Content uploaded by Andrea Discacciati
Author content
All content in this area was uploaded by Andrea Discacciati on Nov 21, 2015
Content may be subject to copyright.
Systematic Reviews and Meta- and Pooled Analyses
Coffee Consumption and Mortality From All Causes, Cardiovascular Disease, and
Cancer: A Dose-Response Meta-Analysis
Alessio Crippa*, Andrea Discacciati, Susanna C. Larsson, Alicja Wolk, and Nicola Orsini
*Correspondence to Alessio Crippa, Institute of Environmental Medicine, Karolinska Institutet, Box 210, SE-171 77 Stockholm, Sweden
(e-mail: alessio.crippa@ki.se).
Initially submitted March 11, 2014; accepted for publication July 1, 2014.
Several studies have analyzed the relationship between coffee consumption and mortality, but the shape of the
association remains unclear. We conducted a dose-response meta-analysis of prospective studies to examine the
dose-response associations between coffee consumption and mortality from all causes, cardiovascular disease
(CVD), and all cancers. Pertinent studies, published between 1966 and 2013, were identified by searching PubMed
and by reviewing the reference lists of the selected articles. Prospective studies in which investigators reported rel-
ative risks of mortality from all causes, CVD, and all cancers for 3 or more categories of coffee consumption were
eligible. Results from individual studies were pooled using a random-effects model. Twenty-one prospective studies,
with 121,915 deaths and 997,464 participants, met the inclusion criteria. There was strong evidence of nonlinear
associations between coffee consumption and mortality for all causes and CVD (Pfor nonlinearity < 0.001). The
largest risk reductions were observed for 4 cups/day for all-cause mortality (16%, 95% confidence interval: 13, 18)
and 3 cups/day for CVD mortality (21%, 95% confidence interval: 16, 26). Coffee consumption was not associated
with cancer mortality. Findings from this meta-analysis indicate that coffee consumption is inversely associated with
all-cause and CVD mortality.
all-cause mortality; cancer mortality; cardiovascular disease mortality; coffee; dose-response relationship; meta-
analysis; prospective studies
Abbreviations: CI, confidence interval; CVD, cardiovascular disease; CYP1A2, cytochrome P-450 1A2 gene.
Coffee is one of the most commonly consumed beverages
around the world. Because of its popularity, even small health
effects could have important public health consequences.
Coffee has been considered potentially unhealthy, since caf-
feine intake has been positively associated with blood pres-
sure (1), serum lipid concentration (2), cholesterol levels
(3), and insulin resistance (4). Prospective studies, however,
have generally not supported adverse health effects associ-
ated with coffee consumption. Besides caffeine, coffee con-
tains several bioactive compounds with potentially beneficial
properties, such as insulin-sensitizing (5) and antiinflamma-
tory (6) effects.
Several quantitative reviews have indicated that coffee con-
sumption may decrease the incidence of common chronic
diseases, including type 2 diabetes (7), heart disease (8), and
specific types of cancers (9). Only 2 meta-analyses have
examined the association between coffee and mortality, pool-
ing relative risks for the highest category of coffee consump-
tion versus the lowest (10,11). A weak inverse association was
found for all-cause mortality (10,11), while an unclear associ-
ation was found for cardiovascular disease (CVD) and no as-
sociation was found for cancer mortality (10). A dose-response
analysis uses all of the exposure-disease information, includ-
ing intermediate categories, and is therefore more efficient
than the highest-versus-lowest approach. In addition, it is
less sensitive to the variability of exposure categories and
more flexible in modeling the relationship under investigation.
In particular, it provides a detailed description of the risk of
death throughout the observed range of exposure, thus allow-
ing identification of those values associated with the highest
or lowest risk. Because the shapes of the associations remain
uncertain, we conducted a dose-response meta-analysis of
763 Am J Epidemiol. 2014;180(8):763–775
American Journal of Epidemiology
© The Author 2014. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of
Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
Vol. 180, No. 8
DOI: 10.1093/aje/kwu194
Advance Access publication:
August 24, 2014
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
prospective studies on coffee and mortality to examine the
exposure-disease associations between coffee consumption
and mortality from all causes, CVD, and cancer.
METHODS
Literature search and selection
We performed a literature search in the PubMed database
for articles published from January 1966 through December
2013, using the terms “(prospective or cohort) and (fatal
or death or mortality) and (hot beverages or coffee or
caffeine).”The search was limited to studies carried out in
humans. We followed the Meta-Analysis of Observational
Studies in Epidemiology (MOOSE) guidelines for conduct-
ing meta-analyses and reporting results (12). Two authors
(A.C., A.D.) separately retrieved the studies reporting data
on the association between coffee consumption and all-cause
mortality, as well as associations for CVD and all-cancer
mortality. Discrepancies were discussed and resolved.
Studies were eligible for inclusion in the meta-analysis if
they met the following criteria: 1) the study had a prospective
design; 2) the exposure of interest was coffee consumption;
3) the outcome was all-cause mortality, CVD mortality, and/
227 Records Identified Through PubMed
Database Search (1966–2013)
136 Records Excluded Because Title
and/or Abstract Not Relevant
91 Records Assessed for Eligibility
32 Articles Eligible for Inclusion in the
Meta-Analysis
61 Articles Excluded (Reviews, Different
Outcome, Risk Estimates Not Reported,
Special Population Analyzed)
2 Additional Articles Identified From
Manual Searches
11 Articles Excluded for Not Satisfying the
Inclusion Criteria
2 duplicate reports on same population
5 with results not adjusted for smoking
1 report on caffeinated beverages
3 reporting only 1 nonreference category
Cardiovascular
Disease Mortality:
16 Studies
Overall Mortality:
18 Studies
Cancer Mortality:
9 Studies
21 Studies Included in the Meta-Analysis
3 Included Only in the Sensitivity Analysis
Figure 1. Selection of studies for inclusion in a meta-analysis of coffee consumption and mortality from all causes, cardiovascular disease, and all
cancers, 1966–2013.
764 Crippa et al.
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Table 1. Characteristics of Prospective Studies Included in a Meta-Analysis of Coffee Consumption and Mortality From All Causes, Cardiovascular Disease, and Cancer, 1966–2013
First Author,
Year
(Reference No.)
Study Name Country
No. of Cases No. of Noncases
Years of Enrollment
Duration of
Follow-up,
years
Beverage Type Cause of
Death Adjustment Variables
Men Women Total Men Women Total
LeGrady,
1987 (23)
Chicago Western
Electric Company
Study
United States 452 1,910 1959–1978 19 Coffee All causes;
CHD;
all other
causes
Age, smoking, blood pressure,
and cholesterol
Rosengren,
1991 (24)
Multifactor Primary
Prevention Trial
Sweden 478 6,765 1974/1977–1983 7.1 Coffee All causes;
CHD;
cancer;
all other
causes
Age, smoking, alcohol abuse,
occupational class, BMI,
physical activity, blood
pressure, diabetes, family
history of myocardial
infarction, and mental stress
Klatsky,
1993 (25)
Northern California
Kaiser
Permanente
Medical Care
Program
United States 2,695 1,806 128,934 1978/1985–1988 8 Coffee +
decaffeinated
coffee
All causes Age, sex, race, smoking, alcohol,
education, marital status, and
BMI
Hart,
1997 (41)
Scotland,
United
Kingdom
625 5,766 1970/1973–1994 21 Coffee CHD Age, smoking, social class, age
upon leaving full-time
education, BMI, diastolic blood
pressure, cholesterol, angina,
and electrocardiographic
ischemia
Woodward,
1999 (26)
Scottish Heart
Health Study
Scotland,
United
Kingdom
372 201 5,754 5,875 1984/1987–1993 7.7 Coffee All causes;
CHD
Age, smoking, cotinine, alcohol,
housing tenure, BMI, physical
activity, blood pressure,
fibrinogen, cholesterol,
high-density lipoprotein
cholesterol, triglycerides,
vitamin C, tea drinking, and
Bortner score (57)
Kleemola,
2000 (27)
Finland 1,201 444 10,075 10,387 1972/1977–1982 10 Coffee All causes;
CHD
Age, smoking, education, BMI,
cholesterol, blood pressure,
and history of acute
myocardial infarction
Iwai,
2002 (28)
Japan 246 115 1,404 1,451 1989–1999 9.9 Coffee All causes;
apoplexy;
cancer
Age, education (age at final
graduation), physical activity,
and history of selected
diseases
Jazbec,
2003 (29)
Croatia 568 382 1,571 1,739 1972–1999 Coffee All causes;
CVD
Age, residence, smoking,
diastolic blood pressure,
history of gastric/duodenal
ulcer, and feeling of well-being
Andersen,
2006 (30)
Iowa Women’s
Health Study
United States 4,265 27,312 1986–2001 15 Coffee;
decaffeinated
coffee
All causes;
CVD;
cancer;
all other
causes
Age; smoking; alcohol;
education; BMI; waist:hip ratio;
physical activity; intake of
whole grain, refined grain, red
meat, fish and seafood, fruit,
vegetables, and energy; use of
estrogens; and use of
multivitamin supplements
Paganini-Hill,
2007 (31)
Leisure World
Cohort Study
United States 11,386 4,980 8,644 1981–2004 23 Coffee;
decaffeinated
coffee
All causes Age, sex, smoking, alcohol, BMI,
physical activity, hypertension,
diabetes, and history of
angina, heart attack, stroke,
rheumatoid arthritis, and
cancer
Table continues
Coffee and Mortality 765
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Table 1. Continued
First Author,
Year
(Reference No.)
Study Name Country
No. of Cases No. of Noncases
Years of Enrollment
Duration of
Follow-up,
years
Beverage Type Cause of
Death Adjustment Variables
Men Women Total Men Women Total
Lopez-Garcia,
2008 (32)
Nurses’Health
Study; Health
Professionals
Follow-up Study
United States 6,888 11,095 41,736 86,214 HPFS: 1986–2004;
NHS: 1980–2004
HPFS: 18;
NHS: 24
Caffeinated
coffee;
decaffeinated
coffee
All causes;
CVD;
cancer;
all other
causes
Age; smoking; alcohol; BMI;
physical activity; intake of n-3,
polyunsaturated, saturated,
and trans-fats, folic acid, and
total energy; glycemic load;
use of multivitamin and vitamin
E supplements; and family
history of myocardial
infarction. For women, also
menopausal status and
hormone replacement therapy.
Happonen,
2008 (33)
Finland 251 372 311 506 1991/1992–2005 14.5 Coffee All causes;
cancer;
CVD;
all other
causes
Age, sex, calendar year,
smoking, education, previous
occupation, marital status,
BMI, diabetes, history of acute
myocardial infarction, physical
disability, cognitive
impairment, and self-rated
health
Ahmed,
2009 (34)
Cohort of Swedish
Men
Sweden 37,315 1998–2006 9 Coffee All causes Age, smoking, alcohol,
education, marital status, BMI,
physical activity, tea drinking,
fat intake, sodium intake,
cholesterol, aspirin, and family
history of myocardial infarction
de Koning
Gans,
2010 (35)
EPIC-NL
(Prospect-EPIC
and
MORGEN-EPIC
cohorts)
The
Netherlands
1,405 37,514 1993/1997–2006 13 Coffee All causes;
CHD;
stroke
Age; sex; study cohort; smoking;
alcohol; education; waist
circumference; physical
activity; intake of tea, saturated
fat, fiber, vitamin C, total fluids,
and total energy;
hypertension; diabetes; and
hypercholesterolemia. For
women, also menopausal
status and hormone
replacement therapy.
Sugiyama,
2010 (36)
Miyagi Cohort
Study
Japan 1,647 807 18,287 19,455 1990–2001 10.3 Coffee All causes;
CVD;
CHD;
stroke;
all other
causes
Age; sex; smoking; alcohol;
education; BMI; walking time;
intake of green tea, oolong tea,
black tea, rice, miso soup,
meat, dairy products, fish,
vegetables, fruits, and energy;
hypertension; and diabetes
Leurs,
2010 (42)
Netherlands Cohort
Study
The
Netherlands
1,669 828 58,279 62,573 1986–1996 10 Coffee Ischemic
heart
disease;
stroke
Age, smoking and number of
cigarettes smoked per day,
and total energy intake
Tamakoshi,
2011 (37)
Japan Collaborative
Cohort Study for
Evaluation of
Cancer Risk
Japan 11,178 8,354 40,672 57,081 1988/1990–2006 16 Coffee All causes;
cancer
Age; smoking; alcohol;
education; marital status; BMI;
daily walking; intake of green
leafy vegetables and green
tea; sleep length; stress; and
history of cancer, myocardial
infarction, and stroke
Table continues
766 Crippa et al.
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Table 1. Continued
First Author,
Year
(Reference No.)
Study Name Country
No. of Cases No. of Noncases
Years of Enrollment
Duration of
Follow-up,
years
Beverage Type Cause of
Death Adjustment Variables
Men Women Total Men Women Total
Mineharu,
2011 (43)
Japan Collaborative
Cohort Study for
Evaluation of
Cancer Risk
Japan 1,681 1,436 34,345 48,310 1988–2003 13.1 Coffee CVD; CHD;
stroke
Age; smoking; alcohol;
education; BMI; walking hours;
sports participation; intake of
fruit, vegetables, beans, meat,
fish, seaweed, and energy;
multivitamin and vitamin E
supplement use;
hypertension; diabetes; and
mental stress
Freedman,
2012 (38)
NIH-AARP Diet and
Health Study
United States 33,731 18,784 229,119 173,141 1995–2008 13.6 Coffee All causes;
cancer;
heart
disease;
stroke
Age; race; smoking; alcohol;
education; marital status; BMI;
physical activity; intake of
fruits, vegetables, meat,
saturated fats, and total
energy; use of vitamin
supplements; health status;
and diabetes. For cancers,
also family history of cancer.
For women, also hormone
replacement therapy.
Liu,
2013 (39)
Aerobics Center
Longitudinal
Study
United States 2,198 314 33,900 9,827 1971–2002 17 Coffee All causes;
CVD
Age, baseline examination year,
decaffeinated coffee use,
regular tea use, decaffeinated
or herbal tea use, physical
inactivity, BMI, smoking,
alcohol consumption,
diabetes, hypertension,
hypercholesterolemia,
parental history of CVD, and
cardiorespiratory fitness
Gardener,
2013 (40)
Northern Manhattan
Study
United States 863 2,461 1993–2001 11 Caffeinated
coffee;
decaffeinated
coffee
Age, sex, race/ethnicity,
education, demographic
factors, smoking, behavioral
risk factors, diet, BMI, previous
cardiac disease, diabetes,
hypertension, and
hypercholesterolemia
Abbreviations: BMI, body mass index; CHD, coronary heart disease; CVD, cardiovascular disease; EPIC, European Prospective Investigation into Cancer and Nutrition; EPIC-NL, EPIC Netherlands; HPFS, Health
Professionals Follow-up Study; MORGEN, Monitoring Project on Risk Factors for Chronic Diseases; NHS, Nurses’Health Study; NIH, National Institutes of Health.
Coffee and Mortality 767
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
or all-cancer mortality; 4) the investigators reported relative
risks with 95% confidence intervals for 3 or more quantitative
categories of coffee consumption; and 5) the reported relative
risks had been adjusted at least for smoking status.
Data extraction
The following information was extracted from each study:
first author’s surname, publication year, study location, study
period, duration of follow-up (years), sex, number of subjects
(total number of deaths and total cohort size or total number of
deaths and person-years of follow-up), mortality outcomes,
coffee consumption categories, type of coffee, covariates ad-
justed for in the multivariable analysis, and relative risks
(with their 95% confidence intervals) for all categories of cof-
fee consumption. We extracted the relative risks that reflected
the greatest degree of adjustment for potentially confounding
variables. If investigators reported the adjusted relative risks
but not the corresponding confidence intervals, we calculated
the confidence intervals for the crude relative risks and related
them to the adjusted relative risks. For studies that presented
data separately on both coronary heart disease and stroke, we
combined the results as indicated by Hamling et al. (13).
For each study, the median or mean coffee consumption
within each exposure interval was assigned the correspond-
ing relative risk. When median or mean consumption per cat-
egory was not reported, we assigned the midpoint of the
upper and lower boundaries for each category as the average
consumption. If the upper bound for the highest category was
not provided, we assumed that the category had the same am-
plitude as the adjacent one.
Statistical analysis
We performed a 2-stage random-effects dose-response
meta-analysis to examine a potential nonlinear relationship
between coffee consumption and 3 different outcomes: all-
cause mortality, CVD mortality, and cancer mortality (14,
15). This was done by modeling coffee consumption using re-
stricted cubic splines with 3 knots at fixed percentiles (25%,
50%, and 75%) of the distribution (15). In the first stage, a re-
stricted cubic spline model with 2 spline transformations (3
knots minus 1) was fitted taking into account the correlation
within each set of published relative risks (14,15). In the sec-
ond stage, we combined the 2 regression coefficients and the
variance/covariance matrices that had been estimated within
each study, using the multivariate extension of the method
of moments in a multivariate random-effects meta-analysis
(16). We calculated an overall Pvalue by testing that the 2 re-
gression coefficients were simultaneously equal to zero. We
calculated a Pvalue for nonlinearity by testing that the coeffi-
cient of the second spline was equal to zero (17).
We excluded from the main analysis those studies that did
not report the number of subjects (total number of deaths and
total cohort size or total number of deaths and person-years
of follow-up) in order to avoid biases in the estimates for the
variances (15). We considered the excluded studies in a sen-
sitivity analysis.
We performed stratified analysis by study location, sex, type
of smoking adjustment (smoking status, categories of cigarette
smoking, or number of cigarettes smoked per day (continuous
variable)), and alcohol adjustment. Statistical heterogeneity
among studies was assessed using the χ
2
test and was defined
as a Pvalue lessthan 0.10. Statistical heterogeneity was further
quantified through the multivariate generalization of the I
2
sta-
tistic (18). Low heterogeneity is defined by I
2
values less than
25%, while values greater than 75% are indicative of high het-
erogeneity. Publication bias was assessed with Egger’sregres-
sion test (19). All statistical analyses were conducted with the
dosresmeta (20) and metafor (21) packages in R (R Foundation
for Statistical Computing, Vienna, Austria) (22). Pvalues less
than 0.05 were considered statistically significant.
RESULTS
Study characteristics
The search strategy identified 227 articles on humans, 136
of which were excluded after review of the title or abstract
(Figure 1). Of the 91 publications selected, 61 were not in-
cluded, for at least one of the following reasons: 1) the article
did not report original results from the study (9 articles);
2) the article did not provide relative risks and corresponding
confidence intervals (10 articles); 3) disease incidence and
mortality were combined (18 articles); 4) the study analyzed
subpopulations (e.g., persons with diabetes or hypertension)
(8 articles); and 5) the study investigated relationships with
specific types of cancer (22 articles). The reference lists of
the remaining 30 articles were checked to obtain other perti-
nent publications, and 2 additional reports were identified.
We further excluded 11 studies: 2 represented duplicate
publication; 5 did not adjust for smoking; 1 considered only
total caffeine intake; and 3 analyzed only 2 coffee consump-
tion categories. The Web Appendix (available at http://aje.
oxfordjournals.org/) details the reasons for exclusion of indi-
vidual studies.
02468
Coffee Consumption, cups/day
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
1.10
Relative Risk
Figure 2. Pooled dose-response association between coffee con-
sumption and all-cause mortality (solid line) in a meta-analysis,
1966–2013. Coffee consumption was modeled with restricted cubic
splines in a multivariate random-effects dose-response model. The
relative risks are plotted on the log scale. Dashed lines represent
the 95% confidence intervals for the spline model. No coffee con-
sumption (0 cups/day) served as the referent group.
768 Crippa et al.
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Table 2. Adjusted Relative Risk of All-Cause Mortality According to Coffee Consumption (Versus No Consumption) in Prospective Studies, by Study Location, Sex, Type of Smoking
Adjustment, and Alcohol Use, 1966–2013
Coffee Consumption, cups/day
1234567 8
RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI
All studies (n= 15) 0.92 0.91, 0.94 0.87 0.84, 0.90 0.85 0.82, 0.88 0.84 0.82, 0.87 0.85 0.83, 0.87 0.86 0.83, 0.88 0.86 0.83, 0.90 0.87 0.83, 0.92
Inclusion of 3 excluded articles
a
(n= 18) 0.91 0.88, 0.94 0.84 0.80, 0.89 0.82 0.78, 0.87 0.82 0.78, 0.86 0.84 0.80, 0.87 0.85 0.81, 0.89 0.86 0.81, 0.92 0.88 0.81, 0.95
Study location
Europe (n= 6) 0.90 0.86, 0.95 0.83 0.76, 0.91 0.79 0.71, 0.88 0.78 0.70, 0.86 0.78 0.71, 0.86 0.78 0.70, 0.86 0.77 0.70, 0.86 0.77 0.69, 0.87
United States (n= 6) 0.94 0.92, 0.96 0.89 0.85, 0.92 0.87 0.83, 0.91 0.86 0.83, 0.90 0.87 0.83, 0.91 0.87 0.83, 0.92 0.88 0.82, 0.93 0.88 0.82, 0.95
Japan (n= 3) 0.85 0.76, 0.94 0.75 0.64, 0.88 0.75 0.67, 0.85 0.82 0.75, 0.90 0.92 0.73, 1.14
Sex
Men (n= 10) 0.91 0.87, 0.95 0.85 0.79, 0.91 0.83 0.77, 0.89 0.83 0.79, 0.88 0.86 0.83, 0.89 0.88 0.84, 0.91 0.90 0.85, 0.96 0.92 0.84, 1.01
Women (n= 8) 0.91 0.89, 0.93 0.85 0.82, 0.88 0.82 0.79, 0.84 0.81 0.78, 0.83 0.81 0.76, 0.85 0.80 0.74, 0.88 0.80 0.71, 0.91 0.80 0.69, 0.94
Both sexes (n= 4) 0.96 0.93, 0.98 0.92 0.88, 0.96 0.90 0.86, 0.95 0.90 0.85, 0.95 0.90 0.84, 0.97 0.90 0.82, 1.00 0.91 0.80, 1.03 0.91 0.77, 1.06
Type of smoking adjustment
b
Status (n= 12) 0.90 0.87, 0.94 0.83 0.77, 0.88 0.80 0.74, 0.86 0.80 0.74, 0.86 0.81 0.74, 0.87 0.81 0.74, 0.89 0.82 0.74, 0.92 0.83 0.73, 0.95
Categories (n= 7) 0.90 0.85, 0.96 0.83 0.74, 0.93 0.80 0.72, 0.89 0.80 0.75, 0.85 0.81 0.77, 0.85 0.82 0.75, 0.90 0.83 0.71, 0.97
Continuous (n= 3) 0.94 0.92, 0.96 0.90 0.86, 0.94 0.88 0.83, 0.93 0.87 0.82, 0.93 0.87 0.81, 0.94 0.87 0.80, 0.95 0.87 0.79, 0.96 0.87 0.78, 0.97
Alcohol adjustment
No (n= 4) 0.82 0.69, 0.97 0.70 0.52, 0.94 0.66 0.49, 0.89 0.68 0.54, 0.85 0.71 0.60, 0.84 0.74 0.61, 0.90 0.78 0.59, 1.02 0.81 0.56, 1.19
Yes (n= 11) 0.93 0.91, 0.94 0.87 0.84, 0.90 0.85 0.82, 0.88 0.84 0.81, 0.87 0.85 0.82, 0.88 0.85 0.82, 0.88 0.85 0.82, 0.89 0.86 0.82, 0.89
Abbreviations: CI, confidence interval; RR, relative risk.
a
Includes 3 studies that did not provide information about the number of subjects.
b
Smoking status, category of cigarette smoking, or number of cigarettes smoked per day (continuous variable).
Coffee and Mortality 769
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Thus, the meta-analysis included 21 independent prospec-
tive studies, the main characteristics of which are described in
Table 1. Eighteen studies provided estimates for all-cause
mortality (23–40), 16 provided estimates for CVD mortality
(23,24,26,27,29,30,32,33,35,36,38–43), and 9 provided
estimates for all-cancer mortality (24,28,30,32,33,36–38,
40). Three studies (23,24,26) did not provide confidence in-
tervals for the adjusted relative risks but reported sufficient
data to back-calculate them. Three studies (35,38,42)pro-
vided results for coronary heart disease and stroke mortality
separately. Three studies (29,39,40) did not report infor-
mation about the distribution of cases and noncases across
exposure levels, and therefore they were included only in
the sensitivity analysis.
Combined, these studies included 121,915 deaths and
997,464 study participants. Nine studies were conducted in
Europe, 8 in the United States, and 4 in Japan (Table 1).
One study considered only elderly people (33), while the re-
maining studies included persons from the general popula-
tion. All of the studies but 5 included male and female
participants, but only 11 reported sex-specific results. The in-
cluded studies provided relative risk estimates adjusted for
age (all 21 studies), body mass index (15 studies), alcohol
consumption (14 studies), hypertension or blood pressure
(11 studies), physical activity (11 studies), and history of di-
abetes (8 studies).
Association between coffee consumption and all-cause
mortality
We found strong evidence of a nonlinear association
between coffee consumption and all-cause mortality (overall
P<0.001; Pfor nonlinearity < 0.001) based on 15 studies
(Figure 2). Compared with no coffee consumption, the pooled
relative risks for all-cause mortality were 0.92 (95% confi-
dence interval (CI): 0.91, 0.94) for 1 cup/day, 0.87 (95%
CI: 0.84, 0.90) for 2 cups/day, 0.85 (95% CI: 0.82, 0.88)
for 3 cups/day, 0.84 (95% CI: 0.82, 0.87) for 4 cups/day,
and 0.86 (95% CI: 0.83, 0.88) for 6 cups/day. There
was between-study heterogeneity (I
2
= 58.1%; P< 0.001).
Egger’s regression test provided no evidence of substantial
publication bias (P= 0.26).
The associations were similar for men and women (Pfor
heterogeneity = 0.19), although at high levels of coffee con-
sumption the inverse association was more pronounced in
women. Moreover, the associations were similar across strata
of type of smoking adjustment (Pfor heterogeneity = 0.99)
and alcohol adjustment (Pfor heterogeneity = 0.13). There
was evidence of differences according to geographical region
(Pfor heterogeneity < 0.001); in particular, the inverse rela-
tionship was slightly stronger among studies conducted in
Europe than among those conducted in the United States.
In the 3 studies conducted in Japan, the association was sta-
tistically significant only for moderate coffee consumption
(<4 cups/day) (Table 2).
Association between coffee consumption and CVD
mortality
Similar to all-cause mortality, we found strong evidence
of a nonlinear association between coffee consumption
and CVD mortality (overall P< 0.001; Pfor nonlinearity <
0.001) based on 13 studies (Figure 3). Compared with no cof-
fee consumption, the pooled relative risks of CVD mortality
were 0.89 (95% CI: 0.86, 0.91) for 1 cup/day, 0.81 (95% CI:
0.77, 0.85) for 2 cups/day, 0.79 (95% CI: 0.74, 0.84) for 3
cups/day, 0.80 (95% CI: 0.74, 0.86) for 4 cups/day, and
0.85 (95% CI: 0.75, 0.95) for 6 cups/day. There was evidence
of moderate between-study heterogeneity (I
2
= 58.8%; P<
0.001). Egger’s regression test provided no evidence of sub-
stantial publication bias (P= 0.29).
No relevant differences were found by sex (Pfor hetero-
geneity = 0.60) or alcohol adjustment (Pfor heterogeneity =
0.99) (Table 3). Differences were found for geographical region
(Pfor heterogeneity < 0.001); in particular, studies conducted in
Japanshowedaninverseassociationonlyforlowcoffeecon-
sumption (2 cups/day), while studies conducted in Europe
and the United States provided similar results. The associations
differed across strata of type of smoking adjustment, with non–
statistically significant results for studies that adjusted only for
smoking status.
Association between coffee consumption and cancer
mortality
Coffee consumption was not statistically significantly
associated with cancer mortality (overall P= 0.07; Pnon-
linearity = 0.06) based on 8 studies (Figure 4). Compared
with no coffee consumption, the pooled relative risks for
total cancer mortality were 0.98 (95% CI: 0.96, 1.01) for 1
cup/day, 0.97 (95% CI: 0.93, 1.01) for 2 cups/day, 0.98
(95% CI: 0.93, 1.02) for 3 cups/day, 0.99 (95% CI: 0.95,
1.03) for 4 cups/day, and 1.03 (95% CI: 0.99, 1.08) for 6
cups/day. There was low between-study heterogeneity (I
2
=
1%; P= 0.45). Egger’s regression test provided no evidence
of publication bias (P= 0.52).
02468
Coffee Consumption, cups/day
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
1.10
Relative Risk
Figure 3. Pooled dose-response association between coffee
consumption and cardiovascular disease mortality (solid line) in a
meta-analysis, 1966–2013. Coffee consumption was modeled with re-
stricted cubic splines in a multivariate random-effects dose-response
model. The relative risks are plotted on the log scale. Dashed lines
represent the 95% confidence intervals for the spline model. No coffee
consumption (0 cups/day) served as the referent group.
770 Crippa et al.
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Table 3. Adjusted Relative Risk of Cardiovascular Disease Mortality According to Coffee Consumption (Versus No Consumption) in Prospective Studies, by Study Location, Sex, Type of
Smoking Adjustment, and Alcohol Use, 1966–2013
Coffee Consumption, cups/day
1234567 8
RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI
All studies (n= 13) 0.89 0.86, 0.91 0.81 0.77, 0.85 0.79 0.74, 0.84 0.80 0.74, 0.86 0.82 0.75, 0.90 0.85 0.75, 0.95 0.87 0.76, 1.00 0.90 0.76, 1.06
Inclusion of 3 excluded articles
a
(n= 16) 0.88 0.83, 0.94 0.80 0.72, 0.90 0.78 0.69, 0.88 0.80 0.70, 0.90 0.83 0.73, 0.93 0.86 0.75, 0.98 0.89 0.77, 1.04 0.93 0.78, 1.10
Study location
Europe (n= 7) 0.89 0.83, 0.95 0.81 0.72, 0.92 0.78 0.68, 0.90 0.78 0.68, 0.91 0.80 0.68, 0.93 0.81 0.69, 0.97 0.83 0.68, 1.01 0.84 0.68, 1.05
United States (n= 4) 0.88 0.84, 0.91 0.79 0.73, 0.85 0.76 0.71, 0.81 0.76 0.72, 0.81 0.78 0.72, 0.85 0.81 0.72, 0.91 0.83 0.70, 0.97
Japan (n= 2) 0.82 0.72, 0.93 0.75 0.60, 0.94 0.91 0.62, 1.35 1.36 0.66, 2.80
Sex
Men (n= 10) 0.90 0.86, 0.94 0.83 0.76, 0.90 0.81 0.74, 0.90 0.83 0.75, 0.92 0.87 0.77, 0.98 0.90 0.78, 1.04 0.94 0.78, 1.12 0.98 0.79, 1.21
Women (n= 8) 0.88 0.84, 0.92 0.79 0.73, 0.86 0.76 0.67, 0.85 0.76 0.65, 0.88 0.77 0.63, 0.94 0.78 0.61, 1.00 0.79 0.59, 1.06 0.80 0.57, 1.13
Both sexes (n= 2) 1.03 0.85, 1.23 1.04 0.75, 1.44 1.03 0.71, 1.49 1.00 0.70, 1.42 0.96 0.69, 1.34 0.93 0.66, 1.30 0.89 0.62, 1.28
Type of smoking adjustment
b
Status (n= 11) 0.92 0.86, 0.99 0.87 0.77, 0.99 0.87 0.74, 1.03 0.90 0.74, 1.09 0.95 0.75, 1.20 1.00 0.75, 1.32 1.05 0.75, 1.47 1.10 0.74, 1.64
Categories (n= 5) 0.88 0.79, 0.96 0.78 0.67, 0.92 0.75 0.63, 0.88 0.74 0.65, 0.84 0.74 0.65, 0.85 0.75 0.62, 0.90 0.75 0.58, 0.98
Continuous (n= 4) 0.87 0.85, 0.90 0.78 0.75, 0.82 0.75 0.71, 0.79 0.76 0.72, 0.80 0.78 0.72, 0.84 0.80 0.72, 0.88 0.82 0.72, 0.93
Alcohol adjustment
No (n= 5) 0.89 0.81, 0.97 0.81 0.68, 0.95 0.78 0.64, 0.96 0.79 0.64, 0.98 0.82 0.66, 1.02 0.84 0.66, 1.07 0.87 0.66, 1.13 0.89 0.66, 1.21
Yes (n= 8) 0.89 0.86, 0.92 0.81 0.76, 0.86 0.78 0.73, 0.84 0.79 0.72, 0.86 0.81 0.72, 0.91 0.84 0.72, 0.97 0.86 0.72, 1.03 0.88 0.71, 1.10
Abbreviations: CI, confidence interval; RR, relative risk.
a
Includes 3 studies that did not provide information about the number of subjects.
b
Smoking status, category of cigarette smoking, or number of cigarettes smoked per day (continuous variable).
Coffee and Mortality 771
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Subgroup analysis was limited by the small number of
studies. We found weak suggestions of differences according
to sex (Pfor heterogeneity = 0.11); in particular, results for
men suggested a more pronounced positive association, al-
though the finding was significant only for high consumption
(≥5 cups/day). Results were similaracross geographical areas
(Pfor heterogeneity = 0.39), by type of smoking adjustment
(Pfor heterogeneity = 0.76), and by adjustment for alcohol
intake (Pfor heterogeneity = 0.87).
Sensitivity analysis
Inclusion of the 3 studies that did not report information
about the number of subjects by category of coffee consump-
tion did not materially change the results (Tables 2–4). We
obtained similar results when we removed data points
above 6 cups of coffee per day, and there was still evidence
of nonlinearity (P< 0.001) for the associations between
coffee consumption and all-cause and CVD mortality. For
cancer mortality, the association was still not statistically sig-
nificant (overall P= 0.26).
DISCUSSION
Findings from the current meta-analysis, including 21
prospective studies, indicate that coffee consumption may
be inversely associated with all-cause and CVD mortality.
Nonlinear dose-response relationships were found, with the
strongest association being observed for 4 cups/day for all-
cause mortality (16% lower risk) and 3 cups/day for CVD
mortality (21% lower risk). The results for the association
between coffee consumption and all-cause mortality were
comparable to those reported in the previous 2 published
meta-analyses (10,11), where no further risk reduction was
observed for high coffee consumption (≥4 cups/day) as com-
pared with moderate coffee consumption (2–4 cups/day). No
association between coffee consumption and cancer mortal-
ity was found. It is still unclear how coffee consumption may
have an effect on mortality. Coffee is a composite brew with
several bioactive compounds whose health effects are contra-
dictory. In the past, coffee consumption was considered un-
healthy because of its content of caffeine, which has been
related to increased blood pressure (4), insulin resistance
(44), and serum lipid concentration (2). Nonetheless, habitual
coffee consumers seem to develop a partial tolerance to the
acute effect of caffeine (45). In addition, other bioactive com-
pounds besides caffeine may play an important role; indeed,
the phenolic compounds make coffee a major source of anti-
oxidants, with potential beneficial health effects (46). More-
over, several epidemiologic studies have indicated an inverse
relationship between coffee consumption and risk of suicide
(47), Parkinson’s disease (48), and gallstones (49). Further
studies have found that coffee consumption is inversely asso-
ciated with specific markers of inflammation (50) that are re-
sponsible for progression of atherosclerosis (51), coronary
heart disease (52), and cancer (53). Cardiovascular disease
and cancer are some of the most frequent causes of mortality;
thus, the beneficial compounds in coffee may lead to a reduc-
tion in mortality by slowing the progression of disease.
Strengths of this meta-analysis include the dose-response
analysis, which provides a comprehensive description of the
shape of the studied association. Another strength is the pro-
spective design of the included studies, which should have
eliminated potential selection and recall biases that can affect
the results of retrospective case-control studies. The relatively
large total number of cases provided high statistical power,
which contributes to stable risk estimates. The large number
of studies enabled us to conduct several subgroup analyses to
assess potential sources of heterogeneity. Lastly, we did not
find evidence of publication bias, which could affect the re-
sults of a meta-analysis.
Our meta-analysis also had several potential limitations.
First, the observational design of the studies did not exclude
the presence of residual or unmeasured confounding from
other mortality risk factors. However, all of the included stud-
ies adjusted for smoking and age, and some of them (11 out
of 21) also adjusted for other important confounders, includ-
ing body mass index, alcohol intake, and physical activity. In
particular, the analysis stratified according to type of smoking
adjustment indicated the presence of some residual con-
founding for the association between coffee consumption
and CVD mortality, suggesting that the relationship may be
even stronger, especially for high coffee consumption. An-
other potential limitation is misclassification of exposure,
which was inevitable since coffee consumption was self-
reported. Nevertheless, results from validation studies sug-
gest that coffee consumption can be assessed with relatively
high validity (54). Third, we found the presence of heteroge-
neity among studies, which may be related to geographical
area. We observed a stronger inverse relationship between
coffee consumption and all-cause mortality among studies
conducted in Europe than among those conducted in the
United States. Because coffee composition can vary substan-
tially, differences in the association across countries may be
related to different types of coffee powder, different methods
of preparation, and different serving sizes (4). In addition,
02468
Coffee Consumption, cups/day
0.70
0.75
0.80
0.85
0.90
0.95
1.00
1.05
1.10
Relative Risk
Figure 4. Pooled dose-response association between coffee con-
sumption and cancer mortality (solid line) in a meta-analysis, 1966–
2013. Coffee consumption was modeled with restricted cubic splines
in a multivariate random-effects dose-response model. The relative
risks are plotted on the log scale. Dashed lines representthe 95% con-
fidence intervals for the spline model. No coffee consumption (0 cups/
day) served as the referent group.
772 Crippa et al.
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Table 4. Adjusted Relative Risk of All-Cancer Mortality According to Coffee Consumption (Versus No Consumption) in Prospective Studies, by Study Location, Sex, Type of Smoking
Adjustment, and Alcohol Use, 1966–2013
Coffee Consumption, cups/day
1234567 8
RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI RR 95% CI
All studies (n= 8) 0.98 0.96, 1.01 0.97 0.93, 1.01 0.98 0.93, 1.02 0.99 0.95, 1.03 1.01 0.97, 1.05 1.03 0.99, 1.08 1.06 1.00, 1.11 1.08 1.01, 1.15
Inclusion of 1 excluded article
a
(n= 9) 0.98 0.95, 1.00 0.96 0.92, 1.00 0.96 0.92, 1.01 0.98 0.93, 1.03 0.99 0.95, 1.05 1.01 0.96, 1.07 1.03 0.97, 1.10 1.05 0.98, 1.13
Study location
Europe (n= 2) 1.04 0.83, 1.30 1.07 0.71, 1.60 1.08 0.67, 1.73 1.07 0.67, 1.71 1.05 0.67, 1.65 1.04 0.67, 1.61 1.02 0.66, 1.59 1.01 0.64, 1.59
United States (n= 3) 0.99 0.97, 1.00 0.98 0.95, 1.01 0.98 0.95, 1.02 1.00 0.96, 1.03 1.01 0.97, 1.06 1.03 0.98, 1.08 1.05 0.99, 1.12
Japan (n= 3) 0.94 0.84, 1.06 0.91 0.74, 1.11 0.92 0.75, 1.14 0.98 0.79, 1.21 1.05 0.81, 1.36
Sex
Men (n= 6) 1.00 0.97, 1.02 0.99 0.95, 1.04 1.01 0.96, 1.06 1.03 0.98, 1.08 1.05 1.00, 1.10 1.08 1.01, 1.14 1.10 1.02, 1.18 1.13 1.03, 1.23
Women (n= 5) 0.96 0.91, 1.01 0.93 0.86, 1.01 0.93 0.86, 1.01 0.95 0.89, 1.01 0.98 0.93, 1.03 1.00 0.93, 1.08
Type of smoking adjustment
b
Status (n= 4) 0.95 0.88, 1.03 0.91 0.79, 1.05 0.90 0.76, 1.07 0.90 0.74, 1.09 0.91 0.71, 1.15 0.91 0.68, 1.23 0.92 0.64, 1.32 0.93 0.61, 1.43
Categories (n= 6) 0.98 0.93, 1.03 0.96 0.88, 1.06 0.96 0.88, 1.05 0.97 0.91, 1.04 0.98 0.91, 1.06 1.00 0.90, 1.11 1.01 0.87, 1.18
Continuous (n= 2) 0.99 0.97, 1.01 0.99 0.95, 1.03 1.00 0.95, 1.04 1.02 0.97, 1.06 1.04 0.99, 1.08 1.06 1.01, 1.11 1.08 1.01, 1.15
Alcohol adjustment
No (n= 2) 1.04 0.82, 1.31 1.07 0.71, 1.61 1.07 0.67, 1.72 1.06 0.66, 1.72 1.05 0.62, 1.76 1.03 0.56, 1.89 1.01 0.49, 2.09 1.00 0.42, 2.35
Yes (n= 6) 0.98 0.96, 1.01 0.97 0.93, 1.01 0.97 0.93, 1.02 0.99 0.95, 1.03 1.01 0.97, 1.05 1.03 0.99, 1.08 1.06 1.00, 1.12 1.08 1.01, 1.16
Abbreviations: CI, confidence interval; RR, relative risk.
a
Includes 1 study that did not provide information about the number of subjects.
b
Smoking status, category of cigarette smoking, or number of cigarettes smoked per day (continuous variable).
Coffee and Mortality 773
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
different genotypes and gene-environment interactions may
partially explain the observed variation among studies. For
example, cytochrome P-450 1A2 (CYP1A2) genotype has
been shown to be responsible for 95% of caffeine metabolism
and thus for the effect of coffee (55). Another limitation
could be related to reverse causation, since persons with
chronic disease and poor health might abstain from coffee
drinking. However, we considered only studies that targeted
the general population, excluding all studies that analyzed
specific subpopulations (e.g., persons with diabetes or hyper-
tension). Additionally, 3 different reports (31,37,50) evalu-
ated reverse causation by excluding subjects with a history of
cancer or CVD or deaths that occurred during the first few
years of follow-up. With these exclusions, the estimates
showed no remarkable alteration of the associations. A sim-
ilar result was obtained in a sensitivity analysis in a recent
meta-analysis of the relationship between coffee and all-cause
mortality (11). Finally, the shape of the observed associations
might have been influenced by observations in high-dose cat-
egories (56). However, removal of data points above 6 cups of
coffee per day did not substantially change the results.
In summary, results from this dose-response meta-analysis
indicate that coffee consumption is inversely associated with
all-cause and CVD mortality but not with cancer mortality.
People who regularly drink a moderate amount of coffee
(3–4 cups/day) may have lower risks of death from all causes
and CVD than persons who rarely drink coffee. It is unclear
whether the nonlinear association between coffee and CVD
mortality is related to the harmful effects of caffeine or is
due to other risk factors related to coffee consumption.
ACKNOWLEDGMENTS
Author affiliations: Unit of Nutritional Epidemiology, In-
stitute of Environmental Medicine, Karolinska Institutet,
Stockholm, Sweden (Alessio Crippa, Andrea Discacciati,
Susanna C. Larsson, Alicja Wolk, Nicola Orsini); and
Unit of Biostatistics, Institute of Environmental Medicine,
Karolinska Institutet, Stockholm, Sweden (Alessio Crippa,
Andrea Discacciati, Nicola Orsini).
This work was partly supported by a Young Scholar
Award from the Karolinska Institutet’s Strategic Program
in Epidemiology.
The funders played no role in the study design, data col-
lection, or analysis, the decision to publish, or manuscript
preparation.
Conflict of interest: none declared.
REFERENCES
1. Noordzij M, Uiterwaal CSPM, Arends LR, et al. Blood pressure
response to chronic intake of coffee and caffeine: a
meta-analysis of randomized controlled trials. J Hypertens.
2005;23(5):921–928.
2. Hartley TR, Lovallo WR, Whitsett TL. Cardiovascular effects
of caffeine in men and women. Am J Cardiol. 2004;93(8):
1022–1026.
3. Jee SH, He J, Appel LJ, et al. Coffee consumption and serum
lipids: a meta-analysis of randomized controlled clinical trials.
Am J Epidemiol. 2001;153(4):353–362.
4. Rebello SA, van Dam RM. Coffee consumption and
cardiovascular health: getting to the heart of the matter. Curr
Cardiol Rep. 2013;15(10):403.
5. van Dam RM, Feskens EJM. Coffee consumption and risk
of type 2 diabetes mellitus. Lancet. 2002;360(9344):
1477–1478.
6. Kempf K, Herder C, Erlund I, et al. Effects of coffee
consumption on subclinical inflammation and other risk factors
for type 2 diabetes: a clinical trial. Am J Clin Nutr. 2010;91(4):
950–957.
7. Ding M, Bhupathiraju SN, Chen M, et al. Caffeinated and
decaffeinated coffee consumption and risk of type 2 diabetes: a
systematic review and a dose-response meta-analysis. Diabetes
Care. 2014;37(2):569–586.
8. Ding M, Bhupathiraju SN, Satija A, et al. Long-term coffee
consumption and risk of cardiovascular disease: a systematic
review and a dose-response meta-analysis of prospective cohort
studies. Circulation. 2014;129(6):643–659.
9. Yu X, Bao Z, Zou J, et al. Coffee consumption and risk of
cancers: a meta-analysis of cohort studies. BMC Cancer. 2011;
11:96.
10. Malerba S, Turati F, Galeone C, et al. A meta-analysis of
prospective studies of coffee consumption and mortality for all
causes, cancers and cardiovascular diseases. Eur J Epidemiol.
2013;28(7):527–539.
11. Je Y, Giovannucci E. Coffee consumption and total mortality: a
meta-analysis of twenty prospective cohort studies. Br J Nutr.
2014;111(7):1162–1173.
12. Stroup DF, Berlin JA, Morton SC, et al. Meta-analysis of
observational studies in epidemiology: a proposal for reporting.
JAMA. 2000;283(15):2008–2012.
13. Hamling J, Lee P, Weitkunat R, et al. Facilitating meta-analyses
by deriving relative effect and precision estimates for alternative
comparisons from a set of estimates presented byexposure level
or disease category. Stat Med. 2008;27(7):954–970.
14. Orsini N, Bellocco R, Greenland S. Generalized least squares
for trend estimation of summarized dose-response data. Stata J.
2006;6(1):40–57.
15. Orsini N, Li R, Wolk A, et al. Meta-analysis for linear and
nonlinear dose-response relations: examples, an evaluation of
approximations, and software. Am J Epidemiol. 2012;175(1):
66–73.
16. Jackson D, White IR, Thompson SG. Extending DerSimonian
and Laird’s methodology to perform multivariate random
effects meta-analyses. Stat Med. 2010;29(12):1282–1297.
17. Desquilbet L, Mariotti F. Dose-response analyses using
restricted cubic spline functions in public health research. Stat
Med. 2010;29(9):1037–1057.
18. Higgins JPT, Thompson SG. Quantifying heterogeneity in a
meta-analysis. Stat Med. 2002;21(11):1539–1558.
19. Egger M, Davey Smith G, Schneider M, et al. Bias in
meta-analysis detected by a simple, graphical test. BMJ. 1997;
315(7109):629–634.
20. Crippa A. Dosresmeta: Performing Multivariate Dose-
Response Meta-Analysis. Vienna, Austria: R Foundation for
Statistical Computing; 2013. http://CRAN.R-project.org/
package=dosresmeta. Accessed January 20, 2014.
21. Viechtbauer W. metafor: Meta-Analysis Package for R. Vienna,
Austria: R Foundation for Statistical Computing; 2010. http://
CRAN.R-project.org/package=metafor. Accessed January
20, 2014.
22. R Development Core Team. R: A Language and Environment
for Statistical Computing. Vienna, Austria: R Foundation for
774 Crippa et al.
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from
Statistical Computing; 2009. http://www.R-project.org.
Accessed January 20, 2014.
23. LeGrady D, Dyer AR, Shekelle RB, et al. Coffee consumption
and mortality in the Chicago Western Electric Company Study.
Am J Epidemiol. 1987;126(5):803–812.
24. Rosengren A, Wilhelmsen L. Coffee, coronary heart disease
and mortality in middle-aged Swedish men: findings from the
Primary Prevention Study. J Intern Med. 1991;230(1):67–71.
25. Klatsky AL, Armstrong MA, Friedman GD. Coffee, tea, and
mortality. Ann Epidemiol. 1993;3(4):375–381.
26. Woodward M, Tunstall-Pedoe H. Coffee and tea consumption
in the Scottish Heart Health Study follow up: conflicting
relations with coronary risk factors, coronary disease, and all
cause mortality. J Epidemiol Community Health. 1999;53(8):
481–487.
27. Kleemola P, Jousilahti P, Pietinen P, et al. Coffee consumption
and the risk of coronary heart disease and death. Arch Intern
Med. 2000;160(22):3393–3400.
28. Iwai N, Ohshiro H, Kurozawa Y, et al. Relationship between
coffee and green tea consumption and all-cause mortality in a
cohort of a rural Japanese population. J Epidemiol. 2002;12(3):
191–198.
29. Jazbec A, SimićD, CorovićN, et al. Impact of coffee and other
selected factors on general mortality and mortality due to
cardiovascular disease in Croatia. J Health Popul Nutr. 2003;
21(4):332–340.
30. Andersen LF, Jacobs DR Jr, Carlsen MH, et al. Consumption of
coffee is associated with reduced risk of death attributed to
inflammatory and cardiovascular diseases in the Iowa Women’s
Health Study. Am J Clin Nutr. 2006;83(5):1039–1046.
31. Paganini-Hill A, Kawas CH, Corrada MM. Non-alcoholic
beverage and caffeine consumption and mortality: the Leisure
World Cohort Study. Prev Med. 2007;44(4):305–310.
32. Lopez-Garcia E, van Dam RM, Li TY, et al. The relationship of
coffee consumption with mortality. Ann Intern Med. 2008;
148(12):904–914.
33. Happonen P, Läärä E, Hiltunen L, et al. Coffee consumption
and mortality in a 14-year follow-up of an elderly northern
Finnish population. Br J Nutr. 2008;99(6):1354–1361.
34. Ahmed HN, Levitan EB, Wolk A, et al. Coffee consumption
and risk of heart failure in men: an analysis from the Cohort of
Swedish Men. Am Heart J. 2009;158(4):667–672.
35. de Koning Gans JM, Uiterwaal CSPM, van der Schouw YT,
et al. Tea and coffee consumption and cardiovascular morbidity
and mortality. Arterioscler Thromb Vasc Biol. 2010;30(8):
1665–1671.
36. Sugiyama K, Kuriyama S, Akhter M, et al. Coffee consumption
and mortality due to all causes, cardiovascular disease, and
cancer in Japanese women. J Nutr. 2010;140(5):1007–1013.
37. Tamakoshi A, Lin Y, Kawado M, et al. Effect of coffee
consumption on all-cause and total cancer mortality: findings
from the JACC Study. Eur J Epidemiol. 2011;26(4):285–293.
38. Freedman ND, Park Y, Abnet CC, et al. Association of coffee
drinking with total and cause-specific mortality. N Engl J Med.
2012;366(20):1891–1904.
39. Liu J, Sui X, Lavie CJ, et al. Association of coffee consumption
with all-cause and cardiovascular disease mortality. Mayo Clin
Proc. 2013;88(10):1066–1074.
40. Gardener H, Rundek T, Wright CB, et al. Coffee and tea
consumption are inversely associated with mortality in a
multiethnic urban population. J Nutr. 2013;143(8):1299–1308.
41. Hart C, Smith GD. Coffee consumption and coronary heart
disease mortality in Scottish men: a 21 year follow up study.
J Epidemiol Community Health. 1997;51(4):461–462.
42. Leurs LJ, Schouten LJ, Goldbohm RA, et al. Total fluid and
specific beverage intake and mortality due to IHD and stroke in
the Netherlands Cohort Study. Br J Nutr. 2010;104(8):
1212–1221.
43. Mineharu Y, Koizumi A, Wada Y, et al. Coffee, green tea,
black tea and oolong tea consumption and risk of mortality
from cardiovascular disease in Japanese men and women.
J Epidemiol Community Health. 2011;65(3):230–240.
44. Heckman MA, Weil J, Gonzalez de Mejia E. Caffeine
(1, 3, 7-trimethylxanthine) in foods: a comprehensive review on
consumption, functionality, safety, and regulatory matters.
J Food Sci. 2010;75(3):R77–R87.
45. van Dam RM. Coffee consumption and coronary heart disease:
paradoxical effects on biological risk factors versus disease
incidence. Clin Chem. 2008;54(9):1418–1420.
46. Natella F, Scaccini C. Role of coffee in modulation of diabetes
risk. Nutr Rev. 2012;70(4):207–217.
47. Lucas M, O’Reilly EJ, Pan A, et al. Coffee, caffeine, and risk of
completed suicide: results from three prospective cohorts of
American adults. World J Biol Psychiatry. 2014;15(5):
377–386.
48. Hernán MA, Takkouche B, Caamaño-Isorna F, et al.
A meta-analysis of coffee drinking, cigarette smoking, and
the risk of Parkinson’s disease. Ann Neurol. 2002;52(3):
276–284.
49. Leitzmann MF, Willett WC, Rimm EB, et al. A prospective
study of coffee consumption and the risk of symptomatic
gallstone disease in men. JAMA. 1999;281(22):2106–2112.
50. Lopez-Garcia E, van Dam RM, Qi L, et al. Coffee consumption
and markers of inflammation and endothelial dysfunction in
healthy and diabetic women. Am J Clin Nutr. 2006;84(4):
888–893.
51. Libby P, Ridker PM, Maseri A. Inflammation and
atherosclerosis. Circulation. 2002;105(9):1135–1143.
52. Roivainen M, Viik-Kajander M, Palosuo T, et al. Infections,
inflammation, and the risk of coronary heart disease.
Circulation. 2000;101(3):252–257.
53. Shacter E, Weitzman SA. Chronic inflammation and cancer.
Oncology (Williston Park). 2002;16(2):217–226, 229.
54. Ferraroni M, Tavani A, Decarli A, et al. Reproducibility and
validity of coffee and tea consumption in Italy. Eur J Clin Nutr.
2004;58(4):674–680.
55. Cornelis MC, El-Sohemy A, Kabagambe EK, et al. Coffee,
CYP1A2 genotype, and risk of myocardial infarction. JAMA.
2006;295(10):1135–1141.
56. Bagnardi V, Zambon A, Quatto P, et al. Flexible meta-
regression functions for modeling aggregate dose-response
data, with an application to alcohol and mortality. Am J
Epidemiol. 2004;159(11):1077–1086.
57. Bortner RW, et al. A short rating scale as a potential
measure of pattern A behavior. J Chronic Dis. 1969;22(2):
87–91.
Coffee and Mortality 775
Am J Epidemiol. 2014;180(8):763–775
by guest on November 21, 2015http://aje.oxfordjournals.org/Downloaded from