics and personal and family history. In detail, weight and height were measured on a beam
All measurements
+ 41.3
43.4
3.0 + 1.6
’ Standard International Units ’ Standard International Units ’P < 0.05, noncoffee drinkers ’ P < 0.001, noncoffee drinkers
drinkers
(n = 44)
221.8
(cups/d)
of smokers
Noncoffee
+ 41.2
43.6
(n = 849)
drinkers
221.3
149.3 t
cholesterol
consumption
Coffee
(mg/dL)”
(mg/dL)
the Olivetti Heart Study
(n = 856)
26.3 ? 3.1
251
Jossa et al. SERUM LIPIDS
AND
deviation):
Coffee
All (N = 900)
Variables
Total
CONSUMPTION
drinkers,
in the coffee drinkers
while
higher levels of body mass smoked per day than did
HDL-C
compared
was significantly
to the noncoffee
lower drinkers.
252
Jossaet al. COFFEE CONSUMFTON
AEP Vol. 3, No. 3 May 1993: 250-255
AND SERUM LIF’IDS
TABLE 2. Correlation matrix: coffee consumption, lioids: the Olivetti Heart StudP
body mass index, age, cigarette smoking, alcohol intake, and serum
Body coffee Coffee consumption
Age
mass index
Smoking
0.12lb -0.050 0.043 0.12Ob 0.052 0.011 0.143b
-O.lOlb -0.021 0.142b 0.239b -0.162b 0.092‘
0.047 0.019 0.107b -0.226b 0.048
Alcohol (n = 859)
Cholesterol
High-densitylipoprotein cholesterol
Triglycerides
(cups/d)
Age (Y) Body mass index (kg/m’) Cigarette smoking (cigarettes/d) Alcohol intake (mL/wk) (n = 849) Serum total cholesterol (mg/dL) Serum triglyceridesb (mg/dL) High-density-lipoprotein cholesterol (mg/dL) Low-density-lipoprotein cholesterol (mg/dL)
-0.122b 0.031 0.491b 0.032 0.060 0.105b -0.152b 0.073
0.024 - 0.020 0.141b
0.007
0.40FJb 0.063 0.910b
-0.344b 0.158”
-0.066
”N = 900 unless otherwise indicated. b P I 0.01. r P _c 0.05.
Table 2 reports the correlation matrix for coffee consumption (cups/d) and other coronary heart disease risk factors. Coffee intake (cups/d) was significantly and positively related with number of cigarettes smoked per day and TG level, while it was inversely associated with age and HDL-C level. Serum total cholesterol and TG concentrations were both positively related to body mass index, while TG correlated positively with cigarette smoking and negatively with HDL-C; HDL-C was inversely related to cigarette smoking and body mass index and positively related with alcohol intake. In addition, LDL-C was positively related with age, body mass index, cholesterol, and TG. A significant positive interaction was detected between coffee use (yes/no) and smoking (yes/no) with regard to total serum cholesterol but not TG or HDL-C (data not shown). Therefore, the association between total serum cholesterol and coffee consumption was analyzed separately in smokers and nonsmokers. Table 3 reports the outcome of the multiple regression analysis for LDL-C after stratification for smoking status: Coffee consumption was an independent positive correlate of LDL-C only in smokers. Age was positively and significantly related to LDL-C both in smokers and nonsmokers. As indicated by the regression coefficient in Table 3, one additional cup of coffee was associtaed with an increase in LDL-C levels of 2.01 mg/dL in smokers and of 1.17 mg/ dL in nonsmokers. In the sample as a whole, in a multiple regression analysis including age, body mass index, and cigarette smoking, the relationship between coffee and either serum TG or HDL-C failed to reach statistical significance (data not shown). Tables 4,5, and 6 show the adjusted (age and body mass index) mean values for cholesterol, HDLC, and TG in the different categories of coffee consumption for the whole sample and in smokers and nonsmokers separately. The analysis of covariance indicated an overall significant difference for the average LDL-C level in the various catego-
ries of coffee intake (cups/d) in only the smokers. After participants who smoke but do not drink coffee (n = 9) were excluded, the observed difference was still significant (data not shown). No significant association was detected for coffee and LDL-C in nonsmokers and in the sample as a whole. A decrease in the serum concentration of HDL-C (see Table 5) and an increase in TG (see Table 6) were detectable across coffee categories in smokers, and in the group as a whole, even though in both cases the results of analysis of covariance did not indicate statistical significance. The inclusion of alcohol consumption as a covariate in the model did not change the results (data not shown).
DISCUSSION The relationship between coffee consumption and serum lipids has been investigated in a number of cross-sectional, longitudinal, and experimental studies, with conflicting results (3, 24-32). In a review of 22 cross-sectional studies (3), ten showed a significant positive relationship between coffee and serum total cholesterol in both sexes, while in the remaining 12 TABLE 3. Multiple linear regression coefficients for LDL-C in smokers and nonsmokers: the Olivetti Heart Study Dependent
variable
LDL-C
(mg/dL) Smokers (n = 504) Independent
variables
Coffee consumption
(cups/d)
Age (~1 Body mass index (kg/m*) RI
Nonsmokers (n = 396)
B
SE
B
SE
2.012” 0.4574 0.687”
1.068 0.233 0.667
0.175 0.632” 0.679
1.115 0.190 0.498
0.02
0.04
n P I 0.05. SE = standard error of B; LDL-C = low-density-lipoprotein cholesterol.
AEP Vol. 3, No. 3 May 1993: 250-255
COFFEE CONSUMPnON
Coffee categories (cups/d) 0 l-2 3-4 5+
LDL-C (mg/dL)
n 44 296 417 143
F (overall)
140.0 146.7 148.5 153.4
Smokers (n = 504)
Nonsmokers (n = 396) LDL-C (mg/dL)
n 35 195 143 23
2.414
143.3 146.8 148.5 147.2
LDL-C n
(mg/dL)
9 101 274 120
0.311
124.6 145.3b 148.21‘ 153.0 2.983’
studies the evidence for a link was less consistent. For example, in the Tromso Heart Study (24) coffee consumption was positively associated with cholesterol and TG levels in both sexes and was inversely associated with HDL-C in females, while in the Tecumseh Study (25) an association was reported only in women. In a cross-sectional study of nearly 5000 Australians, Shirlow and Mathers (26) reported that caffeine consumption was significantly associated with cholesterol levels only in females, while use of caffeinated coffee, but not total caffeine consumption, was positively related to cholesterol levels in males. Kark and colleagues (27) reported a positive association of coffee drinking (but not tea) with cholesterol in both sexes in Jerusalem. A recent report from the Health Professional Follow-up Study detected an increase in cholesterol levels with increasing consumption of decaffeinated coffee but not regular coffee (28). Finally, no significant relationship between coffee consumption and cholesterol has been found in the Framingham Study (29), the National Health and Nutrition Examination Survey (30), and the Western Electric Study (31). Lack of association in the present study, after multivariate adjustment, between coffee and TG and HDL-C is consistent with most of the findings of previous reports (see the article by Thelle and coauthors (3) for review). The TABLE 5. Adjuste& mean values of HDL-C by coffee consumption categories: the Olivetti Heart Study
Coffee categories (cups/d) 0 l-2 3-4 5+ F (overall)
n 44 296 417 143
HDL-C (mg/dL) 47.3 46.0 46.4 45.6 1.527
Nonsmokers (n = 396)
n 35 195 143 23
’Adjusted for age and body mass index. HDLC = high-density-lipoprotein cholesterol.
All (N = 900) Coffee categories (cups/d) 0 l-2 3-4 5+ F (overall)
n 44 296 417 143
TG (mg/dL) 140.8 139.1 151.2 156.0 1.677
Nonsmokers (n = 396)
n 35 195 143 23
TG (mg/dL) 138.5 134.3 142.2 131.3 1.591
Smokers (n = 504) TG n 9 101 274 120
(mg/dL) 124.8 139.2 158.0 158.6 2.456
’Adjusted for age and body mass index TG = triglycerides.
’Adjusted for age and body mass index. b P < 0.01. c P < 0.05. LDL-C = low-density-lipoprotein cholesterol.
All (N = 900)
253
TABLE 6. Adjusted” mean values of serum TG by coffee consumption categories: the Olivetti Heart Study
TABLE 4. Adjusteh mean values of LDL-C by coffee consumption categories: the Olivetti Heart Study All (N = 900)
Jossaet al. AND SERUM LIPIDS
HDL-C (mg/dL) 48.3 47.5 45.9 50.0 1.277
Smokers (n = 504) HDL-C n 9 101 274 120
(mg/dL) 43.7 44.0 41.0 40.9 2.254
results of the studies on the relationship of coffee consumption with coronary heart disease are contradictory as well. Several studies suggested that excessive coffee drinking is associated with an increased risk of coronary heart disease, but many others failed to show such an association (see earlier publications (1, 2) for reviews). Case-control and cross-sectional studies more often showed a positive association, while longitudinal studies generally did not support this finding (1, 2). The discrepancies on the association between coffee and cholesterol among studies have been postulated to be due to (a) the inaccurate measurement of coffee and caffeine intake (32), (b) differences in brewing methods (14-17), and/or (c) the possible role of confounding variables (33). With regard to the brewing methods, some investigators speculated that boiled or percolated coffee contains a lipid-raising factor which is removed by filtering (14-17). Therefore consumption of filtered coffee should not be associated with an increase in cholesterol. However, in contrast with this hypothesis, Pietinen and coworkers (34) reported in a Belgian study with approximately 18,000 participants, that consumption of filtered coffee was associated with increased cholesterol levels. The present study has a number of differences compared to previous studies that need to be emphasized. First, Italians drink “espresso” coffee, the grinding procedure and brewing method for which differs from those for boiled, filtered, or percolated coffee. Espresso is brewed in machines in which hot water under forced steam passes in a short time through darkly roasted, finely ground coffee beans. Compared to boiled and percolated coffee, brewed by a prolonged contact between coffee beans and hot water, espresso is characterized by a very short period of contact between water and the coffee grounds. Second, a demitasse of espresso (approximately 30 to 50 mL) contains on average 50 to 70 mg of caffeine (35), while the amount of caffeine@ a cup of boiled, filtered, or percolated coffee (approximately 150 to 200 mL) is much higher (100 to 150 mg) (36). Third, Italians drink their coffee black so that cream or milk consumption with coffee, hypothesized to be responsi-
254
Jossaet al. COFFEE
CONSUMPTION
AEP Vol. 3, No. 3 May 1993: 250-255
AND SERUM LIPIDS
ble for the increased cholesterol levels among coffee drinkers in other countries (37), does not apply to our population. Finally, diet and health behaviors
are different in our
sample compared to those observed in studies from northern European
and North
clude differences sumption,
American
countries
(38). These in-
in diet, cigarette smoking,
alcohol con-
and their intercorrelations.
Since both cigarette smoking and coffee consumption are life habits strongly related in the general population, their possible interactive effect on cholesterol may have important public health implications in the prevention of coronary heart disease. Further
studies are needed to confirm
this finding and to detect the responsible
components
in-
volved.
Despite these differences, we detected a positive relationship between coffee consumption terol and LDL-C
and serum total choles-
but only in smokers.
Dr. Trevisan
Our results support the previous findings of Heyden and associates (39), who found that LDL-C cantly higher and HDL-C significantly) County
levels were signifi-
levels were lower (although
in 361 smoking
participants
Study who consumed
not
of the Evans
five or more cups of coffee
per day compared to nonsmokers
tween smoking and coffee among 2033 Australian
found a significant positive
between coffee and cholesterol
not in nonsmokers,
after controlling
in smokers but
for a number of con-
founders (age, body mass index, alcohol consumption,
and
dietary fats). The observed interactive tween coffee consumption
effect suggests a synergism be-
and cigarette smoking on choles-
terol levels, although neither the pharmacologically agents in cigarette mechanisms
have been elucidated. As experimental
have shown (41, 42), the only established
metabolic
of cigarette smoking and coffee consumption in free fatty acid and TG concentrations
studies effect
is an increase through
release. (In addition,
gested that coffee contains the excretion
active
smoke or coffee nor their physiologic
crease in catecholamine
an in-
Bjelke sug-
components
that may reduce
of bile acids and neutral
sterols, therefore
increasing serum total cholesterol
levels and reducing the
risk factor for colon cancer (43).) It has also been questioned the components
responsible
whether caffeine is one of
for the observed
association
between coffee and cholesterol (12-14, 27, 28,41,44). Two trials found that decaffeinated coffee increased cholesterol levels (28,45)
and a number of studies reported no associa-
tion of cholesterol with caffeinated beverages different from coffee (12, 13, 27). A number of studies reported that the positive association between coffee and cholesterol could be related to the clustering of atherogenic behaviors in coffee drinkers compared to noncoffee drinkers (9, 11, 13): for instance, drinkers may have a more atherogenic
coffee
diet compared
to
noncoffee drinkers. Dietary data are not available in the present study, but a cross-sectional observation carried out in Italy (40) detected a significant association between coffee and cholesterol
after controlling
of a Research
was funded
Heart,
Career
Lung,
in part by a grant from NATO
Development
and Blood
Award
Institute.
for International
(K04
The study
Collaboration
in Research. We would Company,
like to acknowledge
with special thanks
and Ms. Ida Bartolomei
(head
the support
to Dr. Antonio nurse)
received
from the Olivetti
Scottoni
(medical
of the Olivetti
Factory
director)
in Naples.
be-
females
but not in 2724 males. Panic0 and coworkers (40), in the Italian Nine Community Study, a multicenter investigation association
is a recipient
from the National
who did not drink coffee.
Shirlow and Mathers (26) detected a similar interaction
of risk factors for atherosclerosis,
HL02189)
for dietary fats.
REFERENCES 1. Rosmarin Cardiovasc 2. Christiansen consumption 15:391-408.
PC. Coffee and coronary Dis. 1989;32:239-45.
heart
disease:
L, Murray T. A review of the relationship and coronary heart disease, J Community
A review,
Prog
between Health.
coffee 1990;
3. Thelle DS, Heyden S, Fodor JG. Coffee and cholesterol in epidemiolog ical and experimental studies, Atherosclerosis. 1987;67:97-103. 4. Stensvold 1, Tverdal A, Per Foss 0. The effect of coffee on blood lipids and blood pressure. Results from a Norweigian cross-sectional study, men and women, 40-42 years, J Clin Epidemiol. 1989;42:877-84. 5. Salvaggio A, Periti M, Miano L, Zambelli C. Association between habitual coffee consumption and blood pressure levels, J Hypertens. 1990;8:585-90. 6. Klatsky AL, Friedman GD, Siegelaub AB. Coffee drinking prior to acute myocardial infarction. Results from the Kaiser Permanente epidemiologic study of myocardial infarction, JAMA. 1973;226:540-3. 7. Kannel WB. Coffee, Med. 1977;297:443-4. 8. Anonymous. 1283-4.
Coffee
cocktails, and
and
cholesterol
coronary
candidates,
(editorial),
N Engl J
Lancet.
9. Haffner SM, Knapp JA, Stern I’, Hazuda HP, Rosenthal LJ. Coffee consumption, diet, and lipids, Am J Epidemiol. 1-12.
1985;2:
M, France 1985;122:
10. Jacobsen BK, Thelle DS. The Tromso Heart Study: Is coffee drinking an indicator of a life-style with high risk for ischemic heart disease?, Acta Med Stand. 1987;222:215-21. J, Barrett-Connor E, Ganiats TG. Clustering 11. Puccio EM, McPhyllips of atherogenic behaviors in coffee drinkers, Am J Public Health. 1990; 80:1310-3. 12. Curb JD, Reed DM, Kantz JA, Yano K. Coffee, caffeine cholesterol in Japanese men in Hawaii, Am J Epidemiol.
and serum 1986;123:
648-6. 13. Klatsky AL, Petitti DB, Armstrong MA, Friedman and cholesterol, Am J Cardiol. 1985;55:577-8.
GD.
Coffee,
tea
M, Harryvan JL. 14. Zoeck PL, Katan MB, Merkus MP, Van Dusseldorp Effect of lipid rich-fraction from boiled coffee on serum cholesterol, Lancet. 1990;335:1235-7. Grobbee ED. The effect on serum cholesterol levels of 15. Bak AAA, coffee brewed by filtering or boiling, N Engl J Med 1989;321:1432-7. J, Kostiainen 16. Aro A, Tuomilehto coffee increases serum low density lism. 1987;36:1027-30.
E, Uusitalo U, Pietinen lipoprotein concentration,
P. Boiled Metabo-
Jossa et al.
AEP Vol. 3, No. 3 May 1993: 250-255
COFFEE
17. Thelle DS. Coffee, cholesterol is in the brewing,
and coronary
heart disease. The secret
Br Med J. 1991;302:804.
CONSUMPTION
AND SERUM
255
LIPIDS
32. Shreiber GB, Maffeo CE, Robins M, Masters MN, Bond AI’. Measurement of coffee and caffeine intake: Implications for epidemiologic research,
Prev Med. 1988;17:280-94.
18. Farinaro E, Panico S, Oriente P, Paggi E, Mancini M. Prevalence of risk factors in an urban working population of southern Italy. In: Carlson LA, Paoletti R, Sirtori CR, Weber G, eds. International Conference on Atherosclerosis. New York: Raven Press,l978:375-8.
33. Shreiber GB, Robins M, Maffeo CE, Masters MN, Bond AP, Morganstein D. Confounders contributing to the reported associations of coffee or caffeine with disease, Prev Med. 1988; 17:295-309.
19. Rose GA, Blackburn H, Gillum RF, Prineas RJ. Cardiovascular Methods. Geneva: WHO; 1982:936-942.
34. Pietinen P, Geboers J, Kesteloot cholesterol: An epidemiological 1988;17:98-104.
Survey
20. Siedel J, Hagele EO, Ziegenhorn J, Wahlefeld AW. Reagent for the enzymatic determination of serum total cholesterol with improved lipolytic efficiency, Clin Chem. 1983;29:1075-80. 21
Nagele V, Hagele EO, Lauer G, et al. Reagent for the enzymatic determination of serum total triglycerides with improved lipolytic efficiency, J Clin Chem Biochem. 1984;22:165-74.
22
Draeger G, Wahlefeld AW, Ziegenhorn J. A practical HDL-cholesterol quantitation, Lab Med. 1982;6:198-202.
method
23
Kleinbaum DC, K upper LL, Muller KE. Dummy variables in regression. In: Applied Regression Analysis and Other Multivariable Methods. 2nd ed. 1988:260-97.
24. Thelle DS, Arnesen E, Forde OH. The Tromso Heart Study. coffee raise serum cholesterol?, N Engl J Med. 1983;300:1454-7.
for
Does
25. Nichols AB, Ravencroft C, Lamphiear DE, et al. Independence of serum lipid levels and dietary habits: The Tecumseh Study, JAMA. 1976;236:1948-53, 26. Shirlow MJ, Mathers CD. Caffeine consumption terol levels, Int J Epidemiol. 1984;13:422-7.
and serum choles-
27. Kark JD, Friedlander Y, Kaufmann NA, Stein Y. Coffee, tea, and plasma cholesterol: The Jerusalem Lipid Research Clinic prevalence study, Br Med J. 1985;291:699-704. 28. Grobbee DE, Rimm EB, Giovannucci E, et al. Coffee, caffeine and cardiovascular disease in men, N Engl J Med. 1990;323:1026-32. 29. Dawber TR, Kannel WB, Gordon T. Coffee and cardiovascular disease: Observations from the Framingham Study, N Engl J Med. 1974; 291:871-4. 30. Kovan MC, Fulwood R, Feinleib N Engl J Med. 1983;309:1249.
M. Coffee and cholesterol
(letter),
31. Shekelle RB, Gale M, Paul 0, Stamler J. Coffee and cholesterol N Engl J Med. 1983;309:1249-50.
(letter),
H. Coffee consumption and serum study in Belgium, Int J Epidemiol.
35. Finocchiaro G, Pezzana A, Pernigotti L, Bo M. The influence of coffee on plasma lipids. In: Thelle DC, van der Stegen, eds. Coffee and Coronary Heart Disease. The Nordic School of Public Health. 1990: 101-5. 36. Gilbert RM. Caffeine as a drug abuse. In: Gibbins RJ, Israel Y, Kalant H, Popham RE, Schmidt W, Smart RG, eds. Research Advances in Alcohol and Drug Problems. v. 3. John Wiley and Sons; 1976:49178. 37. Roeckel I. Coffee 1248-178.
and cholesterol
(letter).
N Engl J Med.
1983;309:
38. Keys A. Seven Countries. A Multivariate Analysis of Death and Coronary Heart Disease. Cambridge, MA: Harvard University Press; 1980. 39. Heyden S, Heiss G, Manegold C, et al. The combined effect of smoking and coffee drinking on LDL and HDL-cholesterol, Circulation. 1979; 60:22-5. 40. Panic0 S, Celentano E, Krogh V. Coffee and blood lipids in Italy. In: Thelle DC, van der Stegen, eds. Coffee and Coronary Heart Disease. The Nordic School of Public Health. 1990:63-70. 41. Whitsett TL, Manion CV, Christensen HD. Cardiovascular coffee and caffeine, Am J Cardiol. 1984;53:918-22.
effect of
42. Robertson D, Froelich JC, Carr RK, et al. Effect of caffeine on plasma renin activity, catecholamine and blood pressure, N Engl J Med. 1978; 298:181. 43. Bjelke E. Colon
cancer
and blood cholesterol,
Lancet.
1974;1:1116.
44. Brown CR, Benowitz NL. Caffeine and cigarette smoking: Behavioral, cardiovascular and metabolic interactions, Pharmacol Biochem Behav. 1989;34:565-70. 45. Naismith DJ, Akinyanju PA, Szanto S, Yudkin J. The effect in volunteers of coffee and decaffeinated coffee on blood glucose, insulin, plasma lipids, and some factors involved in blood clotting, Nutr Metab. 1970;12:144-51.