Coffee consumption and the risk of colorectal cancer: A prospective cohort study in Japan

Authors

  • Toru Naganuma,

    Corresponding author
    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    • Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, 2-1 Seiryo-machi, Aoba-ku, Sendai 980-8575, Japan
    Search for more papers by this author
    • Fax: +81-22-717-8125.

  • Shinichi Kuriyama,

    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    Search for more papers by this author
  • Munira Akhter,

    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    Search for more papers by this author
  • Masako Kakizaki,

    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    Search for more papers by this author
  • Naoki Nakaya,

    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    Search for more papers by this author
  • Kaori Matsuda-Ohmori,

    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    Search for more papers by this author
  • Taichi Shimazu,

    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    Search for more papers by this author
  • Akira Fukao,

    1. Department of Public Health, Yamagata University Graduate School of Medicine, Yamagata, Japan
    Search for more papers by this author
  • Ichiro Tsuji

    1. Division of Epidemiology, Department of Public Health and Forensic Medicine, Tohoku University Graduate School of Medicine, Sendai, Japan
    Search for more papers by this author

Abstract

An inverse association between coffee consumption and the risk of colorectal cancer has been reported in several case–control studies, but results from prospective cohort studies have been inconclusive. We conducted a prospective cohort study among a Japanese population to clarify the association between coffee consumption and the risk of colorectal cancer incidence. We used data from the Miyagi Cohort Study for this analysis. Usable self-administered questionnaires about coffee consumption were returned from 22,836 men and 24,769 women, aged 40–64 years, with no previous history of cancer. We used the Cox proportional-hazard regression model to estimate hazard ratios and 95% confidence intervals. During 11.6 years of follow-up (425,303 person-years), we identified 457 cases of colorectal cancer. Coffee consumption was not associated with the incidence of colorectal, colon or rectal cancer. The multivariate-adjusted hazard ratio (95% confidence interval) of colorectal cancer incidence for 3 or more cups of coffee per day as compared with no consumption was 0.95 (0.65–1.39) for men and women (p for trend = 0.55), 0.91 (0.56–1.46) for men (p for trend = 0.53) and 1.16 (0.60–2.23) for women (p for trend = 0.996). Coffee consumption was also not associated with incidence of either proximal or distal colon cancer. We conclude that coffee consumption is not associated with the incidence risk of colorectal cancer in the general population in Japan. © 2006 Wiley-Liss, Inc.

Colorectal cancer is one of the most common cancers worldwide.1, 2 In Japan, colorectal cancer ranks second in terms of age-standardized incidence rate among all cancers, and it is increasing rapidly.3 Therefore, primary prevention of colorectal cancer worldwide, including Japan, is a considerable public health concern.

Coffee is considered to be a protective factor against colorectal cancer through activity of its anticarcinogenic constituents, cafestol and kahweol.4 It may also decrease the risk of colorectal cancer by reducing the excretion of bile acids and neutral sterols into the colon.5 Some epidemiological studies have examined the relationship between coffee consumption and the risk of colorectal cancer.6 Sixteen out of 20 case–control studies supported an inverse association between high coffee consumption and colorectal cancer risk.7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26 However, the results of those studies may have been affected by selection bias or recall bias due to retrospective assessment of coffee consumption and other lifestyles-related factors after diagnosis of cancer. In contrast, results from prospective cohort studies have been inconclusive.27, 28, 29, 30, 31, 32, 33, 34, 35, 36 Among 10 such studies, 1 suggested an increasing risk,27 3 showed an inverse association29, 34, 36 and the others observed no association.28, 30, 31, 32, 33, 35 However, as the studies that observed any association had some statistical limitations, the results have to be appraised with caution.

Therefore, we conducted a population-based, prospective cohort study of men and women in Japan, where the consumption of coffee is one of the highest in the world,37 to further clarify the association between coffee consumption and the risk of colorectal cancer incidence.

Abbreviations:

BMI, body mass index; CI, confidence interval; DCO, death certificates only; HR, hazard ratio; HRT, hormone replacement therapy; ICD-O-2, international classification of disease for oncology second edition; SD, standard deviation.

Material and methods

Study cohorts

Our present study was based on the Miyagi Cohort Study, whose study design has been described in detail elsewhere.38 Briefly, all 51,921 residents (25,279 men and 26,642 women) aged 40–64 years living in 14 municipalities, which were randomly selected from the 62 ones of Miyagi Prefecture, Northeastern Japan, were entered into cohort subjects in April 1, 1990. From June through August 1990, we delivered a self-administered questionnaire on various health habits to them. The questionnaires were delivered to, and collected from the subjects' residences by members of health-promotion committees appointed by the municipal governments. Usable questionnaires were returned from 47,605 subjects (22,836 men and 24,769 women) and the response rate was 91.7%. All the residents in study area were entered into our cohort subjects and the response rate of questionnaires was very high; thus, we considered that our subjects were sufficiently representative of this area. The study protocol was approved by the institutional review board of Tohoku University Graduate School of Medicine. We considered the return of self-administered questionnaires signed by the subjects to imply their consent to participate in the study.

Dietary assessment

Dietary intake was assessed by a baseline survey using a self-administered food frequency questionnaire. In this questionnaire, we asked participants to report their frequency of recent consumption of 40 food and beverage items, including coffee. The questionnaire provided 5 categories of response to describe the participant's frequency of coffee consumption: never, occasionally, 1–2 cups per day, 3–4 cups per day, and 5 or more cups per day. No question about the type of coffee and the method used to brew it was asked. The volume of a typical cup of coffee was 150 ml in the study region.

The reproducibility and validity of coffee consumption data among the subjects have been reported previously.39 Spearman's coefficient for the correlation between consumption assessed by the food frequency questionnaire and four 3-day diet records was 0.70, and the correlation between consumption measured by 2 food frequency questionnaires during the 1-year interval was 0.72.

Ascertainment of cases and follow-up

The end point of our analysis was incidence of colorectal cancer defined as the topography code C18.0-C20.9 according to the International Classification of Disease for Oncology, Second Edition (ICD-O-2).

We ascertained the incidence of cancer through computerized record linkage to the Miyagi Prefecture Cancer Registry, one of the oldest and most accurate population-based cancer registries in Japan.40 In this registry, the relevant cases were abstracted from medical records of hospitals by a medical doctor or trained medical record reviewer, except for the cases reported from an institution to the registry. The percentage registered by death certificates only (DCO) for colon and rectum cancer was 9% for men and 14% for women during 1993–1997.40

Of the 47,605 subjects who responded to the questionnaire, we excluded 1,113 (427 men and 686 women) those who were given a diagnosis of cancer before the baseline survey, that we ascertained from self-reports and the cancer registry. Then, we also excluded 7,791 subjects (3,542 men and 4,249 women) who entered incomplete responses about coffee consumption. Consequently, all 38,701 eligible subjects (18,867 men and 19,834 women) including a total of 457 subjects (284 men and 173 women) with colorectal cancer were entered into the analysis.

For follow-up, we established a Follow-up Committee that consisted of Miyagi Cancer Society, the Community Health Divisions of all 14 municipalities, the Department of Health and Welfare of Miyagi Prefectural Government, and the Division of Epidemiology, Tohoku University Graduate School of Medicine. The Committee periodically reviewed the Residential Registration Record of each municipality. With this review, we identified subjects who had either died or emigrated during the observation period. Follow-up of subjects who had moved from the study municipalities was discontinued because the Committee could not review the Residential Registration Record from outside the study area. During the study period, 1,994 subjects (955 men and 1,039 women: 5.2% of the total) were lost to follow-up.

Statistical analysis

We counted person-years of follow-up for each of the subjects from June 1, 1990, until the date of diagnosis of colorectal cancer, the date of emigration from the study area, the date of death or the end of follow-up (December 31, 2001), whichever occurred first. The mean follow-up period was 11.0 years, and the maximum follow-up period was 11.6 years.

We combined the upper 2 categories of coffee consumption (3–4 cups per day and 5 or more cups per day) into the single category “3 or more cups per day” because of the small number of subjects and cases in each category. Incidence rates of colorectal cancer were calculated by dividing the number of incident cases by the number of person-years in each coffee consumption category. Hazard ratio (HR) was computed as the incidence rate among subjects in each coffee consumption category divided by the rate among subjects in the “never” category. We chose the non-intake category (subjects who responded “never” to coffee consumption) as the referent group. We used the Cox proportional-hazard regression model to estimate hazard ratios and 95% confidence intervals (CI) of colorectal cancer incidence according to categories of coffee consumption and to adjust for potentially confounding variables, by using SAS version 9.1 statistical software (SAS Inc., Cary, NC). The p values for the analysis of linear trends were calculated by treating the coffee consumption category as a continuous variable. All reported p values were 2-tailed and were considered statistically significant if less than 0.05. All analyses were conducted for total subjects, and sex separately. We also conducted analysis after dividing colorectal cancer cases into colon (ICD-O-2 code C18.0-18.9) and rectal (C19.0-20.9) cancer. The colon cancer category was further divided into proximal colon (C18.0-18.5) and distal colon (C18.6, 18.7) cancer. Colon cancer cases of code C18.8 (overlapping lesion of colon) and C18.9 (colon, not otherwise specified) were excluded from separate analysis.

We considered the following variables as potential confounders: age (in years), sex, family history of colorectal cancer (yes/no), education level (≤15 years, 16–18 years, ≥19 years), body mass index (BMI) (<18.5 kg/m2, 18.5–24.9 kg/m2, ≥25.0 kg/m2), walking time (≤30 min/day, 30 min/day–1 h/day, ≥1 h/day), cigarette smoking (never smoked, former smoker, current smoker of <20 cigarettes/day, current smoker of ≥20 cigarettes/day), alcohol drinking (never drank, former drinker, current drinker), green tea consumption (never, occasionally, 1–2 cups/day, ≥3 cups/day), black tea consumption (never, occasionally, ≥1 cups/day), consumption of meat, vegetables, fruits (≤1–2 times/month, 1–2 times/week, 3–4 times/week, every day) and total caloric intake (continuous variable, kcal/day). For women, we added menopausal status (premenopausal/postmenopausal), age at menarche (≤13 years, 14–15 years, ≥16 years), number of pregnancies (0 times, 1–3 times, ≥4 times), age at first delivery (≤21 years, 22–25 years, ≥26 years) and number of deliveries (0 births, 1–2 births, ≥3 births).

We performed additional analyses after excluding all subjects who had given a diagnosis of colorectal cancer within the first 3 years of follow-up, and after excluding all subjects with appendiceal cancer (ICD-O-2 code C18.1). We also carried out separate analyses for women after dividing them into premenopausal or postmenopausal status, and including the use of hormone replacement therapy (HRT) (ever/never) into confounders.

Results

The characteristics of the subjects at the baseline, categorized by coffee consumption, are presented in Table I. The proportion of men was higher among subjects who consumed 3 or more cups of coffee per day (60.6%) than among subjects in other coffee consumption categories (46.8–47.0%). The subjects showing higher coffee consumption tended to be younger with a BMI of 18.5 to less than 25.0 kg/m2. Higher coffee consumption was associated with less walking time, a higher frequency of smoking, higher meat consumption and lower vegetable consumption in both men and women. In women, higher coffee consumption was also associated with higher alcohol drinking.

Table I. Characteristics of the Subjects According to Coffee Consumption Status
CharacteristicsCoffee consumption
NeverOccasionally1–2 cups/day≥3 cups/day
  • SD, standard deviation.

  • 1

    The maximum intake of beef, pork and chicken.

  • 2

    The maximum intake of spinach, carrot or pumpkin, tomato and cabbage.

  • 3

    The maximum intake of orange and other fruits.

 Men and Women (n = 38,701)
Subjects (n)6,95514,12912,4525,165
Mean age, years (SD)54.2 (7.0)52.8 (7.2)49.5 (7.2)47.5 (6.7)
Men (%)47.046.847.060.6
Family history of colorectal cancer (%)1.51.71.61.6
Education level (%)
 ≤15 years47.941.034.029.9
 16–18 years41.145.950.152.0
 ≥19 years10.913.115.918.2
Body mass index (%)
 <18.5 kg/m22.72.02.42.6
 18.5-24.9 kg/m265.167.270.672.7
 ≥25.0 kg/m232.330.827.124.7
Walking time (%)
 ≤30 min/day29.828.532.734.9
 30 min/day – 1 hr/day24.024.824.724.3
 ≥1 hr/day46.146.742.640.8
Smoking (%)
 Never55.954.651.334.2
 Former13.712.410.29.6
 Current (<20 cigarettes/day)10.410.510.810.7
 Current (≥20 cigarettes/day)20.122.627.745.6
Alcohol drinking (%)
 Never45.343.040.334.2
 Former6.55.04.96.4
 Current48.252.054.959.4
Green tea consumption (%)
 Never16.47.59.411.1
 Occasionally16.319.521.424.6
 1–2 cups/day19.420.829.026.1
 ≥3 cups/day47.952.240.338.2
Black tea consumption (%)
 Never86.955.751.952.1
 Occasionally11.842.344.343.2
 ≥1 cups/day1.42.03.84.7
Meat consumption1 (%)
 ≤1–2 times/month28.521.117.316.5
 1–2 times/week49.453.153.849.9
 3–4 times/week19.022.825.628.1
 Everyday3.23.13.45.5
Vegetables consumption2 (%)
 ≤1–2 times/month3.42.62.54.0
 1–2 times/week17.316.718.620.1
 3–4 times/week31.936.338.338.0
 Everyday47.444.440.637.9
Fruits consumption3 (%)
 ≤1–2 times/month13.89.39.312.5
 1–2 times/week20.020.620.322.2
 3–4 times/week24.128.129.427.6
 Everyday42.143.041.037.7
Mean total caloric intake, kcal/day (SD)1,587 (604)1,623 (598)1,598 (572)1,795 (604)

During 425,303 person-years of follow-up (205,494 person-years for men and 219,809 person-years for women), we documented 457 colorectal cancer cases (in 284 men and 173 women), which included 281 cases of colon cancer (in 175 men and 106 women) and 180 cases of rectal cancer (in 112 men and 68 women).

Table II presents the association between coffee consumption and the incidence risk of colorectal, colon and rectal cancer. We found no significant association between coffee consumption and the incidence of colorectal, colon or rectal cancer either in men or in women. The multivariate-adjusted hazard ratio (95% CI) of colorectal cancer incidence for 3 or more cups of coffee per day when compared with no consumption was 0.95 (0.65–1.39) for men and women (p for trend = 0.55), 0.91 (0.56–1.46) for men (p for trend = 0.53) and 1.16 (0.60–2.23) for women (p for trend = 0.996).

Table II. Hazard Ratio and 95% Confidence Interval (CI) of Colorectal, Colon, and Rectal Cancer Incidence According to Coffee Consumption
 Coffee consumptionp
NeverOccasionally1–2 cups/day≥3 cups/day
  • 1

    Adjusted for age (in years), sex, family history of colorectal cancer (yes/no), education level (≤15 years, 16–18 years, ≥19 years), body mass index (<18.5 kg/m2, 18.5–24.9 kg/m2, ≥25.0 kg/m2), walking time (≤30 min/day, 30 min/day–1 hr/day, ≥1 h/day), smoking(never smoked, former smoker, current smoker of <20 cigarettes/day, current smoker of ≥20 cigarettes/day), alcohol drinking (never drank, former drinker, current drinker), green tea consumption (never, occasionally, 1–2 cups/day, ≥3 cups/day), black tea consumption (never, occasionally, ≥1 cups/day), consumption of meat, vegetables, fruits (≤1–2 times/month, 1–2 times/week, 3–4 times/week, everyday), and total caloric intake (continuous variable, kcal/day). Models stratified by sex did not include variable for sex. For women, in addition for menopausal status (premenopausal/postmenopausal), age at menarche (≤13 years, 14–15 years, ≥16 years), number of pregnancies (0 times, 1–3 times, ≥4 times), age at first delivery (≤21 years, 22–25 years, ≥26 years), and number of deliveries (0 births, 1–2 births, ≥3 births).

Colorectal cancer
Men and Women
 Person-years75,959155,899137,21956,226 
 Number of incidence cases9319512346 
 Age, sex-adjusted HR (95%CI)1.00 (referent)1.11 (0.87–1.42)1.00 (0.76–1.31)0.98 (0.68–1.41)0.72
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.14 (0.87–1.46)0.98 (0.74–1.32)0.95 (0.65–1.39)0.55
Men
 Person-years35,34672,19463,92334,030 
 Number of incidence cases521208032 
 Age-adjusted HR (95%CI)1.00 (referent)1.20 (0.87–1.66)1.10 (0.77–1.57)0.97 (0.62–1.52)0.85
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.16 (0.82–1.63)1.03 (0.70–1.49)0.91 (0.56–1.46)0.53
Women
 Person-years40,61383,70573,29622,196 
 Number of incidence cases41754314 
 Age-adjusted HR (95%CI)1.00 (referent)0.99 (0.68–1.46)0.84 (0.54–1.31)1.06 (0.57–1.99)0.70
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.13 (0.76–1.70)0.95 (0.59–1.51)1.16 (0.60–2.23)0.996
Colon cancer
Men and Women
 Person-years76,051156,121137,37756,298 
 Number of incidence cases561287225 
 Age, sex-adjusted HR (95%CI)1.00 (referent)1.23 (0.90–1.68)1.02 (0.72–1.46)0.96 (0.59–1.56)0.69
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.26 (0.90–1.76)1.04 (0.72–1.52)0.96 (0.58–1.59)0.66
Men
 Person-years35,37672,34163,99534,077 
 Number of incidence cases33764818 
 Age-adjusted HR (95%CI)1.00 (referent)1.20 (0.80–1.81)1.08 (0.69–1.69)0.92 (0.52–1.65)0.76
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.20 (0.78–1.85)1.06 (0.66–1.72)0.91 (0.49–1.69)0.68
Women
 Person-years40,67583,78173,38222,221 
 Number of incidence cases2352247 
 Age-adjusted HR (95%CI)1.00 (referent)1.26 (0.77–2.06)0.91 (0.51–1.65)1.08 (0.45–2.57)0.79
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.40 (0.83–2.36)1.05 (0.56–1.95)1.16 (0.47–2.88)0.96
Rectal cancer
Men and Women
 Person-years76,131156,385137,46656,309 
 Number of incidence cases38695122 
 Age, sex-adjusted HR (95%CI)1.00 (referent)0.94 (0.63–1.40)0.92 (0.60–1.42)1.00 (0.58–1.72)0.91
 Multivariate-adjusted HR (95%CI)11.00 (referent)0.96 (0.63–1.46)0.88 (0.56–1.39)0.94 (0.53–1.66)0.67
Men
 Person-years35,47072,49364,08434,097 
 Number of incidence cases20453215 
 Age-adjusted HR (95%CI)1.00 (referent)1.15 (0.68–1.95)1.07 (0.61–1.88)1.06 (0.54–2.10)0.95
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.07 (0.61–1.87)0.93 (0.51–1.71)0.92 (0.45–1.90)0.66
Women
 Person-years40,66183,89273,38222,213 
 Number of incidence cases1824197 
 Age-adjusted HR (95%CI)1.00 (referent)0.70 (0.38–1.28)0.74 (0.38–1.45)1.00 (0.40–2.48)0.75
 Multivariate-adjusted HR (95%CI)11.00 (referent)0.84 (0.44–1.60)0.85 (0.42–1.72)1.08 (0.42–2.82)0.94

Coffee consumption was not associated with incidence of either proximal or distal colon cancer (Table III). The multivariate-adjusted hazard ratio (95% CI) for 3 or more cups of coffee per day when compared with no consumption was 1.00 (0.52–1.94) for proximal colon cancer (p for trend = 0.73), and 0.88 (0.37–2.09) for distal colon cancer (p for trend = 0.97).

Table III. Hazard Ratio and 95% Confidence Interval (CI) of Proximal and Distal Colon Cancer Incidence According to Coffee Consumption
 Coffee consumptionp
NeverOccationally1–2 cups/day≥3 cups/day
  • 1

    Adjusted for age (in years), sex, family history of colorectal cancer (yes/no), education level (≤15 years, 16–18 years, ≥19 years), body mass index (<18.5 kg/m2, 18.5–24.9 kg/m2, ≥25.0 kg/m2), walking time (≤30 min/day, 30 min/day–1 hr/day, ≥1 h/day), smoking(never smoked, former smoker, current smoker of <20 cigarettes/day, current smoker of ≥20 cigarettes/day), alcohol drinking (never drank, former drinker, current drinker), green tea consumption (never, occasionally, 1–2 cups/day, ≥3 cups/day), black tea consumption (never, occasionally, ≥1 cups/day), consumption of meat, vegetables, fruits (≤1–2 times/month, 1–2 times/week, 3–4 times/week, every day), and total caloric intake (continuous variable, kcal/day). Models stratified by sex did not include variable for sex. For women, in addition for menopausal status (premenopausal/postmenopausal), age at menarche (≤13 years, 14–15 years, ≥16 years), number of pregnancies (0 times, 1–3 times, ≥4 times), age at first delivery (≤21 years, 22–25 years, ≥26 years), and number of deliveries (0 births, 1–2 births, ≥3 births).

Proximal colon cancer
Men and Women
 Person-years76,136156,392137,52056,338 
 Number of incidence cases33593515 
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.01 (0.64–1.58)0.86 (0.51–1.44)1.00 (0.52–1.94)0.73
Men
 Person-years35,45072,53964,10334,016 
 Number of incidents cases16262312 
 Multivariate-adjusted HR (95%CI)11.00 (referent)0.90 (0.46–1.73)1.04 (0.52–2.08)1.17 (0.51–2.66)0.62
Women
 Person-years40,68583,85473,41722,232 
 Number of incidence cases1733123 
 Multivariate-adjusted1.00 (referent)1.15 (0.61–2.16)0.70 (0.31–1.57)0.69 (0.19–2.52)0.32
Distal colon cancer
Men and Women
 Person-years76,149156,428137,52356,355 
 Number of incidence cases2143308 
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.21 (0.70–2.09)1.21 (0.66–2.21)0.88 (0.37–2.09)0.97
Men
 Person-years35,44272,50564,07834,126 
 Number of incidents cases1636215 
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.13 (0.60–2.14)1.05 (0.21–2.12)0.64 (0.22–1.85)0.51
Women
 Person-years40,70883,92373,44622,228 
 Number of incidence cases51293 
 Multivariate-adjusted HR (95%CI)11.00 (referent)1.65 (0.54–5.03)2.11 (0.64–6.93)2.88 (0.60–13.71)0.14

When we divided women into premenopausal and postmenopausal status, there were no statistically significant associations between coffee consumption and colorectal cancer (data not shown). The risk did not change even after we included the use of HRT into confounders for women (data not shown). After we excluded colorectal cancer cases diagnosed during the first 3 years of follow-up to avoid any possible bias due to the influence of undiagnosed colorectal cancer present at the baseline, the results did not differ essentially (data not shown). We also performed additional analysis after excluding cases of appendiceal cancer, but again the results were not changed appreciably (data not shown).

Discussion

In our analysis of a population-based prospective cohort of Japanese containing 457 cases of colorectal cancer, we observed no association between coffee consumption and the incidence risk of colorectal, colon or rectal cancer. Ten prospective studies have been conducted to assess the relationship between coffee consumption and colorectal cancer.27, 28, 29, 30, 31, 32, 33, 34, 35, 36 Among them, 1 argued an increasing risk,27 3 suggested an inverse association,29, 34, 36 and 6 observed no association.28, 30, 31, 32, 33, 35 Our study was consistent with the latter 6 studies.

Only 1 prospective study has reported a positive association between coffee consumption and colorectal cancer mortality.27 However, in this Seventh-Day Adventists Study, less than 30% of the subjects reported consuming coffee, which was prohibited by their church. Thus, this positive association might have been due to unidentified confounders or selection bias. A prospective cohort study from Norway suggested an inverse association between coffee consumption and the incidence risk of colon cancer, but this relationship was observed only among those who were younger than 65 years of age at the baseline.29 Another prospective cohort study from the USA indicated a lower risk of rectal cancer incidence with higher consumption of decaffeinated coffee, although there was no association between consumption of caffeinated coffee and the risk of colorectal cancer.34 In a study conducted in Japan, daily coffee consumption was associated with a reduced risk of colon cancer in comparison with no or rare consumption only among women (adjusted relative risk = 0.43, 95% CI = 0.22–0.85), but not among men (adjusted relative risk = 0.81, 95% CI = 0.46–1.42).36 For these studies, inverse associations between coffee consumption and the risk of colorectal cancer were observed only in subgroups, and not in all subjects. The subgroups of the subjects showing a significant association were inconsistent among the studies, and thus the results could have been due to chance.

Among prospective studies, no association between coffee consumption and the risk of colorectal cancer was indicated in the USA and Europe. Our findings are meaningful because our study indicated a null result in an Asian population whose diet was different from a “western” one. To our knowledge, this is the largest prospective cohort study on coffee consumption and the risk of colorectal cancer based on an Asian population. Only 1 prospective study of the association between coffee consumption and the risk of colorectal cancer has been carried out in Asia, and this estimated only the risk of colon cancer.36

Our present study had several strengths. It was based on subjects from the general population in Japan, and we identified a large number of colorectal cases (457 cases) during a long follow-up period (425,303 person-years). Furthermore, the validity and reproducibility of coffee consumption among the subjects in our validation study were 0.70 and 0.72, respectively, and thus reasonably high.39

Our study also had some limitations. First, we collected the information on coffee consumption only once before the follow-up period. Therefore, measurement error caused by changes in coffee consumption over time among the subjects could have distorted our result. Second, we excluded 7,791 subjects from our analysis because they incompletely answered, or did not answer, the question on coffee consumption. In this group, 119 cases of colorectal cancer (72 in men and 47 in women) were diagnosed. The multivariate-adjusted hazard ratio of colorectal cancer in the subjects who did not report their coffee consumption (N = 7,791), when compared to those who provided a complete report (N = 38,701), was not statistically significant (HR = 1.16, 95% CI = 0.88–1.52). We also found that the characteristics of the subjects who did not answer the question on coffee consumption were not different from those of subjects who did. There were also no differences in mean age (55.9 and 51.3 years, respectively) or gender (men made up 45.5% and 48.8%, respectively). The 2 groups were also similar with regard to the prevalence of risk factors for colorectal cancer: current alcohol drinkers (53.4% and 53.3%, respectively), current smokers (41.9% and 37.7%, respectively), subjects with a BMI of 25.0 or higher (32.4% and 29.0%, respectively), subjects who walked an hour or more per day (52.2% and 44.5%, respectively) and subjects who ate vegetable everyday (45.7% and 42.8%, respectively). Therefore, our results would not be substantially biased by exclusion of the subjects who did not answer the question on coffee consumption. Third, we did not investigate the type of coffee consumed, such as whether it was filtered or boiled, caffeinated or decaffeinated. Since boiled or decaffeinated coffee is not commonly consumed in Japan, most of the subjects would have drank instant or filtered caffeinated coffee.41 Fourth, the generalizability of our subjects to other population may be limited, because our study was conducted among middle-aged rural Japanese population.

The association of coffee consumption with cancer site, such as the pancreas, liver, bladder and colorectum, has been investigated. Accumulated evidence now suggests that coffee consumption is associated with a decreased risk of liver cancer,39, 42 but has no association with pancreatic cancer43 or bladder cancer.44 Our findings may therefore serve to end the controversial debate about whether there is a relationship between coffee consumption and the risk of colorectal cancer.

In conclusion, we have found that coffee consumption is not associated with a risk of colorectal cancer in the general population in Japan.

Ancillary