Eur J Epidemiol DOI 10.1007/s10654-015-0052-3
MORTALITY
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality in Swedish women Marie Lo¨f1 • Sven Sandin2 • Li Yin2 • Hans-Olov Adami2,3 Elisabete Weiderpass2,4,5,6
•
Received: 28 December 2014 / Accepted: 28 May 2015 Springer Science+Business Media Dordrecht 2015
Abstract We investigated whether coffee consumption was associated with all-cause, cancer, or cardiovascular mortality in a prospective cohort of 49,259 Swedish women. Of the 1576 deaths that occurred in the cohort, 956 were due to cancer and 158 were due to cardiovascular disease. We used Cox proportional hazard models with adjustment for potential confounders to estimate multivariable relative risks (RR) and 95 % confidence intervals (CI). Compared to a coffee consumption of 0–1 cups/day, the RR for all cause-mortality was 0.81 (95 % CI 0.69–0.94) for 2–5 cups/day and 0.88 (95 % CI 0.74–1.05) for [5 cups/day. Coffee consumption was not associated with cancer mortality or cardiovascular mortality when analyzed in the entire cohort. However, in supplementary analyses of women over 50 years of age, the RR for all cause-mortality was 0.74 (95 % CI 0.62–0.89) for 2–5
cups/day and 0.86 (95 % CI 0.70–1.06) for [5 cups/day when compared to 0–1 cups/day. In this same subgroup, the RRs for cancer mortality were 1.06 (95 % CI 0.81–1.38) for 2–5 cups/day and 1.40 (95 % CI 1.05–1.89) for [5 cups/day when compared to 0–1 cups/day. No associations between coffee consumption and all-cause mortality, cancer mortality, or cardiovascular mortality were observed among women below 50 years of age. In conclusion, higher coffee consumption was associated with lower all-cause mortality when compared to a consumption of 0–1 cups/day. Furthermore, coffee may have differential effects on mortality before and after 50 years of age. Keywords Coffee All-cause mortality Cancer mortality Cardiovascular mortality Cohort
Introduction Electronic supplementary material The online version of this article (doi:10.1007/s10654-015-0052-3) contains supplementary material, which is available to authorized users. & Marie Lo¨f [email protected] 1
Department of Biosciences and Nutrition, Karolinska Institutet, NOVUM, 141 83 Huddinge, Stockholm, Sweden
2
Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
3
Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
4
Department of Community Medicine, Faculty of Health Sciences, University of Tromsø, The Arctic University of Norway, Tromsø, Norway
5
Department of Research, Cancer Registry of Norway, Oslo, Norway
6
Samfundet Folkha¨lsan, Helsinki, Finland
Coffee has been a component of the human diet for centuries and has well known acute effects that may be harmful to human health, such as increased blood pressure and inhibited insulin activity [1, 2]. Coffee also contains beneficial compounds such as antioxidants [3], though the net effect of these compounds on human health is uncertain. The quantitative relationship between coffee and all-cause mortality, cancer, and cardiovascular disease has been the topic of a large number of prospective studies with inconsistent results for all-cause and cardiovascular mortality and have generally shown null results for cancer mortality. Four recent metaanalyses examined the association between coffee consumption and mortality [4–7]. All four reported an inverse association with all-cause mortality [4–7]; an unclear [6] or inverse association [4] with cardiovascular mortality; and no association with cancer mortality [4, 6].
123
M. Lo¨f et al.
Clearly, the associations between coffee consumption and mortality remain unclear, which calls for further studies. Previous studies investigated coffee consumption and all-cause or cause-specific mortality in populations consisting of primarily postmenopausal women [8–12], or covered a wide range of ages, for instance 20–69 years [13] and 40–75 years [14]. All previous studies had a mean or median age at baseline that was above 40 years. To the best of our knowledge, only two studies so far have evaluated whether associations between coffee consumption and allcause or cause-specific mortality differ by age [15, 16]. Tamakoshi et al. [16] reported similar risk estimates for women aged 40–59 years and 60–79 years at baseline for all-cause and cancer mortality. Liu et al. [15] reported a positive association between coffee consumption and allcause mortality, but only in men and women younger than 55 years at baseline. The Swedish Women’s Lifestyle and Health Cohort is a prospective cohort characterized by the young age of its participants at baseline (30–49 years). Thus, our specific aim was to evaluate whether coffee consumption is associated with all-cause mortality, cancer mortality, and cardiovascular mortality in young and middle-aged women.
Methods Study cohort The study cohort and exposure assessment have been previously described in detail [17]. Briefly, roughly 96,000 women aged 30–49 years (born 1942–1962) and residing in the Uppsala Health Care Region in 1991–1992 were randomly selected from four age strata (30–34, 35–39, 40–44, and 45–49 years) and invited to participate in the Swedish component of the Scandinavian Women’s Lifestyle and Health Cohort. The study questionnaires were returned by 49,259 women (9 % postmenopausal) who were enrolled in the study. We excluded 4119 women from the initial cohort due to death before enrollment (n = 2), emigration before final enrolment in the cohort (n = 10), energy intakes below the 1st or above the 99th percentile (n = 1476), and missing information on body mass index (BMI), education, or age at first birth (n = 2631). Thus, the final study sample included 45,140 women. The study was approved by the ethical committee at the University of Uppsala. Coffee consumption was assessed in an open-ended question on how many cups of coffee per day or per week the women consumed during the last year. The women were told that one cup was equivalent to 1.5 dl (150 ml) of coffee. Information on BMI, education, parity, smoking,
123
physical activity, diabetes, and alcohol consumption were also obtained from the baseline questionnaire. Follow-up for death among the study participants was conducted through linkages with existing nationwide population and health registries, using the individually unique national registration number assigned to all residents in Sweden. Follow-up was virtually complete with respect to death and emigration [18]. For death, information on date and cause of death was obtained from the nationwide Cause of Death Registry while for emigration information was obtained from the Emigration registry. The start of follow-up was defined as the date of return of the questionnaire. Observation time was calculated from date of entry into the cohort until the date of emigration (first emigration), date of death or 31 December 2010, whichever came first. Codes from the International Classification of Diseases, revisions 9 and 10 (ICD-9, ICD-10), were used (cardiovascular death: ICD-9 codes 410–414, 430–438, and 440; ICD-10 codes I20–I25, I60–I70, and I74; cancer death: ICD-9 codes 140–209 and ICD-10 codes: C00–C97). Statistical analysis The relative risk (RR) of death, comparing exposed and non-exposed women, was estimated by calculating hazard ratios using Cox proportional hazard models with attained age as an underlying time scale [19]. The RR was considered statistically significant when the associated twosided 95 % Wald-type confidence interval (CI) did not include unity, corresponding to a two-sided 5 % level of significance. First we fitted crude models (including only coffee consumption in categories, i.e. 0–1, 2–5, [5 cups/day), and thereafter multivariable models in which we adjusted for the following pre-selected potential confounders assessed at baseline: education (0–10, 11–13, [13 years), BMI (\25, 25–29.9, C30 kg/m2), parity (0, 1, 2, 3, C4), age at first birth (years), smoking habits (never smoker, former smoker of \10, 10–14, 15–19, or C20 cigarettes/day, and current smokers of\10, 10–14, 15–19, or C20 cigarettes/day) and alcohol consumption (g/day). For parity and alcohol we have complete data and for BMI, education, and age at first birth we have almost complete data and thus we excluded women with missing data on these variables. For physical activity, we conducted supplementary analyses where we adjusted for physical activity and compared the results with the original models (see below). We also conducted supplementary analyses. Firstly, we checked whether our associations were affected by adjustment for physical activity level (level 1: sedentary lifestyle, level 3: moderately active lifestyle characterized
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality…
by a few walks per week, and level 5: active lifestyle with exercise a couple of times per week) or diabetes at baseline (yes or no). We also fitted separate models for all-cause, cancer, and cardiovascular mortality occurring before and after 50 years of age in order to evaluate risk factors in different periods of life, e.g. the effect of menopausal status or the accumulated life-long effect of diet. In Sweden, the average age at menopause is 50 years [20]. As coffee consumption and smoking are positively correlated, residual confounding by smoking was of concern; therefore we also fitted adjusted models with and without smoking. To evaluate the functional form of the associations between coffee consumption and all-cause, cancer, cardiovascular and other mortality, we fitted splines for each outcome. In addition, we re-ran our analyses excluding women with a diagnosis of cancer, cardiovascular disease, or diabetes in the Patient Registry up to 2 years before cohort entry. Finally, we re-ran our analyses, restricting follow-up time to 10 years after baseline in order to investigate whether our associations were attenuated by time. As in a previous study [21], we used a coffee consumption of 0–1 cups/day as the reference category, as non-consumers are scarce in Sweden (in this cohort only 9 % of women reported no coffee consumption and only 112 deaths occurred among these non-consumers). We checked the proportional hazards assumption by plotting the Schoenfeld residuals versus time [22]. All statistical analyses were carried out using the SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).
Results Women were followed for an average of 18 years (844, 706 person-years). The baseline characteristics of the study women classified by coffee consumption are shown in Table 1. Women in the two highest categories of coffee consumption (2–5 cups/day and [5 cups/day) were less educated than women in the lowest category (0–1 cups/day). Furthermore, the prevalence of current smokers was higher in the two highest categories of coffee consumption compared to that in the lowest category. During follow-up, there were a total of 1576 deaths: 956 due to cancer, 158 due to cardiovascular disease, and 462 due to other causes. The major reasons for deaths due to other causes were accidents or suicides (34 %), diseases of the nervous system (19 %), and respiratory diseases (13 %). The proportional hazards assumption was verified by graphic evaluation of the plotted Schoenfeld residuals for all models. Table 2 shows the observed associations between coffee consumption and all-cause, cancer, and cardiovascular mortality. In the crude models, a coffee consumption of[5 cups/day was associated with increased risks for all-cause,
cancer, and cardiovascular mortality when compared to a coffee consumption of 0–1 cups/day (all-cause mortality: RR 1.27, 95 % CI 1.08–1.51; cancer mortality: RR 1.61, 1.29–2.03, and cardiovascular mortality: RR 1.60, 95 % CI 0.93–2.78). None of these RRs remained statistically significant after adjustment for confounders and the RRs for all-cause and cardiovascular mortality switched from a positive to an inverse direction after adjustment. When adjusting for confounders, coffee consumption of 2–5 cups/day was associated with a statistically significant, lower risk of all-cause mortality compared to 0–1 cups/day (RR 0.81, 95 % CI 0.69–0.94). The corresponding RR for all-cause mortality for a coffee consumption of [5 cups/day compared 0–1 cups/day was similar, i.e. 0.88, although not statistically significant (95 % CI 0.74–1.05). Coffee consumption was not associated with cancer or cardiovascular mortality in the adjusted models. Finally, coffee consumptions of 2–5 and [5 cups/day were associated with decreased mortality due to other causes compared to 0–1 cups/day (RR 0.59, 95 % CI 0.46–0.75 and RR 0.50, 95 % CI 0.37–0.69, respectively). When we excluded smoking from the analyses, the RRs were similar to those in the crude models for all four outcomes (Table 2). As shown in supplementary Table S1, exclusion of individual confounders from the adjusted models produced estimates similar to those obtained in the fully-adjusted models. Table S2 shows results stratified by age at follow-up. The RRs for all-cause mortality, cancer mortality, cardiovascular mortality, and mortality due to other causes were not statistically significant among women below 50 years of age. Among women older than 50 years of age, coffee consumption of 2–5 cups/day was associated with a statistically significant, lower risk of all-cause mortality compared to 0–1 cups/day. The corresponding RR for a coffee consumption of [5 cups/day was similar, i.e. 0.86, although not statistically significant (95 % CI 0.70–1.06). Furthermore, among women aged 50 years and older, the RR for cancer mortality was 1.06 (0.81–1.38) for 2–5 cups/day and 1.40 (95 % CI 1.05–1.89) for [5 cups/day when compared to 0–1 cups/day. Finally, a coffee consumption of 2–5 and[5 cups/day was associated with lower mortality due to other causes after 50 years of age compared to drinking 0–1 cups/day (RR 0.42, 95 % CI 0.32–0.56 and RR 0.37, 95 % CI 0.26–0.53, respectively). Splines for RRs and 95 % CIs for coffee consumption of 1–7 cups/day in relation to all-cause mortality, cancer mortality, cardiovascular disease mortality, and other cause mortality are shown in Figure S1. There was no support for any substantial change in risk by increasing coffee intake when considering individual cups of coffee compared with the approach aggregating the data into categories of coffee intake.
123
M. Lo¨f et al. Table 1 Baseline characteristics and number of deaths according to categories of coffee consumption
Characteristic
Coffee consumption (cups/day) 0–1
2–5
[5
6766 (15)
29,930 (66)
8444 (19)
Q1
33
35
35
Mean
38
40
40
Q3
43
45
45
Q1
57
58
58
Mean
64
64
66
Q3
69
69
72
Total number of women N (%) Age at enrollment (years)
Body weight (kg)
Height (cm) Q1
162
162
162
Mean
166
166
166
Q3
170
170
170
Q1
20.5
21.0
21.5
Mean
23.2
23.4
24.0
Q3
25.0
25.0
25.5
0–10 years
(1344; 20)
(8612; 29)
(3315; 39)
11–13 years
(2730; 40)
(11,712; 39)
(3279; 39)
[13 years
(2692; 40)
(9606; 32)
(1850; 22)
Q1
0
0
0
Mean
2
4
9
Q3
0
7
17
Body mass index (kg/m2)
Education (n; %)
Number of cigarettes/day
Smoking status (n; %) Never
(4173; 62)
(12,673; 42)
(1810; 22)
Former
(1633; 24)
(9364; 31)
(2399; 28)
Current
(960; 14)
(7893; 26)
(4235; 50)
Alcohol consumption (n; %) \5 g/day
(5553; 82)
(21,901; 73)
(6233; 74)
5–25 g/day
(1188; 18)
(7916; 26)
(2149; 25)
[25 g/day
(25; 1)
(113; 1)
(62; 1)
1
(335; 5)
(978; 3)
(478; 6)
2
(765; 11)
(2962; 10)
(908; 11)
3
(3590; 53)
(17,383; 58)
(4845; 57)
4
(1100; 16)
(5057; 17)
(1227; 15)
5
(557; 8)
(2434; 8)
(664; 8)
Missing
(419;6)
(1116; 4)
(322; 4)
Physical activity level (n; %)
Diabetes (n; %) Yes No
(98; 1.5)
(343; 1.1)
(155; 1.8)
(6668; 98.5)
(29,587; 98.9)
(8289; 98.2)
52 (65)
503 (159)
1103 (325)
Coffee consumption (g/day) Mean (SD)
123
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality… Table 1 continued
Characteristic
Coffee consumption (cups/day) 0–1
2–5
[5
Any death
208
979
389
Any cancer deatha
105
600
251
Any cardiovascular death
18
96
44
85
283
94
Number of deaths
b
Other causes
The Swedish women’s lifestyle and health cohort study Cup 150 ml; n number of women, Q1 1st quartile, Q3 3rd quartile, SD standard deviation Number of deaths due to breast cancer (25 %), lung cancer (19 %), other cancers (56 %)
a
b
Number of deaths due to suicide and accidents (34 %), diseases of the nervous system (19 %), respiratory diseases (14 %), and other (33 %)
Table 2 Relative risks (RR) and 95 % confidence intervals (CI) of coffee consumption in relation to all-cause and cause-specific mortality Cause of death All-cause mortality
Cancer mortality
Cardiovascular disease mortality
Other causes
Cases
Cups of coffee (cups/day)
Crude RR (95 % CI)
Adjusted RR (95 % CI)a
Adjusted RR (95 % CI)—smokingb
208
0–1
Reference
Reference
Reference
979
2–5
0.89 (0.77–1.03)
0.81 (0.69–0.94)
0.88 (0.76–1.03)
389
[5
1.27 (1.08–1.51)
0.88 (0.74–1.05)
1.17 (0.99–1.39)
105
0–1
Reference
Reference
Reference
600
2–5
1.07 (0.87–1.32)
0.99 (0.81–1.23)
1.06 (0.86–1.31)
251
[5
1.61 (1.29–2.03)
1.25 (0.98–1.58)
1.53 (1.21–1.93)
18 96
0–1 2–5
Reference 0.97 (0.59–1.61)
Reference 0.76 (0.46–1.26)
Reference 0.93 (0.56–1.54)
44
[5
1.60 (0.93–2.78)
0.74 (0.42–1.30)
1.35 (0.78–2.36)
85
0–1
Reference
Reference
Reference
283
2–5
0.65 (0.51–0.83)
0.59 (0.46–0.75)
0.65 (0.51–0.83)
94
[5
0.77 (0.58–1.04)
0.50 (0.37–0.69)
0.68 (0.51–0.92)
a
Adjusted for BMI, education, smoking, alcohol, parity and age at first birth
b
Adjusted for BMI, education, alcohol, parity and age at first birth
We found no evidence that the RRs for any of the outcomes were substantially attenuated as follow-up time increased. For instance, the RR for all-cause mortality was 0.87 (95 % CI 0.67–1.13) when a coffee consumption of 2–5 cups/day was compared to 0–1 cups/day within 10 years of follow-up, while the corresponding estimate for a follow-up time of more than 10 years was 0.77 (95 % CI 0.64–0.94). Excluding women that had diabetes, cardiovascular disease or cancer disease 2 years prior to baseline did not affect our results in any major way. For instance, when excluding these women, the RRs for all-cause mortality were 0.79 (95 % CI 0.67–0.93) for 2–5 cups/day and 0.91 (95 % CI 0.76–1.10) for [5 cups/day versus 0–1 cups/day. Finally, adjustment for physical activity level or diabetes at baseline did not affect any of the observed RRs (data not shown).
Discussion In this large and detailed prospective cohort of young and middle-aged Swedish women, a coffee consumption of 2–5 cups/day was associated with a 19 % lower risk of allcause mortality compared to a consumption of 0–1 cups/day. This finding is in agreement with several large, recent studies, including the Nurses’ Health Study and the National Institutes of Health—AARP Diet and Health Study [9, 14]. In the Nurses’ Health Study, a coffee consumption of 2–3, 4–5, and C6 cups/day was associated with an 18, 26 and 17 % lower mortality risk, respectively, when compared to no coffee consumption [14]. The corresponding results for the National Institute of Health— AARP Diet and Health Study were 13, 16, and 15 % [9]. Furthermore, our finding is in agreement with one recent
123
M. Lo¨f et al.
meta-analyses that reported an 18 % lower risk for allcause mortality among women with a coffee consumption of 3 cups/day compared to no consumption when pooling eight prospective cohorts [4]. The RR for all-cause mortality for a coffee consumption of [5 cups/day compared to 0–1 cups/day was 0.88 (95 % CI 0.74–1.05), which was not very different from the corresponding RR for 2–5 cups/day. Thus, we did not find any evidence that increasing coffee consumption above 4 cups/day further reduced the risk of all-cause mortality. This finding is in agreement with the results of the Nurses’ Health Study and the National Institute of Health—AARP Diet and Health Study [9, 14]. It is also consistent with three recent meta-analyses [4–6]. As in many previous prospective cohort studies carried out among women [8, 9, 23–25], coffee consumption was not associated with cancer mortality in our study, which is also in agreement with two recent meta-analyses [4, 6]. Two prospective cohort studies in Japanese women reported some evidence for an inverse association between coffee consumption and cancer mortality [16, 26]. However, the effect of coffee consumption differs by cancer site. Indeed, inverse associations have been reported between coffee consumption and risk of colorectal [27], endometrial [28], and liver cancer [29], while a null association was reported for breast cancer [30]. In this context, it is relevant to note that one-fourth of the cancer deaths in our cohort were due to breast cancer. In supplementary analyses, the estimated RRs for allcause mortality and mortality due to other causes were shifted downwards, but were still statistically significant for women above 50 years of age. In this same age group, there was also evidence of a reverse effect i.e. an increased risk of cancer mortality. The RRs for all-cause and cancer mortality were close to unity and not statistically significant for women below 50 years of age. These patterns may be due to true differences in risk between pre- and postmenopausal women. However, this interpretation may not necessarily be causal due to the possible confounding effect of attained age during follow-up for older women. Thus, these results need confirmation. In this study, the RR for cardiovascular mortality was 0.76 (95 % CI 0.46–1.26) for a coffee consumption of 2–5 cups/day and 0.74 (95 % CI 0.42–1.30) for [5 cups/day when compared to 0–1 cups/day. In comparison, the corresponding RRs from the Nurses’ Health Study for 2–3 and C6 cups/day, compared to no consumption, was 0.75 (95 % CI 0.66–0.86) and 0.81 (95 % CI 0.61–1.06), respectively [14]. The corresponding results from the National Institute of Health—AARP Diet and Health Study were 0.85 (95 % CI 0.76–0.95) and 0.72 (95 % CI 0.59–0.88) [9]. Our findings also agree with a recent meta-analysis that reported a RR for cardiovascular
123
mortality of 0.76 (95 % CI 0.67–0.85) among women with a coffee consumption of 3 cups/day compared to nonconsumers [4]. Thus, it is reasonable to conclude that our results for cardiovascular mortality are similar in magnitude and direction to previous, larger cohorts of women, but our results failed to reach statistical significance, possibly due to limited numbers. When we excluded smoking from the adjusted models, the RRs were similar to the corresponding crude models. Smoking induces changes in hepatic enzymes that increase caffeine clearance, which leads to higher coffee consumption in smokers [31]. Tobacco smoking was indeed the strongest confounder in the recent analyses within the large National Institute of Health—AARP Diet and Health Study [9]. Thus, our results confirm that smoking is a strong confounder of the association between coffee consumption and mortality. Our adjustment for smoking was more detailed than that performed in the studies by Mineharu et al. [32] and Liu et al. [15], which might explain the positive associations they reported between increased coffee consumption and cardiovascular mortality in women. The report by Liu et al. [15] also showed a positive association between coffee consumption and all-cause mortality in men and women younger than 55 years. In a reply to that report, Gretzky et al. suggested that the observed age-dependent effect could be due to the fact that the antioxidant compounds in coffee may change the balance of normal body oxidation in an unfavorable manner in young people, who have lower levels of reactive oxygen species [33]. Our study contradicts these findings because we found decreased all-cause mortality with increased coffee consumption, especially among deaths occurring after 50 years of age. One possible explanation for these contradictory findings is residual confounding by smoking, as Liu et al. [15] assessed smoking status as a binary variable, while we adjusted for the number of cigarettes smoked. Furthermore, the study by Liu et al. stratified by baseline age (\55 years and C55 years), while we fitted models for all-cause and cause-specific mortality occurring before and after 50 years of age, which is superior in order to evaluate if the risk differs by age. It is also relevant to note that the predominant causes of cancer-related deaths in Japanese women are colorectal, stomach and lung cancer while they are breast, colorectal and lung cancer in Swedish women [34]. Finally, we cannot exclude that our observed associations for women above 50 years of age may be due to certain risk backgrounds (for instance genetic predisposition or underlying diseases at baseline). Further studies are needed in order to clarify whether there is a true age-dependent effect of coffee consumption on allcause mortality. The major causes in the category other causes of death were accidents or suicides, diseases of the nervous system,
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality…
and respiratory diseases. There is no obvious explanation for how coffee consumption could reduce mortality from these causes. However, a recent study showed that coffee was associated with lower risk of suicide, perhaps due to an accelerated turnover of serotonin, which is involved in depression [35]. Furthermore, a meta-analysis reported an inverse association between coffee consumption and the risk of Parkinson’s disease [36], possible mechanisms being that caffeine acts as antagonists of the adenosine receptors or activates specific neuroprotection signaling pathways [37]. We cannot exclude the possibility that the association we found with other causes of death was caused by residual confounding with socioeconomic status, dietary risk factors, or health status. We did adjust for socioeconomic status via the variable education; however, we cannot rule out the presence of remaining confounding due to things such as differences in work and life settings. The strengths of our study include its prospective design, with access to very detailed information on possible confounding factors, its large size and the complete follow-up. Misclassification of coffee consumption due to measurement error in the food frequency questionnaire and due to lack of follow-up on coffee consumption is unavoidable, but given the study design it is likely to be non-differential and thus to attenuate any true association. We found no evidence of a substantial attenuation when we evaluated the associations for a follow-up of more than 10 years compared to the first 10 years. A limitation of our study is that we did not have any information on time since smoking cessation for former smokers, nor on age at smoking initiation among current and former smokers. Our results did not change in any major way when we excluded women with diagnosed diabetes, cardiovascular disease or cancer 2 years before baseline. However, one major reason for people to stop drinking coffee may be unspecific gastrointestinal symptoms, such as heartburn, which was not part of our questionnaire and cannot be obtained from medical journals (and are not included in our database for this study). Hence, we cannot completely exclude the possibility of reverse causality, since women may have stopped drinking coffee prior to study enrollment due to poor health. Finally, the number of cardiovascular deaths in our study was low, limiting our power to detect statistically significant associations for this outcome. In conclusion, in this prospective cohort of Swedish women aged 30–49 years at baseline, consuming 2–5 cups of coffee/day was associated with reduced all-cause mortality when compared to a coffee consumption of 0–1 cups/day. Furthermore, our results do not rule out the possibility that coffee consumption may have differential effects on all-cause mortality and cancer mortality before and after 50 years of age.
Acknowledgments Council.
This study was funded by the Swedish Research
Conflict of interest of interest.
The authors declare that they have no conflict
References 1. Higdon JV, Frei B. Coffee and health: a review of recent human research. Crit Rev Food Sci Nutr. 2006;46(2):101–23. doi:10. 1080/10408390500400009. 2. Noordzij M, Uiterwaal CS, Arends LR, Kok FJ, Grobbee DE, Geleijnse JM. Blood pressure response to chronic intake of coffee and caffeine: a meta-analysis of randomized controlled trials. J Hypertens. 2005;23(5):921–8. ´ , Leake DS, Ames JM. In vitro antioxidant ac3. Go´mez-Ruiz JA tivity of coffee compounds and their metabolites. J Agric Food Chem. 2007;55(17):6962–9. doi:10.1021/jf0710985. 4. Crippa A, Discacciati A, Larsson SC, Wolk A, Orsini N. Coffee consumption and mortality from all causes, cardiovascular disease, and cancer: a dose-response meta-analysis. Am J Epidemiol. 2014;180(8):763–75. doi:10.1093/aje/kwu194. 5. Je Y, Giovannucci E. Coffee consumption and total mortality: a meta-analysis of twenty prospective cohort studies. Br J Nutr. 2013;. doi:10.1017/S0007114513003814. 6. 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):527–39. doi:10.1007/s10654-013-9834-7. 7. Zhao Y, Wu K, Zheng J, Zuo R, Li D. Association of coffee drinking with all-cause mortality: a systematic review and metaanalysis. Public Health Nutr. 2014;. doi:10.1017/S1368980014 001438. 8. Andersen LF, Jacobs DR Jr, Carlsen MH, Blomhoff R. 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):1039–46. 9. Freedman ND, Park Y, Abnet CC, Hollenbeck AR, Sinha R. Association of coffee drinking with total and cause-specific mortality. N Engl J Med. 2012;366(20):1891–904. doi:10.1056/ NEJMoa1112010. 10. Happonen P, Laara E, Hiltunen L, Luukinen H. Coffee consumption and mortality in a 14-year follow-up of an elderly northern Finnish population. Br J Nutr. 2008;99(6):1354–61. doi:10.1017/S0007114507871650. 11. Harris HR, Bergkvist L, Wolk A. Coffee and black tea consumption and breast cancer mortality in a cohort of Swedish women. Br J Cancer. 2012;107(5):874–8. doi:10.1038/bjc.2012. 337. 12. 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):305–10. doi:10.1016/j. ypmed.2006.12.011. 13. de Koning Gans JM, Uiterwaal CS, van der Schouw YT, et al. Tea and coffee consumption and cardiovascular morbidity and mortality. Arterioscler Thromb Vasc Biol. 2010;30(8):1665–71. doi:10.1161/ATVBAHA.109.201939. 14. Lopez-Garcia E, van Dam RM, Li TY, Rodriguez-Artalejo F, Hu FB. The relationship of coffee consumption with mortality. Ann Intern Med. 2008;148(12):904–14. doi:10.7326/0003-4819-14812-200806170-00003.
123
M. Lo¨f et al. 15. 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):1066–74. doi:10.1016/j.mayocp.2013.06.020. 16. Tamakoshi A, Lin Y, Kawado M, Yagyu K, Kikuchi S, Iso H. Effect of coffee consumption on all-cause and total cancer mortality: findings from the JACC study. Eur J Epidemiol. 2011;26(4):285–93. doi:10.1007/s10654-011-9548-7. 17. Kumle M, Weiderpass E, Braaten T, Persson I, Adami H. Use of oral contraceptives and breast cancer risk: the Norwegian– Swedish women’s lifestyle and health cohort study. Cancer Epidemiol Biomark Prev. 2002;11:1375–81. 18. Statistics Sweden. http://www.scb.se. Accessed 15 May 2015. 19. Korn E, Graubard B, Midthune D. Time-to-event analysis of longitudinal follow-up of a survey: choice of the time-scale. Am J Epidemiol. 1997;145:72–80. 20. Weiderpass E, Baron J, Adami H. Low-potency oestrogen and risk of endometrial cancer: a case–control study. Lancet. 1999;353:1824–8. 21. Weiderpass E, Sandin S, Lof M, et al. Endometrial cancer in relation to coffee, tea, and caffeine consumption: a prospective cohort study among middle-aged women in Sweden. Nutr Cancer. 2014;66(7):1132–43. doi:10.1080/01635581.2014.948214. 22. Grambsch P, Therneau T. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika. 1994;81:515–26. 23. Gardener H, Rundek T, Wright CB, Elkind MSV, Sacco RL. Coffee and tea consumption are inversely associated with mortality in a multiethnic urban population. J Nutr. 2013;143(8):1299–308. doi:10.3945/jn.112.173807. 24. 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):191–8. 25. Klatsky AL, Armstrong MA, Friedman GD. Coffee, tea, and mortality. Ann Epidemiol. 1993;3(4):375–81. doi:10.1016/10472797(93)90064-B. 26. 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):1007–13. doi:10. 3945/jn.109.109314. 27. Tian C, Wang W, Hong Z. Coffee consumption and risk of colorectal cancer: a dose-response analysis of observational studies. Cancer Causes Control. 2013;24:1265–8. doi:10.1007/s10552013-0200-6.
123
28. Je Y, Giovannucci E. Coffee consumption and risk of endometrial cancer; findings from a large up-to-date meta-analysis. Int J Cancer. 2012;131:1700–10. doi:10.1002/ijc.27408 Epub 2012 Jan 31. 29. Setiawan VW, Wilkens LR, Lu SC, Hernandez BY, Le Marchand L, Henderson BE. Association of coffee intake with reduced incidence of liver cancer and death from chronic liver disease in the US multiethnic cohort. Gastroenterology. 2015;148(1):118–25. doi:10.1053/j.gastro.2014.10.005 quiz e15. 30. Li X, Ren Z, Qin J. Coffee consumption and risk of breast cancer: an up-to-date meta-analysis. PLoS One. 2013;8:e52681. doi:10. 1371/journal.pone.0052681. 31. Wagner MG, Kapoor KG, Wagner AL. Factors affecting the association of coffee consumption with all-cause and cardiovascular disease mortality. Mayo Clin Proc. 2013;88(12):1491–2. doi:10.1016/j.mayocp.2013.10.010. 32. Mineharu Y, Koizumi A, Wada Y. 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):230–40. doi:10.1136/jech.2009. 097311. 33. Greitzer E. Heavy coffee drinking and age-dependent all-cause mortality. Mayo Clin Proc. 2013;88(12):1492–3. doi:10.1016/j. mayocp.2013.10.004. 34. Globocan 2012. Estimated cancer incidence, mortality and prevalence worldwide in 2012. http://globocan.iarc.fr/Default. aspx. Accessed 15 May 2015. 35. 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):377–86. doi:10.3109/15622975.2013.795243. 36. Hernan MA, Takkouche B, Caamano-Isorna F, Gestal-Otero JJ. A meta-analysis of coffee drinking, cigarette smoking, and the risk of Parkinson’s disease. Ann Neurol. 2002;52(3):276–84. doi:10. 1002/ana.10277. 37. Costa J, Lunet N, Santos C, Santos J, Vaz-Carneiro A. Caffeine exposure and the risk of Parkinson’s disease: a systematic review and meta-analysis of observational studies. J Alzheimers Dis. 2010;20(Suppl 1):S221–38. doi:10.3233/JAD-2010-091525.
MORTALITY
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality in Swedish women Marie Lo¨f1 • Sven Sandin2 • Li Yin2 • Hans-Olov Adami2,3 Elisabete Weiderpass2,4,5,6
•
Received: 28 December 2014 / Accepted: 28 May 2015 Springer Science+Business Media Dordrecht 2015
Abstract We investigated whether coffee consumption was associated with all-cause, cancer, or cardiovascular mortality in a prospective cohort of 49,259 Swedish women. Of the 1576 deaths that occurred in the cohort, 956 were due to cancer and 158 were due to cardiovascular disease. We used Cox proportional hazard models with adjustment for potential confounders to estimate multivariable relative risks (RR) and 95 % confidence intervals (CI). Compared to a coffee consumption of 0–1 cups/day, the RR for all cause-mortality was 0.81 (95 % CI 0.69–0.94) for 2–5 cups/day and 0.88 (95 % CI 0.74–1.05) for [5 cups/day. Coffee consumption was not associated with cancer mortality or cardiovascular mortality when analyzed in the entire cohort. However, in supplementary analyses of women over 50 years of age, the RR for all cause-mortality was 0.74 (95 % CI 0.62–0.89) for 2–5
cups/day and 0.86 (95 % CI 0.70–1.06) for [5 cups/day when compared to 0–1 cups/day. In this same subgroup, the RRs for cancer mortality were 1.06 (95 % CI 0.81–1.38) for 2–5 cups/day and 1.40 (95 % CI 1.05–1.89) for [5 cups/day when compared to 0–1 cups/day. No associations between coffee consumption and all-cause mortality, cancer mortality, or cardiovascular mortality were observed among women below 50 years of age. In conclusion, higher coffee consumption was associated with lower all-cause mortality when compared to a consumption of 0–1 cups/day. Furthermore, coffee may have differential effects on mortality before and after 50 years of age. Keywords Coffee All-cause mortality Cancer mortality Cardiovascular mortality Cohort
Introduction Electronic supplementary material The online version of this article (doi:10.1007/s10654-015-0052-3) contains supplementary material, which is available to authorized users. & Marie Lo¨f [email protected] 1
Department of Biosciences and Nutrition, Karolinska Institutet, NOVUM, 141 83 Huddinge, Stockholm, Sweden
2
Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
3
Department of Epidemiology, Harvard School of Public Health, Boston, MA, USA
4
Department of Community Medicine, Faculty of Health Sciences, University of Tromsø, The Arctic University of Norway, Tromsø, Norway
5
Department of Research, Cancer Registry of Norway, Oslo, Norway
6
Samfundet Folkha¨lsan, Helsinki, Finland
Coffee has been a component of the human diet for centuries and has well known acute effects that may be harmful to human health, such as increased blood pressure and inhibited insulin activity [1, 2]. Coffee also contains beneficial compounds such as antioxidants [3], though the net effect of these compounds on human health is uncertain. The quantitative relationship between coffee and all-cause mortality, cancer, and cardiovascular disease has been the topic of a large number of prospective studies with inconsistent results for all-cause and cardiovascular mortality and have generally shown null results for cancer mortality. Four recent metaanalyses examined the association between coffee consumption and mortality [4–7]. All four reported an inverse association with all-cause mortality [4–7]; an unclear [6] or inverse association [4] with cardiovascular mortality; and no association with cancer mortality [4, 6].
123
M. Lo¨f et al.
Clearly, the associations between coffee consumption and mortality remain unclear, which calls for further studies. Previous studies investigated coffee consumption and all-cause or cause-specific mortality in populations consisting of primarily postmenopausal women [8–12], or covered a wide range of ages, for instance 20–69 years [13] and 40–75 years [14]. All previous studies had a mean or median age at baseline that was above 40 years. To the best of our knowledge, only two studies so far have evaluated whether associations between coffee consumption and allcause or cause-specific mortality differ by age [15, 16]. Tamakoshi et al. [16] reported similar risk estimates for women aged 40–59 years and 60–79 years at baseline for all-cause and cancer mortality. Liu et al. [15] reported a positive association between coffee consumption and allcause mortality, but only in men and women younger than 55 years at baseline. The Swedish Women’s Lifestyle and Health Cohort is a prospective cohort characterized by the young age of its participants at baseline (30–49 years). Thus, our specific aim was to evaluate whether coffee consumption is associated with all-cause mortality, cancer mortality, and cardiovascular mortality in young and middle-aged women.
Methods Study cohort The study cohort and exposure assessment have been previously described in detail [17]. Briefly, roughly 96,000 women aged 30–49 years (born 1942–1962) and residing in the Uppsala Health Care Region in 1991–1992 were randomly selected from four age strata (30–34, 35–39, 40–44, and 45–49 years) and invited to participate in the Swedish component of the Scandinavian Women’s Lifestyle and Health Cohort. The study questionnaires were returned by 49,259 women (9 % postmenopausal) who were enrolled in the study. We excluded 4119 women from the initial cohort due to death before enrollment (n = 2), emigration before final enrolment in the cohort (n = 10), energy intakes below the 1st or above the 99th percentile (n = 1476), and missing information on body mass index (BMI), education, or age at first birth (n = 2631). Thus, the final study sample included 45,140 women. The study was approved by the ethical committee at the University of Uppsala. Coffee consumption was assessed in an open-ended question on how many cups of coffee per day or per week the women consumed during the last year. The women were told that one cup was equivalent to 1.5 dl (150 ml) of coffee. Information on BMI, education, parity, smoking,
123
physical activity, diabetes, and alcohol consumption were also obtained from the baseline questionnaire. Follow-up for death among the study participants was conducted through linkages with existing nationwide population and health registries, using the individually unique national registration number assigned to all residents in Sweden. Follow-up was virtually complete with respect to death and emigration [18]. For death, information on date and cause of death was obtained from the nationwide Cause of Death Registry while for emigration information was obtained from the Emigration registry. The start of follow-up was defined as the date of return of the questionnaire. Observation time was calculated from date of entry into the cohort until the date of emigration (first emigration), date of death or 31 December 2010, whichever came first. Codes from the International Classification of Diseases, revisions 9 and 10 (ICD-9, ICD-10), were used (cardiovascular death: ICD-9 codes 410–414, 430–438, and 440; ICD-10 codes I20–I25, I60–I70, and I74; cancer death: ICD-9 codes 140–209 and ICD-10 codes: C00–C97). Statistical analysis The relative risk (RR) of death, comparing exposed and non-exposed women, was estimated by calculating hazard ratios using Cox proportional hazard models with attained age as an underlying time scale [19]. The RR was considered statistically significant when the associated twosided 95 % Wald-type confidence interval (CI) did not include unity, corresponding to a two-sided 5 % level of significance. First we fitted crude models (including only coffee consumption in categories, i.e. 0–1, 2–5, [5 cups/day), and thereafter multivariable models in which we adjusted for the following pre-selected potential confounders assessed at baseline: education (0–10, 11–13, [13 years), BMI (\25, 25–29.9, C30 kg/m2), parity (0, 1, 2, 3, C4), age at first birth (years), smoking habits (never smoker, former smoker of \10, 10–14, 15–19, or C20 cigarettes/day, and current smokers of\10, 10–14, 15–19, or C20 cigarettes/day) and alcohol consumption (g/day). For parity and alcohol we have complete data and for BMI, education, and age at first birth we have almost complete data and thus we excluded women with missing data on these variables. For physical activity, we conducted supplementary analyses where we adjusted for physical activity and compared the results with the original models (see below). We also conducted supplementary analyses. Firstly, we checked whether our associations were affected by adjustment for physical activity level (level 1: sedentary lifestyle, level 3: moderately active lifestyle characterized
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality…
by a few walks per week, and level 5: active lifestyle with exercise a couple of times per week) or diabetes at baseline (yes or no). We also fitted separate models for all-cause, cancer, and cardiovascular mortality occurring before and after 50 years of age in order to evaluate risk factors in different periods of life, e.g. the effect of menopausal status or the accumulated life-long effect of diet. In Sweden, the average age at menopause is 50 years [20]. As coffee consumption and smoking are positively correlated, residual confounding by smoking was of concern; therefore we also fitted adjusted models with and without smoking. To evaluate the functional form of the associations between coffee consumption and all-cause, cancer, cardiovascular and other mortality, we fitted splines for each outcome. In addition, we re-ran our analyses excluding women with a diagnosis of cancer, cardiovascular disease, or diabetes in the Patient Registry up to 2 years before cohort entry. Finally, we re-ran our analyses, restricting follow-up time to 10 years after baseline in order to investigate whether our associations were attenuated by time. As in a previous study [21], we used a coffee consumption of 0–1 cups/day as the reference category, as non-consumers are scarce in Sweden (in this cohort only 9 % of women reported no coffee consumption and only 112 deaths occurred among these non-consumers). We checked the proportional hazards assumption by plotting the Schoenfeld residuals versus time [22]. All statistical analyses were carried out using the SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).
Results Women were followed for an average of 18 years (844, 706 person-years). The baseline characteristics of the study women classified by coffee consumption are shown in Table 1. Women in the two highest categories of coffee consumption (2–5 cups/day and [5 cups/day) were less educated than women in the lowest category (0–1 cups/day). Furthermore, the prevalence of current smokers was higher in the two highest categories of coffee consumption compared to that in the lowest category. During follow-up, there were a total of 1576 deaths: 956 due to cancer, 158 due to cardiovascular disease, and 462 due to other causes. The major reasons for deaths due to other causes were accidents or suicides (34 %), diseases of the nervous system (19 %), and respiratory diseases (13 %). The proportional hazards assumption was verified by graphic evaluation of the plotted Schoenfeld residuals for all models. Table 2 shows the observed associations between coffee consumption and all-cause, cancer, and cardiovascular mortality. In the crude models, a coffee consumption of[5 cups/day was associated with increased risks for all-cause,
cancer, and cardiovascular mortality when compared to a coffee consumption of 0–1 cups/day (all-cause mortality: RR 1.27, 95 % CI 1.08–1.51; cancer mortality: RR 1.61, 1.29–2.03, and cardiovascular mortality: RR 1.60, 95 % CI 0.93–2.78). None of these RRs remained statistically significant after adjustment for confounders and the RRs for all-cause and cardiovascular mortality switched from a positive to an inverse direction after adjustment. When adjusting for confounders, coffee consumption of 2–5 cups/day was associated with a statistically significant, lower risk of all-cause mortality compared to 0–1 cups/day (RR 0.81, 95 % CI 0.69–0.94). The corresponding RR for all-cause mortality for a coffee consumption of [5 cups/day compared 0–1 cups/day was similar, i.e. 0.88, although not statistically significant (95 % CI 0.74–1.05). Coffee consumption was not associated with cancer or cardiovascular mortality in the adjusted models. Finally, coffee consumptions of 2–5 and [5 cups/day were associated with decreased mortality due to other causes compared to 0–1 cups/day (RR 0.59, 95 % CI 0.46–0.75 and RR 0.50, 95 % CI 0.37–0.69, respectively). When we excluded smoking from the analyses, the RRs were similar to those in the crude models for all four outcomes (Table 2). As shown in supplementary Table S1, exclusion of individual confounders from the adjusted models produced estimates similar to those obtained in the fully-adjusted models. Table S2 shows results stratified by age at follow-up. The RRs for all-cause mortality, cancer mortality, cardiovascular mortality, and mortality due to other causes were not statistically significant among women below 50 years of age. Among women older than 50 years of age, coffee consumption of 2–5 cups/day was associated with a statistically significant, lower risk of all-cause mortality compared to 0–1 cups/day. The corresponding RR for a coffee consumption of [5 cups/day was similar, i.e. 0.86, although not statistically significant (95 % CI 0.70–1.06). Furthermore, among women aged 50 years and older, the RR for cancer mortality was 1.06 (0.81–1.38) for 2–5 cups/day and 1.40 (95 % CI 1.05–1.89) for [5 cups/day when compared to 0–1 cups/day. Finally, a coffee consumption of 2–5 and[5 cups/day was associated with lower mortality due to other causes after 50 years of age compared to drinking 0–1 cups/day (RR 0.42, 95 % CI 0.32–0.56 and RR 0.37, 95 % CI 0.26–0.53, respectively). Splines for RRs and 95 % CIs for coffee consumption of 1–7 cups/day in relation to all-cause mortality, cancer mortality, cardiovascular disease mortality, and other cause mortality are shown in Figure S1. There was no support for any substantial change in risk by increasing coffee intake when considering individual cups of coffee compared with the approach aggregating the data into categories of coffee intake.
123
M. Lo¨f et al. Table 1 Baseline characteristics and number of deaths according to categories of coffee consumption
Characteristic
Coffee consumption (cups/day) 0–1
2–5
[5
6766 (15)
29,930 (66)
8444 (19)
Q1
33
35
35
Mean
38
40
40
Q3
43
45
45
Q1
57
58
58
Mean
64
64
66
Q3
69
69
72
Total number of women N (%) Age at enrollment (years)
Body weight (kg)
Height (cm) Q1
162
162
162
Mean
166
166
166
Q3
170
170
170
Q1
20.5
21.0
21.5
Mean
23.2
23.4
24.0
Q3
25.0
25.0
25.5
0–10 years
(1344; 20)
(8612; 29)
(3315; 39)
11–13 years
(2730; 40)
(11,712; 39)
(3279; 39)
[13 years
(2692; 40)
(9606; 32)
(1850; 22)
Q1
0
0
0
Mean
2
4
9
Q3
0
7
17
Body mass index (kg/m2)
Education (n; %)
Number of cigarettes/day
Smoking status (n; %) Never
(4173; 62)
(12,673; 42)
(1810; 22)
Former
(1633; 24)
(9364; 31)
(2399; 28)
Current
(960; 14)
(7893; 26)
(4235; 50)
Alcohol consumption (n; %) \5 g/day
(5553; 82)
(21,901; 73)
(6233; 74)
5–25 g/day
(1188; 18)
(7916; 26)
(2149; 25)
[25 g/day
(25; 1)
(113; 1)
(62; 1)
1
(335; 5)
(978; 3)
(478; 6)
2
(765; 11)
(2962; 10)
(908; 11)
3
(3590; 53)
(17,383; 58)
(4845; 57)
4
(1100; 16)
(5057; 17)
(1227; 15)
5
(557; 8)
(2434; 8)
(664; 8)
Missing
(419;6)
(1116; 4)
(322; 4)
Physical activity level (n; %)
Diabetes (n; %) Yes No
(98; 1.5)
(343; 1.1)
(155; 1.8)
(6668; 98.5)
(29,587; 98.9)
(8289; 98.2)
52 (65)
503 (159)
1103 (325)
Coffee consumption (g/day) Mean (SD)
123
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality… Table 1 continued
Characteristic
Coffee consumption (cups/day) 0–1
2–5
[5
Any death
208
979
389
Any cancer deatha
105
600
251
Any cardiovascular death
18
96
44
85
283
94
Number of deaths
b
Other causes
The Swedish women’s lifestyle and health cohort study Cup 150 ml; n number of women, Q1 1st quartile, Q3 3rd quartile, SD standard deviation Number of deaths due to breast cancer (25 %), lung cancer (19 %), other cancers (56 %)
a
b
Number of deaths due to suicide and accidents (34 %), diseases of the nervous system (19 %), respiratory diseases (14 %), and other (33 %)
Table 2 Relative risks (RR) and 95 % confidence intervals (CI) of coffee consumption in relation to all-cause and cause-specific mortality Cause of death All-cause mortality
Cancer mortality
Cardiovascular disease mortality
Other causes
Cases
Cups of coffee (cups/day)
Crude RR (95 % CI)
Adjusted RR (95 % CI)a
Adjusted RR (95 % CI)—smokingb
208
0–1
Reference
Reference
Reference
979
2–5
0.89 (0.77–1.03)
0.81 (0.69–0.94)
0.88 (0.76–1.03)
389
[5
1.27 (1.08–1.51)
0.88 (0.74–1.05)
1.17 (0.99–1.39)
105
0–1
Reference
Reference
Reference
600
2–5
1.07 (0.87–1.32)
0.99 (0.81–1.23)
1.06 (0.86–1.31)
251
[5
1.61 (1.29–2.03)
1.25 (0.98–1.58)
1.53 (1.21–1.93)
18 96
0–1 2–5
Reference 0.97 (0.59–1.61)
Reference 0.76 (0.46–1.26)
Reference 0.93 (0.56–1.54)
44
[5
1.60 (0.93–2.78)
0.74 (0.42–1.30)
1.35 (0.78–2.36)
85
0–1
Reference
Reference
Reference
283
2–5
0.65 (0.51–0.83)
0.59 (0.46–0.75)
0.65 (0.51–0.83)
94
[5
0.77 (0.58–1.04)
0.50 (0.37–0.69)
0.68 (0.51–0.92)
a
Adjusted for BMI, education, smoking, alcohol, parity and age at first birth
b
Adjusted for BMI, education, alcohol, parity and age at first birth
We found no evidence that the RRs for any of the outcomes were substantially attenuated as follow-up time increased. For instance, the RR for all-cause mortality was 0.87 (95 % CI 0.67–1.13) when a coffee consumption of 2–5 cups/day was compared to 0–1 cups/day within 10 years of follow-up, while the corresponding estimate for a follow-up time of more than 10 years was 0.77 (95 % CI 0.64–0.94). Excluding women that had diabetes, cardiovascular disease or cancer disease 2 years prior to baseline did not affect our results in any major way. For instance, when excluding these women, the RRs for all-cause mortality were 0.79 (95 % CI 0.67–0.93) for 2–5 cups/day and 0.91 (95 % CI 0.76–1.10) for [5 cups/day versus 0–1 cups/day. Finally, adjustment for physical activity level or diabetes at baseline did not affect any of the observed RRs (data not shown).
Discussion In this large and detailed prospective cohort of young and middle-aged Swedish women, a coffee consumption of 2–5 cups/day was associated with a 19 % lower risk of allcause mortality compared to a consumption of 0–1 cups/day. This finding is in agreement with several large, recent studies, including the Nurses’ Health Study and the National Institutes of Health—AARP Diet and Health Study [9, 14]. In the Nurses’ Health Study, a coffee consumption of 2–3, 4–5, and C6 cups/day was associated with an 18, 26 and 17 % lower mortality risk, respectively, when compared to no coffee consumption [14]. The corresponding results for the National Institute of Health— AARP Diet and Health Study were 13, 16, and 15 % [9]. Furthermore, our finding is in agreement with one recent
123
M. Lo¨f et al.
meta-analyses that reported an 18 % lower risk for allcause mortality among women with a coffee consumption of 3 cups/day compared to no consumption when pooling eight prospective cohorts [4]. The RR for all-cause mortality for a coffee consumption of [5 cups/day compared to 0–1 cups/day was 0.88 (95 % CI 0.74–1.05), which was not very different from the corresponding RR for 2–5 cups/day. Thus, we did not find any evidence that increasing coffee consumption above 4 cups/day further reduced the risk of all-cause mortality. This finding is in agreement with the results of the Nurses’ Health Study and the National Institute of Health—AARP Diet and Health Study [9, 14]. It is also consistent with three recent meta-analyses [4–6]. As in many previous prospective cohort studies carried out among women [8, 9, 23–25], coffee consumption was not associated with cancer mortality in our study, which is also in agreement with two recent meta-analyses [4, 6]. Two prospective cohort studies in Japanese women reported some evidence for an inverse association between coffee consumption and cancer mortality [16, 26]. However, the effect of coffee consumption differs by cancer site. Indeed, inverse associations have been reported between coffee consumption and risk of colorectal [27], endometrial [28], and liver cancer [29], while a null association was reported for breast cancer [30]. In this context, it is relevant to note that one-fourth of the cancer deaths in our cohort were due to breast cancer. In supplementary analyses, the estimated RRs for allcause mortality and mortality due to other causes were shifted downwards, but were still statistically significant for women above 50 years of age. In this same age group, there was also evidence of a reverse effect i.e. an increased risk of cancer mortality. The RRs for all-cause and cancer mortality were close to unity and not statistically significant for women below 50 years of age. These patterns may be due to true differences in risk between pre- and postmenopausal women. However, this interpretation may not necessarily be causal due to the possible confounding effect of attained age during follow-up for older women. Thus, these results need confirmation. In this study, the RR for cardiovascular mortality was 0.76 (95 % CI 0.46–1.26) for a coffee consumption of 2–5 cups/day and 0.74 (95 % CI 0.42–1.30) for [5 cups/day when compared to 0–1 cups/day. In comparison, the corresponding RRs from the Nurses’ Health Study for 2–3 and C6 cups/day, compared to no consumption, was 0.75 (95 % CI 0.66–0.86) and 0.81 (95 % CI 0.61–1.06), respectively [14]. The corresponding results from the National Institute of Health—AARP Diet and Health Study were 0.85 (95 % CI 0.76–0.95) and 0.72 (95 % CI 0.59–0.88) [9]. Our findings also agree with a recent meta-analysis that reported a RR for cardiovascular
123
mortality of 0.76 (95 % CI 0.67–0.85) among women with a coffee consumption of 3 cups/day compared to nonconsumers [4]. Thus, it is reasonable to conclude that our results for cardiovascular mortality are similar in magnitude and direction to previous, larger cohorts of women, but our results failed to reach statistical significance, possibly due to limited numbers. When we excluded smoking from the adjusted models, the RRs were similar to the corresponding crude models. Smoking induces changes in hepatic enzymes that increase caffeine clearance, which leads to higher coffee consumption in smokers [31]. Tobacco smoking was indeed the strongest confounder in the recent analyses within the large National Institute of Health—AARP Diet and Health Study [9]. Thus, our results confirm that smoking is a strong confounder of the association between coffee consumption and mortality. Our adjustment for smoking was more detailed than that performed in the studies by Mineharu et al. [32] and Liu et al. [15], which might explain the positive associations they reported between increased coffee consumption and cardiovascular mortality in women. The report by Liu et al. [15] also showed a positive association between coffee consumption and all-cause mortality in men and women younger than 55 years. In a reply to that report, Gretzky et al. suggested that the observed age-dependent effect could be due to the fact that the antioxidant compounds in coffee may change the balance of normal body oxidation in an unfavorable manner in young people, who have lower levels of reactive oxygen species [33]. Our study contradicts these findings because we found decreased all-cause mortality with increased coffee consumption, especially among deaths occurring after 50 years of age. One possible explanation for these contradictory findings is residual confounding by smoking, as Liu et al. [15] assessed smoking status as a binary variable, while we adjusted for the number of cigarettes smoked. Furthermore, the study by Liu et al. stratified by baseline age (\55 years and C55 years), while we fitted models for all-cause and cause-specific mortality occurring before and after 50 years of age, which is superior in order to evaluate if the risk differs by age. It is also relevant to note that the predominant causes of cancer-related deaths in Japanese women are colorectal, stomach and lung cancer while they are breast, colorectal and lung cancer in Swedish women [34]. Finally, we cannot exclude that our observed associations for women above 50 years of age may be due to certain risk backgrounds (for instance genetic predisposition or underlying diseases at baseline). Further studies are needed in order to clarify whether there is a true age-dependent effect of coffee consumption on allcause mortality. The major causes in the category other causes of death were accidents or suicides, diseases of the nervous system,
Prospective study of coffee consumption and all-cause, cancer, and cardiovascular mortality…
and respiratory diseases. There is no obvious explanation for how coffee consumption could reduce mortality from these causes. However, a recent study showed that coffee was associated with lower risk of suicide, perhaps due to an accelerated turnover of serotonin, which is involved in depression [35]. Furthermore, a meta-analysis reported an inverse association between coffee consumption and the risk of Parkinson’s disease [36], possible mechanisms being that caffeine acts as antagonists of the adenosine receptors or activates specific neuroprotection signaling pathways [37]. We cannot exclude the possibility that the association we found with other causes of death was caused by residual confounding with socioeconomic status, dietary risk factors, or health status. We did adjust for socioeconomic status via the variable education; however, we cannot rule out the presence of remaining confounding due to things such as differences in work and life settings. The strengths of our study include its prospective design, with access to very detailed information on possible confounding factors, its large size and the complete follow-up. Misclassification of coffee consumption due to measurement error in the food frequency questionnaire and due to lack of follow-up on coffee consumption is unavoidable, but given the study design it is likely to be non-differential and thus to attenuate any true association. We found no evidence of a substantial attenuation when we evaluated the associations for a follow-up of more than 10 years compared to the first 10 years. A limitation of our study is that we did not have any information on time since smoking cessation for former smokers, nor on age at smoking initiation among current and former smokers. Our results did not change in any major way when we excluded women with diagnosed diabetes, cardiovascular disease or cancer 2 years before baseline. However, one major reason for people to stop drinking coffee may be unspecific gastrointestinal symptoms, such as heartburn, which was not part of our questionnaire and cannot be obtained from medical journals (and are not included in our database for this study). Hence, we cannot completely exclude the possibility of reverse causality, since women may have stopped drinking coffee prior to study enrollment due to poor health. Finally, the number of cardiovascular deaths in our study was low, limiting our power to detect statistically significant associations for this outcome. In conclusion, in this prospective cohort of Swedish women aged 30–49 years at baseline, consuming 2–5 cups of coffee/day was associated with reduced all-cause mortality when compared to a coffee consumption of 0–1 cups/day. Furthermore, our results do not rule out the possibility that coffee consumption may have differential effects on all-cause mortality and cancer mortality before and after 50 years of age.
Acknowledgments Council.
This study was funded by the Swedish Research
Conflict of interest of interest.
The authors declare that they have no conflict
References 1. Higdon JV, Frei B. Coffee and health: a review of recent human research. Crit Rev Food Sci Nutr. 2006;46(2):101–23. doi:10. 1080/10408390500400009. 2. Noordzij M, Uiterwaal CS, Arends LR, Kok FJ, Grobbee DE, Geleijnse JM. Blood pressure response to chronic intake of coffee and caffeine: a meta-analysis of randomized controlled trials. J Hypertens. 2005;23(5):921–8. ´ , Leake DS, Ames JM. In vitro antioxidant ac3. Go´mez-Ruiz JA tivity of coffee compounds and their metabolites. J Agric Food Chem. 2007;55(17):6962–9. doi:10.1021/jf0710985. 4. Crippa A, Discacciati A, Larsson SC, Wolk A, Orsini N. Coffee consumption and mortality from all causes, cardiovascular disease, and cancer: a dose-response meta-analysis. Am J Epidemiol. 2014;180(8):763–75. doi:10.1093/aje/kwu194. 5. Je Y, Giovannucci E. Coffee consumption and total mortality: a meta-analysis of twenty prospective cohort studies. Br J Nutr. 2013;. doi:10.1017/S0007114513003814. 6. 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):527–39. doi:10.1007/s10654-013-9834-7. 7. Zhao Y, Wu K, Zheng J, Zuo R, Li D. Association of coffee drinking with all-cause mortality: a systematic review and metaanalysis. Public Health Nutr. 2014;. doi:10.1017/S1368980014 001438. 8. Andersen LF, Jacobs DR Jr, Carlsen MH, Blomhoff R. 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):1039–46. 9. Freedman ND, Park Y, Abnet CC, Hollenbeck AR, Sinha R. Association of coffee drinking with total and cause-specific mortality. N Engl J Med. 2012;366(20):1891–904. doi:10.1056/ NEJMoa1112010. 10. Happonen P, Laara E, Hiltunen L, Luukinen H. Coffee consumption and mortality in a 14-year follow-up of an elderly northern Finnish population. Br J Nutr. 2008;99(6):1354–61. doi:10.1017/S0007114507871650. 11. Harris HR, Bergkvist L, Wolk A. Coffee and black tea consumption and breast cancer mortality in a cohort of Swedish women. Br J Cancer. 2012;107(5):874–8. doi:10.1038/bjc.2012. 337. 12. 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):305–10. doi:10.1016/j. ypmed.2006.12.011. 13. de Koning Gans JM, Uiterwaal CS, van der Schouw YT, et al. Tea and coffee consumption and cardiovascular morbidity and mortality. Arterioscler Thromb Vasc Biol. 2010;30(8):1665–71. doi:10.1161/ATVBAHA.109.201939. 14. Lopez-Garcia E, van Dam RM, Li TY, Rodriguez-Artalejo F, Hu FB. The relationship of coffee consumption with mortality. Ann Intern Med. 2008;148(12):904–14. doi:10.7326/0003-4819-14812-200806170-00003.
123
M. Lo¨f et al. 15. 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):1066–74. doi:10.1016/j.mayocp.2013.06.020. 16. Tamakoshi A, Lin Y, Kawado M, Yagyu K, Kikuchi S, Iso H. Effect of coffee consumption on all-cause and total cancer mortality: findings from the JACC study. Eur J Epidemiol. 2011;26(4):285–93. doi:10.1007/s10654-011-9548-7. 17. Kumle M, Weiderpass E, Braaten T, Persson I, Adami H. Use of oral contraceptives and breast cancer risk: the Norwegian– Swedish women’s lifestyle and health cohort study. Cancer Epidemiol Biomark Prev. 2002;11:1375–81. 18. Statistics Sweden. http://www.scb.se. Accessed 15 May 2015. 19. Korn E, Graubard B, Midthune D. Time-to-event analysis of longitudinal follow-up of a survey: choice of the time-scale. Am J Epidemiol. 1997;145:72–80. 20. Weiderpass E, Baron J, Adami H. Low-potency oestrogen and risk of endometrial cancer: a case–control study. Lancet. 1999;353:1824–8. 21. Weiderpass E, Sandin S, Lof M, et al. Endometrial cancer in relation to coffee, tea, and caffeine consumption: a prospective cohort study among middle-aged women in Sweden. Nutr Cancer. 2014;66(7):1132–43. doi:10.1080/01635581.2014.948214. 22. Grambsch P, Therneau T. Proportional hazards tests and diagnostics based on weighted residuals. Biometrika. 1994;81:515–26. 23. Gardener H, Rundek T, Wright CB, Elkind MSV, Sacco RL. Coffee and tea consumption are inversely associated with mortality in a multiethnic urban population. J Nutr. 2013;143(8):1299–308. doi:10.3945/jn.112.173807. 24. 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):191–8. 25. Klatsky AL, Armstrong MA, Friedman GD. Coffee, tea, and mortality. Ann Epidemiol. 1993;3(4):375–81. doi:10.1016/10472797(93)90064-B. 26. 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):1007–13. doi:10. 3945/jn.109.109314. 27. Tian C, Wang W, Hong Z. Coffee consumption and risk of colorectal cancer: a dose-response analysis of observational studies. Cancer Causes Control. 2013;24:1265–8. doi:10.1007/s10552013-0200-6.
123
28. Je Y, Giovannucci E. Coffee consumption and risk of endometrial cancer; findings from a large up-to-date meta-analysis. Int J Cancer. 2012;131:1700–10. doi:10.1002/ijc.27408 Epub 2012 Jan 31. 29. Setiawan VW, Wilkens LR, Lu SC, Hernandez BY, Le Marchand L, Henderson BE. Association of coffee intake with reduced incidence of liver cancer and death from chronic liver disease in the US multiethnic cohort. Gastroenterology. 2015;148(1):118–25. doi:10.1053/j.gastro.2014.10.005 quiz e15. 30. Li X, Ren Z, Qin J. Coffee consumption and risk of breast cancer: an up-to-date meta-analysis. PLoS One. 2013;8:e52681. doi:10. 1371/journal.pone.0052681. 31. Wagner MG, Kapoor KG, Wagner AL. Factors affecting the association of coffee consumption with all-cause and cardiovascular disease mortality. Mayo Clin Proc. 2013;88(12):1491–2. doi:10.1016/j.mayocp.2013.10.010. 32. Mineharu Y, Koizumi A, Wada Y. 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):230–40. doi:10.1136/jech.2009. 097311. 33. Greitzer E. Heavy coffee drinking and age-dependent all-cause mortality. Mayo Clin Proc. 2013;88(12):1492–3. doi:10.1016/j. mayocp.2013.10.004. 34. Globocan 2012. Estimated cancer incidence, mortality and prevalence worldwide in 2012. http://globocan.iarc.fr/Default. aspx. Accessed 15 May 2015. 35. 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):377–86. doi:10.3109/15622975.2013.795243. 36. Hernan MA, Takkouche B, Caamano-Isorna F, Gestal-Otero JJ. A meta-analysis of coffee drinking, cigarette smoking, and the risk of Parkinson’s disease. Ann Neurol. 2002;52(3):276–84. doi:10. 1002/ana.10277. 37. Costa J, Lunet N, Santos C, Santos J, Vaz-Carneiro A. Caffeine exposure and the risk of Parkinson’s disease: a systematic review and meta-analysis of observational studies. J Alzheimers Dis. 2010;20(Suppl 1):S221–38. doi:10.3233/JAD-2010-091525.
