Advertisement
Mayo Clinic Proceedings Home

Association Between Caffeine Intake and All-Cause and Cause-Specific Mortality: A Population-Based Prospective Cohort Study

  • Tetsuro Tsujimoto
    Correspondence
    Correspondence: Address to Tetsuro Tsujimoto, MD, PhD, Department of Diabetes, Endocrinology, and Metabolism, Center Hospital, National Center for Global Health and Medicine, 1-21-1 Toyama, Shinjuku-ku, Tokyo 162-8655, Japan.
    Affiliations
    Department of Diabetes, Endocrinology, and Metabolism, Center Hospital, National Center for Global Health and Medicine, Tokyo, Japan
    Search for articles by this author
  • Hiroshi Kajio
    Affiliations
    Department of Diabetes, Endocrinology, and Metabolism, Center Hospital, National Center for Global Health and Medicine, Tokyo, Japan
    Search for articles by this author
  • Takehiro Sugiyama
    Affiliations
    Department of Clinical Study and Informatics, Center for Clinical Sciences, National Center for Global Health and Medicine, Tokyo, Japan

    Department of Public Health/Health Policy, the University of Tokyo, Tokyo, Japan
    Search for articles by this author
Open AccessPublished:July 08, 2017DOI:https://doi.org/10.1016/j.mayocp.2017.03.010

      Abstract

      Objective

      To assess whether caffeine intake is associated with all-cause and cause-specific mortality.

      Patients and Methods

      We conducted a prospective cohort study using data from the National Health and Nutrition Examination Survey 1999-2010. Cox proportional hazards models were used to compare the multivariate-adjusted hazard ratios (HRs) of participants with a caffeine intake of 10 to 99, 100 to 199, and 200 mg/d or more with those of participants with a caffeine intake of less than 10 mg/d.

      Results

      In total, 17,594 participants were included, and the mean ± SD and median (interquartile range) follow-up was 6.5±2.8 years and 6.4 (3.6-9.5) years, respectively; 17,568 participants (99.8%) completed the follow-up, and 1310 died. Compared with those who had a caffeine intake of less than 10 mg/d, HRs and 95% CIs for all-cause mortality were significantly lower in participants with a caffeine intake of 10 to 99 mg/d (HR, 0.81; 95% CI, 0.66-1.00; P=.05), 100 to 199 mg/d (HR, 0.63; 95% CI, 0.51-0.78; P<.001), and 200 or more mg/d (HR, 0.69; 95% CI, 0.58-0.83; P<.001). A similar association was observed in participants who consumed less than 1 cup of coffee per week, and the HR was lowest in those with a caffeine intake of 100 to 199 mg/d (HR, 0.46; 95% CI, 0.22-0.93). There was no association between caffeine intake and cardiovascular mortality, whereas the HRs for noncardiovascular mortality were significantly lower in those with a caffeine intake of 10 to 99 mg/d (HR, 0.74; 95% CI, 0.57-0.95; P=.01), 100 to 199 mg/d (HR, 0.60; 95% CI, 0.46-0.77; P<.001), and 200 or more mg/d (HR, 0.65; 95% CI, 0.53-0.80; P<.001).

      Conclusion

      Moderate caffeine intake was associated with a decreased risk of all-cause mortality, regardless of the presence or absence of coffee consumption.

      Abbreviations and Acronyms:

      BMI (body mass index), HR (hazard ratio), MEC (mobile examination center), NCHS (National Center for Health Statistics), NHANES (National Health and Nutrition Examination Survey), USDA (US Department of Agriculture)
      Coffee is consumed worldwide. Many observational studies have reported a significant inverse association between coffee consumption and all-cause and cause-specific mortality,
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      • Loftfield E.
      • Freedman N.D.
      • Graubard B.I.
      • et al.
      Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
      • Ding M.
      • Satija A.
      • Bhupathiraju S.N.
      • et al.
      Association of coffee consumption with total and cause-specific mortality in 3 large prospective cohorts.
      • Saito E.
      • Inoue M.
      • Sawada N.
      • et al.
      Association of coffee intake with total and cause-specific mortality in a Japanese population: the Japan Public Health Center-based Prospective Study.
      including mortality due to cardiovascular disease,
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      • Loftfield E.
      • Freedman N.D.
      • Graubard B.I.
      • et al.
      Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
      • Ding M.
      • Satija A.
      • Bhupathiraju S.N.
      • et al.
      Association of coffee consumption with total and cause-specific mortality in 3 large prospective cohorts.
      • Saito E.
      • Inoue M.
      • Sawada N.
      • et al.
      Association of coffee intake with total and cause-specific mortality in a Japanese population: the Japan Public Health Center-based Prospective Study.
      respiratory disease,
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      • Loftfield E.
      • Freedman N.D.
      • Graubard B.I.
      • et al.
      Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
      • Saito E.
      • Inoue M.
      • Sawada N.
      • et al.
      Association of coffee intake with total and cause-specific mortality in a Japanese population: the Japan Public Health Center-based Prospective Study.
      infection,
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      • Loftfield E.
      • Freedman N.D.
      • Graubard B.I.
      • et al.
      Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
      neurologic disease,
      • Ding M.
      • Satija A.
      • Bhupathiraju S.N.
      • et al.
      Association of coffee consumption with total and cause-specific mortality in 3 large prospective cohorts.
      and injuries and accidents.
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      In addition, a recent study suggested that higher coffee consumption may be associated with reduced cancer recurrence and death in patients with colon cancer.
      • Guercio B.J.
      • Sato K.
      • Niedzwiecki D.
      • et al.
      Coffee intake, recurrence, and mortality in stage III colon cancer: results from CALGB 89803 (Alliance).
      Caffeine is the most studied compound of coffee, and coffee consumption accounts for 71% of caffeine intake among adults in the United States.
      • Frary C.D.
      • Johnson R.K.
      • Wang M.Q.
      Food sources and intakes of caffeine in the diets of persons in the United States.
      Excessive caffeine intake can lead to an increase in sympathetic activity and circulating catecholamine concentrations mediated by stimulation of the central nervous system,
      • Corti R.
      • Binggeli C.
      • Sudano I.
      • et al.
      Coffee acutely increases sympathetic nerve activity and blood pressure independently of caffeine content: role of habitual versus nonhabitual drinking.
      which could result in significant cardiovascular stress.
      • Svatikova A.
      • Covassin N.
      • Somers K.R.
      • et al.
      A randomized trial of cardiovascular responses to energy drink consumption in healthy adults.
      However, few studies have investigated the association between daily caffeine intake and mortality, and no study has yet evaluated the effects of caffeine intake compared with no caffeine intake. Therefore, the aim of this study was to assess whether caffeine intake is associated with all-cause and cause-specific mortality. In addition, the association between caffeine intake and mortality among participants who did not consume coffee was also investigated.

      Patients and Methods

      Data Source and Study Population

      We conducted a prospective cohort study using data from the US National Health and Nutrition Examination Survey (NHANES).

      Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey. http://www.cdc.gov/nchs/nhanes.htm. Updated January 25, 2017. Accessed June 30, 2015.

      Written informed consent was obtained from all participants. The National Center for Health Statistics (NCHS) Research Ethics Review Board approved the NHANES protocols.

      Centers for Disease Control and Prevention. NCHS Research Ethics Review Board (ERB) Approval. http://www.cdc.gov/nchs/nhanes/irba98.htm. Updated November 5, 2012. Accessed June 30, 2015.

      The NHANES was conducted by the NCHS at the Centers for Disease Control and Prevention and employed a stratified, multistage probability sampling design to enable representation of the US civilian noninstitutionalized population.

      Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey. http://www.cdc.gov/nchs/nhanes.htm. Updated January 25, 2017. Accessed June 30, 2015.

      In this study, data were collected at home and mobile examination centers (MECs). To assess the health and nutritional status of adults and children in the United States, blood specimens were collected during examinations at the MECs.
      Among the population participating in the NHANES during the period 1999-2010, the unweighted response rate of household interviews was 80.6% and that of the MEC examinations was 77.1%.

      Centers for Disease Control and Prevention. Public-use Linked Mortality File. http://www.cdc.gov/nchs/data/datalinkage/Public_use_linked_mortality_file_readme_text_1_2015.pdf. Updated February 2015. Accessed June 30, 2015.

      We focused on participants aged 20 to 79 years on the date of home interview (survey participation), which resulted in a sample of 29,725 participants. We excluded those with missing information on caffeine intake (n = 2700). In addition, those with missing information on any other potential confounders of this study were also excluded (n = 9431), which resulted in a final sample of 17,594. The study participants were prospectively followed-up from the date of survey participation until December 31, 2011.

      Nutrition and Caffeine Intake

      All NHANES participants were eligible for a 24-hour dietary recall interview. Dietary intake data were used to estimate the type and amount of foods and beverages consumed during the 24-hour period before the interview and to estimate the daily total intake of energy, nutrients, and other components from those foods and beverages. Although the NHANES 2003-2010 data set included 2-day dietary information, only the first dietary recall interview data were used to preserve comparability. All dietary recall interviews were conducted in person during examinations at the MECs. To help the respondent report the volume and dimensions of the food items consumed, each MEC dietary interview room contained a standard set of measuring guides. Participants were given measuring cups, spoons, a ruler, and a food model booklet.

      Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey: measuring guides for the dietary recall interview. https://www.cdc.gov/nchs/nhanes/measuring_guides_dri/measuringguides.htm. Updated May 20, 2010. Accessed February 14, 2017.

      The dietary interview component was conducted as a partnership between the US Department of Agriculture (USDA) and the US Department of Health and Human Services. Interview data files were sent electronically from the field and were imported into Survey Net, a computer-assisted food coding and data management system developed by the USDA.
      • Raper N.
      • Perloff B.
      • Ingwersen L.
      • Steinfeldt L.
      • Anand J.
      An overview of USDA's Dietary Intake Data System.
      The USDA dietary data collection instrument, the Automated Multiple-Pass Method,

      Agricultural Research Service, United States Department of Agriculture. AMPM—USDA Automated Multiple-Pass Method. http://www.ars.usda.gov/Services/docs.htm?docid=7710. Updated September 8, 2016. Accessed October 21, 2014.

      was designed to provide an efficient and accurate means of collecting intake data for large-scale national surveys. Sources of caffeine included coffee, tea, soda, energy drinks, and chocolate- and cocoa-containing products, in consideration of caffeinated and decaffeinated versions.
      • Drewnowski A.
      • Rehm C.D.
      Sources of caffeine in diets of US children and adults: trends by beverage type and purchase location.
      Total caffeine intake was calculated, and the amount of caffeine was divided into 4 groups (<10, 10-99, 100-199, and ≥200 mg/d; 1 cup of coffee contains approximately 100 mg of caffeine).

      Outcome Measurements

      The main outcome measure of this study was all-cause mortality. In addition, cardiovascular, noncardiovascular, and cancer mortality were evaluated. We used the mortality follow-up data provided in the Public-use Linked Mortality Files.
      • Albright A.L.
      • Gregg E.W.
      Preventing type 2 diabetes in communities across the U.S.: the National Diabetes Prevention Program.
      The Public-use Linked Mortality Files are available for NHANES for the period 1999-2010 and have been updated through December 31, 2011. We prospectively followed up study participants from the survey participation interview date until the date of death or until December 31, 2011. To identify the causes of death of study participants, the NHANES used the International Classification of Diseases, Tenth Revision, for deaths occurring in or after 1999. The specific codes used in this study were as follows: I00-I09, I11, I13, I20-I51, and I60-I69 for causes of death from cardiovascular disease (cardiovascular death) and C00-C97 for causes of death from malignant neoplasms (cancer death).

      Centers for Disease Control and Prevention. NCHS surveys: 2011 linked mortality files; public-use data dictionary. http://www.cdc.gov/nchs/data/datalinkage/Public-use_Data_Dictionary.pdf. Updated April 16, 2015. Accessed June 30, 2015.

      Follow-up was censored at the time of death from other causes.

      Potential Confounders

      We extracted data on potential confounders: age, sex, race and ethnicity, education, smoking status, and body mass index (BMI; calculated as weight in kilograms divided by height in meters squared), dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, cancer, and total daily intakes of energy, carbohydrate, fat, and protein. Race and ethnicity were classified into 4 categories: non-Hispanic white, non-Hispanic black, Mexican American, and others including other Hispanics, Asians, and multiracial participants. Education was classified as less than high school, more than high school, and high school graduation or general education development certificate. Smoking status was classified into 3 groups: current smoker, former smoker, and never smoked. The BMI was classified into 5 groups: less than 18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and 35.0 kg/m2 or more. Obesity was defined as a BMI of 30.0 kg/m2 or more. Dyslipidemia was defined as a previous diagnosis of hyperlipidemia or intake of lipid-lowering medications. Hypertension was defined as either a previous diagnosis of hypertension or intake of antihypertensive medications. Diabetes was defined as a previous diagnosis of diabetes, use of antidiabetic medications or insulin, or a glycated hemoglobin (A1c) level of 6.5% or higher (to convert to proportion of total hemoglobin, multiply by 0.01). Coronary heart disease was defined as a previous diagnosis of coronary heart disease, myocardial infarction, or angina pectoris. Heart failure was defined as a previous diagnosis of congestive heart failure. Cancer was defined as a previous diagnosis of cancer/malignancy.

      Statistical Analyses

      Demographic data are presented as number and percentage or mean ± SD. Descriptive statistics for patient characteristics were calculated with analysis of variance for continuous variables and χ2 tests for categorical variables. Linear trends across categories of caffeine intake were assessed by linear regression for continuous variables and χ2 tests for categorical variables. For the analyses of the mortality outcomes, multivariate Cox proportional hazards regression was performed, and hazard ratios (HRs) and 95% CIs in participants with caffeine intake of 10-99, 100-199, and 200 mg/d or more were compared with those with a caffeine intake of less than 10 mg/d. P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of less than 10 mg/d to 3 for the highest intake, used as a continuous variable.
      • Loftfield E.
      • Freedman N.D.
      • Graubard B.I.
      • et al.
      Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
      • Saito E.
      • Inoue M.
      • Sawada N.
      • et al.
      Association of coffee intake with total and cause-specific mortality in a Japanese population: the Japan Public Health Center-based Prospective Study.
      The multivariate-adjusted HRs were also analyzed separately for men and women. Multivariate model 1 included adjustments for potential confounders as follows: age, sex, race and ethnicity, education, smoking status, and BMI. Multivariate model 2 included adjustments for the potential confounders of model 1 plus dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, and cancer. Multivariate model 3 included adjustments for the potential confounders of models 1 and 2 plus total daily intakes of energy, carbohydrate, fat, and protein. In addition, to minimize residual confounding, self-reported health conditions (5 categories: excellent, very good, good, fair, or poor) were adjusted for sensitivity analysis. Furthermore, because the present study had missing information about caffeine intake and other potential confounders, we performed a sensitivity analysis for all-cause mortality using multiple imputation to handle missing data. To impute missing values, we used an iterative Markov chain Monte Carlo method, which filled in missing values of one or more variables using multivariate regression.
      • He Y.
      Missing data analysis using multiple imputation: getting to the heart of the matter.
      We constructed models, including all variables in Table 1 that were potentially related to the missing values and outcomes, and analyzed 20 multiply imputed data sets.
      Table 1Baseline Characteristics of the 17,594 Study Participants, Stratified by Daily Caffeine Intake
      GED = general educational development.
      Following the Centers for Disease Control and Prevention's recommendations for the analysis of NHANES data, we used an appropriate weight for each analysis, based on the variables selected. The weights were provided by the National Center for Health Statistics and accounted for unequal probabilities of selection and non-responses to make unbiased national estimates.
      Data are presented as No. (percentage) of participants or mean ± SD.
      CharacteristicCaffeine intake (mg/d)P value for between-group comparison
      P value was calculated with analysis of variance for continuous variables and χ2 tests for categorical variables.
      P value for trend
      Linear trends across categories of caffeine intake were assessed with linear regression for continuous variables and χ2 tests for categorical variables.
      <1010-99100-199≥200
      Unweighted sample3943432037455586
      Age (y)48.1±14.848.2±13.549.4±11.750.4±9.7<.001<.001
      Female sex59.059.256.545.9<.001<.001
      Race and ethnicity
       Non-Hispanic white60.768.875.787.2<.001<.001
       Non-Hispanic black21.113.09.03.8<.001<.001
       Mexican American7.17.15.53.2<.001<.001
       Other
      The category includes other Hispanics and other races including multiracial participants.
      11.111.19.85.8<.001<.001
      Education
       Less than high school17.015.215.213.2<.001<.001
       High school or GED21.323.722.424.4.04.03
       More than high school61.761.162.462.4.69.42
      Smoking status
       Never65.461.854.439.0<.001<.001
       Former22.323.528.633.7<.001<.001
       Current12.314.717.027.3<.001<.001
      Body mass index (kg/m2)29.4±6.628.9±5.828.9±5.428.8±4.7.04.03
       <18.51.81.21.31.0.05.01
       18.5-24.927.828.727.927.6.87.70
       25.0-29.930.833.133.336.7<.001<.001
       30.0-34.920.521.522.120.2.31.59
       ≥35.019.115.515.414.5<.001.001
      Dyslipidemia37.439.241.140.9.01.002
      Hypertension37.834.834.533.2.01.004
      Diabetes13.711.711.610.4.001<.001
      Macrovascular diseases
       Coronary heart disease7.56.66.57.6.25.63
       Heart failure3.12.32.02.4.07.17
       Stroke3.13.02.52.5.24.06
      Cancer9.99.610.510.2.75.49
      Nutrition
       Total energy (kcal)1947±8332072±8062096±7502291±735<.001<.001
       Total carbohydrate (g)242±111258±106255±97268±96<.001<.001
       Total protein (g)71±3878±3879±3590±36<.001<.001
       Total fat (g)78±3780±3580±3188±31<.001<.001
      a GED = general educational development.
      b Following the Centers for Disease Control and Prevention's recommendations for the analysis of NHANES data, we used an appropriate weight for each analysis, based on the variables selected. The weights were provided by the National Center for Health Statistics and accounted for unequal probabilities of selection and non-responses to make unbiased national estimates.
      c Data are presented as No. (percentage) of participants or mean ± SD.
      d P value was calculated with analysis of variance for continuous variables and χ2 tests for categorical variables.
      e Linear trends across categories of caffeine intake were assessed with linear regression for continuous variables and χ2 tests for categorical variables.
      f The category includes other Hispanics and other races including multiracial participants.
      Although the NHANES measured physical activity, this variable was not included in the main analyses owing to inconsistent measurements, which changed between the 2005-2006 and 2007-2008 periods. Therefore, as another sensitivity analysis, physical activity in the NHANES 1999-2006 period was added as an adjustment to model 3. Physical activity was divided into 2 groups according to a cutoff value of 150 min/wk of walking and/or bicycling.
      • Donnelly J.E.
      • Blair S.N.
      • Jakicic J.M.
      • Manore M.M.
      • Rankin J.W.
      • Smith B.K.
      American College of Sports Medicine
      American College of Sports Medicine Position Stand: appropriate physical activity intervention strategies for weight loss and prevention of weight regain for adults.
      • Hankinson A.L.
      • Daviglus M.L.
      • Bouchard C.
      • et al.
      Maintaining a high physical activity level over 20 years and weight gain.
      Furthermore, other sensitivity analyses were performed using the coffee intake data collected in the NHANES period 2003-2006. To exclude the potential effects of coffee components other than caffeine, the association between caffeine intake and mortality was validated in the participants who consumed less than 1 cup of coffee per week. Furthermore, the total amount of caffeine was assessed according to the number of cups of coffee or decaffeinated coffee.
      All statistical analyses were conducted using Stata statistical software, version 14.1 (StataCorp), accounting for the complex survey design. Following the Centers for Disease Control and Prevention's recommendations for the analysis of NHANES data, we used an appropriate weight for each analysis, based on the variables selected. The weights were provided by the NCHS and accounted for unequal probabilities of selection and nonresponses to make unbiased national estimates. P<.05 was considered statistically significant for all tests.

      Results

      The characteristics of all study participants aged 20 to 79 years are presented in Table 1. Among the 17,594 study participants, 3943 (22.4%) had a caffeine intake of less than 10 mg/d, 4320 (24.6%) consumed 10 to 99 mg/d, 3745 (21.3%) consumed 100 to 199 mg/d, and 5586 (31.7%) consumed 200 mg/d or more. Of the 17,594 study participants, 12,008 (68.3%) had a caffeine intake of less than 200 mg/d. Participants with higher caffeine intakes were associated with older age, more proportion of male sex and non-Hispanic white, less proportion of education attainment of less than high school, more smoking, lower proportion of BMI of ≥35.0 kg/m2, higher prevalence of dyslipidemia, lower prevalence of diabetes and hypertension, and higher daily intakes of total energy, carbohydrate, protein, and fat, with P values of <.05 for trends across all categories. Conversely, caffeine intake was not associated with a history of macrovascular diseases or cancer.
      The HRs for all-cause mortality according to daily caffeine intake are shown in Figure 1. The mean ± SD and median (interquartile range) follow-up periods were 6.5±2.8 and 6.4 (3.6-9.5) years, respectively. Of the total study population, 17,568 (99.8%) completed the follow-up, and a total of 1310 deaths were reported. Compared with those with a caffeine intake of less than 10 mg/d, unadjusted HRs (95% CIs) for all-cause mortality were significantly lower in the participants with a caffeine intake of 10 to 99 mg/d (HR, 0.76; 95% CI, 0.61-0.95; P=.01), 100 to 199 mg/d (HR, 0.60; 95% CI, 0.47-0.76; P<.001), and 200 mg/d or more (HR, 0.70; 95% CI, 0.58-0.85; P<.001). The same inverse associations were observed after multivariate adjustment. Compared with participants with a caffeine intake of <10 mg/day, HRs for all-cause mortality in multivariate model 3 were significantly lower in those with a caffeine intake of 10-99 mg/day (HR, 0.81; 95% CI, 0.66-1.00; P=.05), 100-199 mg/day (HR, 0.63; 95% CI, 0.51-0.78; P<.001), and ≥200 mg/day (HR, 0.69; 95% CI, 0.58-0.83; P<.001). In addition, a similar association was observed after multivariate adjustments for the potential confounders of model 3 and health conditions (10-99 mg/d: HR, 0.90; 95% CI, 0.71-1.17; 100-199 mg/d: HR, 0.60; 95% CI, 0.48-0.74; and ≥200 mg/d: HR, 0.75; 95% CI, 0.60-0.94) and after multivariate adjustment for the potential confounders of model 3 and physical activity (10-99 mg/d: HR, 0.85; 95% CI, 0.67-1.08; 100-199 mg/d: HR, 0.68; 95% CI, 0.52-0.87; and ≥200 mg/d: HR, 0.72; 95% CI, 0.59-0.89). The HRs for all-cause mortality in the participants with complications are presented in Table 2. Compared with participants with a caffeine intake of less than 10 mg/d, the lowest HRs for all-cause mortality in multivariate model 3 were found in those with a caffeine intake of 100 to 199 mg/d (HR, 0.64; 95% CI, 0.47-0.88 for those with obesity; HR, 0.75; 95% CI, 0.56-0.98 for those with dyslipidemia; HR, 0.64; 95% CI, 0.52-0.79 for those with hypertension; HR, 0.79; 95% CI, 0.55-1.11 for those with diabetes; and HR, 0.59; 95% CI, 0.37-0.93 for those with cancer). Although a similar association was observed in the participants with diabetes, there were no significant differences between caffeine intake and all-cause mortality (P=.17). In the participants with coronary heart disease and/or heart failure, the HR for all-cause mortality was lowest in those with a caffeine intake of 10 to 99 mg/d (HR, 0.63; 95% CI, 0.42-0.95). The sensitivity analysis using multiple imputation in multivariate model 3 showed similar associations between caffeine intake and all-cause mortality (10-99 mg/d: HR, 0.83; 95% CI, 0.71-0.96; 100-199 mg/d: HR, 0.63; 95% CI, 0.52-0.75; and ≥200 mg/d: HR, 0.75; 95% CI, 0.64-0.87 compared with participants with a caffeine intake of <10 mg/d).
      Figure thumbnail gr1
      Figure 1Hazard ratios for all-cause mortality according to daily caffeine intake. The reference line at 1.0 corresponds to the reference group with a caffeine intake of less than 10 mg/d.
      Table 2Hazard Ratios for All-Cause Mortality of the 17,594 Study Participants, Stratified by Daily Caffeine Intake
      CHD = coronary heart disease; HF = heart failure; HR = hazard ratio; NA = not applicable; ref = reference.
      ,
      Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      VariableCaffeine (mg/d)P value for trend
      P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      <1010-99100-199≥200
      All participants3943432037455586
       Deaths from any cause343311247409NA
       Event rate (per 1000 person-years)12.29.37.28.6NA
       Unadjusted HR1.00 [ref]0.76 (0.61-0.95)0.60 (0.47-0.76)0.70 (0.58-0.85).001
       Multivariate-adjusted HR, model 1
      Multivariate model 1 included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2).
      1.00 [ref]0.80 (0.65-0.99)0.61 (0.48-0.77)0.65 (0.54-0.78)<.001
       Multivariate-adjusted HR, model 2
      Multivariate model 2 included adjustments for the potential confounders of model 1 plus dyslipidemia, hypertension, diabetes, CHD, HF, stroke, and cancer.
      1.00 [ref]0.81 (0.65-1.00)0.63 (0.50-0.79)0.68 (0.56-0.82)<.001
       Multivariate-adjusted HR, model 3
      Multivariate model 3 included adjustments for the potential confounders of models 1 and 2 plus total daily intake of energy, carbohydrates, fat, and protein. Multivariate analyses limited to participants with obesity, dyslipidemia, hypertension, diabetes, CHD/HF, or cancer did not include the confounder body mass index, dyslipidemia, hypertension, diabetes, CHD/HF, or cancer, respectively.
      1.00 [ref]0.81 (0.66-1.00)0.63 (0.51-0.78)0.69 (0.58-0.83)<.001
      Participants with obesity1644172414942097
       Deaths from any cause13311195150NA
       Event rate (per 1000 person-years)13.49.67.89.5NA
       Unadjusted HR1.00 [ref]0.72 (0.51-1.00)0.58 (0.42-0.79)0.70 (0.52-0.93).02
       Multivariate-adjusted HR, model 11.00 [ref]0.78 (0.53-1.15)0.59 (0.41-0.83)0.65 (0.46-0.90).007
       Multivariate-adjusted HR, model 21.00 [ref]0.84 (0.58-1.21)0.63 (0.46-0.86)0.69 (0.50-0.96).01
       Multivariate-adjusted HR, model 31.00 [ref]0.85 (0.59-1.22)0.64 (0.47-0.88)0.72 (0.52-0.99).02
      Participants with dyslipidemia1550178316532378
       Deaths from any cause154153136191NA
       Event rate (per 1000 person-years)14.312.19.610.5NA
       Unadjusted HR1.00 [ref]0.85 (0.62-1.16)0.67 (0.50-0.89)0.73 (0.56-0.95).01
       Multivariate-adjusted HR, model 11.00 [ref]0.91 (0.66-1.26)0.70 (0.53-0.92)0.71 (0.54-0.92).006
       Multivariate-adjusted HR, model 21.00 [ref]0.90 (0.66-1.23)0.73 (0.55-0.97)0.74 (0.57-0.97).01
       Multivariate-adjusted HR, model 31.00 [ref]0.90 (0.66-1.24)0.75 (0.56-0.98)0.75 (0.57-0.98).02
      Participants with hypertension1672170514962107
       Deaths from any cause227200153235NA
       Event rate (per 1000 person-years)21.717.213.015.5NA
       Unadjusted HR1.00 [ref]0.79 (0.59-1.04)0.61 (0.46-0.78)0.71 (0.56-0.90).004
       Multivariate-adjusted HR, model 11.00 [ref]0.83 (0.64-1.09)0.63 (0.48-0.81)0.71 (0.55-0.91).003
       Multivariate-adjusted HR, model 21.00 [ref]0.83 (0.64-1.07)0.64 (0.51-0.79)0.71 (0.56-0.91).003
       Multivariate-adjusted HR, model 31.00 [ref]0.84 (0.64-1.08)0.64 (0.52-0.79)0.72 (0.56-0.92).003
      Participants with diabetes749706638808
       Deaths from any cause1239881115NA
       Event rate (per 1000 person-years)26.723.118.624.8NA
       Unadjusted HR1.00 [ref]0.86 (0.58-1.27)0.70 (0.48-1.02)0.93 (0.66-1.31).67
       Multivariate-adjusted HR, model 11.00 [ref]0.83 (0.56-1.21)0.67 (0.47-0.95)0.84 (0.59-1.18).29
       Multivariate-adjusted HR, model 21.00 [ref]0.90 (0.62-1.30)0.78 (0.55-1.12)0.85 (0.61-1.18).28
       Multivariate-adjusted HR, model 31.00 [ref]0.90 (0.62-1.30)0.79 (0.55-1.11)0.84 (0.60-1.17).26
      Participants with CHD and/or HF370398363572
       Deaths from any cause1007579117NA
       Event rate (per 1000 person-years)50.031.235.832.2NA
       Unadjusted HR1.00 [ref]0.61 (0.42-0.87)0.71 (0.49-1.04)0.62 (0.43-0.89).03
       Multivariate-adjusted HR, model 11.00 [ref]0.60 (0.40-0.89)0.70 (0.48-1.01)0.69 (0.50-0.96).08
       Multivariate-adjusted HR, model 21.00 [ref]0.64 (0.43-0.96)0.73 (0.50-1.05)0.70 (0.50-0.98).08
       Multivariate-adjusted HR, model 31.00 [ref]0.63 (0.42-0.95)0.72 (0.49-1.05)0.72 (0.51-0.99).11
      Participants with cancer349389371624
       Deaths from any cause716362101NA
       Event rate (per 1000 person-years)36.923.119.026.2NA
       Unadjusted HR1.00 [ref]0.63 (0.37-1.05)0.51 (0.32-0.80)0.70 (0.47-1.05).18
       Multivariate-adjusted HR, model 11.00 [ref]0.64 (0.40-1.02)0.55 (0.35-0.87)0.71 (0.46-1.09).21
       Multivariate-adjusted HR, model 21.00 [ref]0.65 (0.40-1.06)0.57 (0.36-0.91)0.73 (0.49-1.10).20
       Multivariate-adjusted HR, model 31.00 [ref]0.66 (0.40-1.07)0.59 (0.37-0.93)0.72 (0.48-1.10).20
      a CHD = coronary heart disease; HF = heart failure; HR = hazard ratio; NA = not applicable; ref = reference.
      b Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      c P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      d Multivariate model 1 included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2).
      e Multivariate model 2 included adjustments for the potential confounders of model 1 plus dyslipidemia, hypertension, diabetes, CHD, HF, stroke, and cancer.
      f Multivariate model 3 included adjustments for the potential confounders of models 1 and 2 plus total daily intake of energy, carbohydrates, fat, and protein. Multivariate analyses limited to participants with obesity, dyslipidemia, hypertension, diabetes, CHD/HF, or cancer did not include the confounder body mass index, dyslipidemia, hypertension, diabetes, CHD/HF, or cancer, respectively.
      The HRs for cause-specific mortality are presented in Table 3. There was no significant association between caffeine intake and cardiovascular mortality (P=.22), whereas an inverse association was observed between caffeine intake and noncardiovascular mortality (P=.002). The adjusted HRs for noncardiovascular mortality in multivariate model 3 were significantly lower in the participants with a caffeine intake of 10 to 99 mg/d (HR, 0.74; 95% CI, 0.57-0.95), 100 to 199 mg/d (HR, 0.60; 95% CI, 0.46-0.77), and 200 mg/d or more (HR, 0.65; 95% CI, 0.53-0.80) (all P<.001), compared with those with a caffeine intake of less than 10 mg/d. The HRs for cancer mortality were nonsignificantly lower in the participants with a caffeine intake of 10 to 99 mg/d (HR, 0.86; 95% CI, 0.56-1.31), 100 to 199 mg/d (HR, 0.69; 95% CI, 0.47-1.04), and 200 mg/d or more (HR, 0.70; 95% CI, 0.50-1.00) (all P=.04). Similar HRs for cardiovascular, noncardiovascular, and cancer mortalities were observed after multivariate adjustment with physical activity (Table 4).
      Table 3Hazard Ratios for Cardiovascular, Noncardiovascular, and Cancer Mortality in 17,594 Study Participants
      HR = hazard ratio; NA = not applicable; ref = reference.
      ,
      Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      VariableCaffeine (mg/d)P value for trend
      P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      <10 (n=3943)10-99 (n=4320)100-199 (n=3745)≥200 (n=5586)
      Cardiovascular mortality
       Deaths from any cause70726398NA
       Event rate (per 1000 person-years)2.22.31.61.9NA
       Unadjusted HR1.00 [ref]1.05 (0.66-1.67)0.71 (0.45-1.12)0.84 (0.55-1.26).22
       Multivariate-adjusted HR, model 1
      Multivariate model 1 included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2).
      1.00 [ref]1.13 (0.69-1.84)0.73 (0.46-1.16)0.77 (0.50-1.18).08
       Multivariate-adjusted HR, model 2
      Multivariate model 2 included adjustments for the potential confounders of model 1 plus dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, and cancer.
      1.00 [ref]1.16 (0.73-1.85)0.78 (0.49-1.24)0.83 (0.54-1.25).15
       Multivariate-adjusted HR, model 3
      Multivariate model 3 included adjustments for the potential confounders of models 1 and 2 plus total daily intake of energy, carbohydrates, fat, and protein.
      1.00 [ref]1.16 (0.73-1.84)0.80 (0.51-1.27)0.88 (0.58-1.33).29
      Noncardiovascular mortality
       Deaths from any cause271237184309NA
       Event rate (per 1000 person-years)9.96.95.76.7NA
       Unadjusted HR1.00 [ref]0.70 (0.53-0.91)0.57 (0.43-0.76)0.67 (0.53-0.84).002
       Multivariate-adjusted HR, model 11.00 [ref]0.73 (0.57-0.94)0.59 (0.44-0.77)0.62 (0.50-0.78)<.001
       Multivariate-adjusted HR, model 21.00 [ref]0.74 (0.58-0.95)0.61 (0.46-0.79)0.68 (0.55-0.84).001
       Multivariate-adjusted HR, model 31.00 [ref]0.74 (0.57-0.95)0.60 (0.46-0.77)0.65 (0.53-0.80)<.001
      Cancer mortality
       Deaths from any cause919175121NA
       Event rate (per 1000 person-years)3.12.62.22.5NA
       Unadjusted HR1.00 [ref]0.82 (0.56-1.20)0.72 (0.50-1.02)0.79 (0.60-1.05).15
       Multivariate-adjusted HR, model 11.00 [ref]0.86 (0.57-1.29)0.70 (0.47-1.04)0.68 (0.47-0.97).02
       Multivariate-adjusted HR, model 21.00 [ref]0.84 (0.55-1.29)0.68 (0.46-1.01)0.67 (0.48-0.96).02
       Multivariate-adjusted HR, model 31.00 [ref]0.86 (0.56-1.31)0.69 (0.47-1.04)0.70 (0.50-1.00).04
      a HR = hazard ratio; NA = not applicable; ref = reference.
      b Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      c P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      d Multivariate model 1 included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2).
      e Multivariate model 2 included adjustments for the potential confounders of model 1 plus dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, and cancer.
      f Multivariate model 3 included adjustments for the potential confounders of models 1 and 2 plus total daily intake of energy, carbohydrates, fat, and protein.
      Table 4Hazard Ratios for Cardiovascular, Noncardiovascular, and Cancer Mortality After Multivariate Adjustment With Physical Activity
      HR = hazard ratio; NA = not applicable; ref = reference.
      ,
      Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      VariableCaffeine (mg/d)P value for trend
      P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      <10 (n=2375)10-99 (n=2476)100-199 (n=2139)≥200 (n=3330)
      Cardiovascular mortality
       Deaths from any cause59544984NA
       Event rate (per 1000 person-years)2.32.41.72.0NA
       Multivariate-adjusted HR
      Multivariate model included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2), dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, cancer, total daily intake of energy, carbohydrate, fat, and protein, and physical activity.
      1.00 [ref]1.12 (0.67-1.88)0.83 (0.49-1.42)0.86 (0.53-1.39).34
      Noncardiovascular mortality
       Deaths from any cause210189146246NA
       Event rate (per 1000 person-years)10.17.56.07.0NA
       Multivariate-adjusted HR1.00 [ref]0.78 (0.58-1.05)0.64 (0.48-0.87)0.69 (0.54-0.88).004
      Cancer mortality
       Deaths from any cause73746593NA
       Event rate (per 1000 person-years)3.22.92.52.6NA
       Multivariate-adjusted HR1.00 [ref]0.88 (0.54-1.44)0.74 (0.48-1.16)0.74 (0.50-1.09).11
      a HR = hazard ratio; NA = not applicable; ref = reference.
      b Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      c P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      d Multivariate model included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2), dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, cancer, total daily intake of energy, carbohydrate, fat, and protein, and physical activity.
      Kaplan-Meier survival curves for all-cause mortality in men and women are presented in the Supplemental Figure (available online at http://www.mayoclinicproceedings.org), and the HRs for all-cause and cause-specific mortality in men and women are shown in Table 5. Compared with those with a caffeine intake of less than 10 mg/d, the adjusted HRs for all-cause mortality in multivariate model 3 were lower for men and women with a caffeine intake of 10 to 99 mg/d (men: HR, 0.90; 95% CI, 0.66-1.21; women: HR, 0.74; 95% CI, 0.53-1.02), 100 to 199 mg/d (men: HR, 0.71; 95% CI, 0.54-0.94; women: HR, 0.53; 95% CI, 0.37-0.77), and 200 mg/d or more (men: HR, 0.79; 95% CI, 0.59-1.05; women: HR, 0.58; 95% CI, 0.41-0.81). Although not all HRs were statistically significant, they were the lowest and significantly lower in both men (HR, 0.71; 95% CI, 0.54-0.94; P =.01) and women (HR, 0.53; 95% CI, 0.37-0.76; P=.001) with a caffeine intake of 100 to 199 mg/d compared with those with a caffeine intake of less than 10 mg/d. The HRs for all-cause, cardiovascular, and noncardiovascular mortality in multivariate model 3 were the lowest in those with a caffeine intake of 100 to 199 mg/d and lower in women than in men (all-cause mortality: HR, 0.71; 95% CI, 0.54-0.94 for men and HR, 0.53; 95% CI, 0.37-0.77 for women; cardiovascular mortality: HR, 1.03; 95% CI, 0.49-2.14 for men and HR, 0.48; 95% CI, 0.23-1.02 for women; noncardiovascular mortality: HR, 0.64; 95% CI, 0.48-0.85 for men and HR, 0.54; 95% CI, 0.36-0.82 for women). In the model that included the interaction term between sex and caffeine intake (<10 or ≥10 mg/d), we found that the association between caffeine intake and all-cause mortality was not significantly interacted by sex (P=.11 for interaction term).
      Table 5Hazard Ratios for All-Cause, Cardiovascular, and Noncardiovascular Mortality in Men and Women
      HR = hazard ratio; NA = not applicable; ref = reference.
      ,
      Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      VariableCaffeine (mg/d)P value for trend
      P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      <1010-99100-199≥200
      Men1576177217163106NA
       All-cause mortality
      Deaths from any cause181177144279NA
      Event rate (per 1000 person-years)13.411.48.910.6NA
      Unadjusted HR1.00 [ref]0.84 (0.62-1.13)0.66 (0.49-0.89)0.78 (0.59-1.02).07
      Multivariate-adjusted HR, model 1
      Multivariate model 1 included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2).
      1.00 [ref]0.89 (0.65-1.20)0.68 (0.51-0.91)0.74 (0.55-1.01).03
      Multivariate-adjusted HR, model 2
      Multivariate model 2 included adjustments for the potential confounders of model 1 plus dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, and cancer.
      1.00 [ref]0.89 (0.65-1.20)0.71 (0.54-0.94)0.78 (0.58-1.04).07
      Multivariate-adjusted HR, model 3
      Multivariate model 3 included adjustments for the potential confounders of models 1 and 2 plus total daily intake of energy, carbohydrates, fat, and protein.
      1.00 [ref]0.90 (0.66-1.21)0.71 (0.54-0.94)0.79 (0.59-1.05).07
       Cardiovascular mortality
      Deaths from any cause40454078NA
      Event rate (per 1000 person-years)2.63.42.32.6NA
      Unadjusted HR1.00 [ref]1.31 (0.66-2.60)0.89 (0.42-1.87)0.98 (0.51-1.89).59
      Multivariate-adjusted HR, model 11.00 [ref]1.47 (0.75-2.87)0.95 (0.46-1.96)1.03 (0.52-2.00).62
      Multivariate-adjusted HR, model 21.00 [ref]1.51 (0.79-2.89)1.03 (0.50-2.12)1.10 (0.58-2.09).80
      Multivariate-adjusted HR, model 31.00 [ref]1.49 (0.77-2.89)1.03 (0.49-2.14)1.15 (0.60-2.23).96
       Noncardiovascular mortality
      Deaths from any cause139130104201NA
      Event rate (per 1000 person-years)10.87.96.78.1NA
      Unadjusted HR1.00 [ref]0.72 (0.51-1.02)0.61 (0.46-0.82)0.74 (0.54-0.99).09
      Multivariate-adjusted HR, model 11.00 [ref]0.75 (0.52-1.07)0.62 (0.46-0.85)0.68 (0.47-0.98).04
      Multivariate-adjusted HR, model 21.00 [ref]0.73 (0.51-1.06)0.63 (0.47-0.85)0.70 (0.50-0.99).05
      Multivariate-adjusted HR, model 31.00 [ref]0.75 (0.52-1.08)0.64 (0.48-0.85)0.70 (0.50-0.99).05
       Cancer mortality
      Deaths from any cause53534881NA
      Event rate (per 1000 person-years)4.13.42.93.1NA
      Unadjusted HR1.00 [ref]0.83 (0.49-1.40)0.71 (0.44-1.14)0.75 (0.48-1.18).23
      Multivariate-adjusted HR, model 11.00 [ref]0.85 (0.48-1.50)0.72 (0.42-1.22)0.68 (0.39-1.18).15
      Multivariate-adjusted HR, model 21.00 [ref]0.83 (0.47-1.50)0.72 (0.42-1.22)0.68 (0.40-1.17).15
      Multivariate-adjusted HR, model 31.00 [ref]0.85 (0.47-1.54)0.74 (0.43-1.26)0.72 (0.41-1.27).24
      Women2367254820292480NA
       All-cause mortality
      Deaths from any cause162134103130NA
      Event rate (per 1000 person-years)11.37.86.06.4NA
      Unadjusted HR1.00 [ref]0.70 (0.49-1.00)0.53 (0.37-0.76)0.55 (0.40-0.77).001
      Multivariate-adjusted HR, model 11.00 [ref]0.73 (0.53-1.03)0.55 (0.38-0.79)0.54 (0.38-0.77).001
      Multivariate-adjusted HR, model 21.00 [ref]0.73 (0.53-1.03)0.53 (0.37-0.76)0.56 (0.40-0.80).002
      Multivariate-adjusted HR, model 31.00 [ref]0.74 (0.53-1.02)0.53 (0.37-0.77)0.58 (0.41-0.81).002
       Cardiovascular mortality
      Deaths from any cause30272320NA
      Event rate (per 1000 person-years)2.01.51.01.1NA
      Unadjusted HR1.00 [ref]0.79 (0.38-1.61)0.52 (0.25-1.05)0.54 (0.28-1.01).03
      Multivariate-adjusted HR, model 11.00 [ref]0.84 (0.41-1.67)0.52 (0.25-1.08)0.49 (0.25-0.97).02
      Multivariate-adjusted HR, model 21.00 [ref]0.86 (0.43-1.72)0.48 (0.22-1.04)0.48 (0.25-0.93).01
      Multivariate-adjusted HR, model 31.00 [ref]0.85 (0.43-1.68)0.48 (0.23-1.02)0.53 (0.28-1.01).03
       Noncardiovascular mortality
      Deaths from any cause13210780108NA
      Event rate (per 1000 person-years)9.36.24.95.2NA
      Unadjusted HR1.00 [ref]0.68 (0.45-1.02)0.53 (0.35-0.81)0.55 (0.37-0.82).005
      Multivariate-adjusted HR, model 11.00 [ref]0.72 (0.49-1.04)0.56 (0.36-0.85)0.55 (0.36-0.83).005
      Multivariate-adjusted HR, model 21.00 [ref]0.71 (0.49-1.03)0.53 (0.35-0.81)0.57 (0.38-0.85).008
      Multivariate-adjusted HR, model 31.00 [ref]0.71 (0.49-1.03)0.54 (0.36-0.82)0.58 (0.39-0.86).008
       Cancer mortality
      Deaths from any cause38382740NA
      Event rate (per 1000 person-years)2.52.01.71.8NA
      Unadjusted HR1.00 [ref]0.80 (0.45-1.43)0.70 (0.38-1.30)0.72 (0.42-1.23).23
      Multivariate-adjusted HR, model 11.00 [ref]0.83 (0.45-1.50)0.67 (0.34-1.30)0.66 (0.37-1.17).14
      Multivariate-adjusted HR, model 21.00 [ref]0.80 (0.44-1.45)0.59 (0.31-1.13)0.66 (0.37-1.18).13
      Multivariate-adjusted HR, model 31.00 [ref]0.82 (0.45-1.50)0.62 (0.32-1.20)0.69 (0.38-1.24).17
      a HR = hazard ratio; NA = not applicable; ref = reference.
      b Data are presented as number or hazard ratio (95% CI). Boldface type indicates a significant difference (P<.05).
      c P values for linear trend were calculated by assigning scores for categories of caffeine intake, starting from 0 for a caffeine intake of 10 mg/d to 3 for the highest intake, used as a continuous variable.
      d Multivariate model 1 included adjustments for potential confounders as follows: age, sex, race and ethnicity (non-Hispanic white, non-Hispanic black, Mexican American, and others), education (less than high school, high school graduation or general education development certificate, and more than high school), current smoking status (current smoker, former smoker, and never smoked), and body mass index (<18.5, 18.5-24.9, 25.0-29.9, 30.0-34.9, and ≥35.0 kg/m2).
      e Multivariate model 2 included adjustments for the potential confounders of model 1 plus dyslipidemia, hypertension, diabetes, coronary heart disease, heart failure, stroke, and cancer.
      f Multivariate model 3 included adjustments for the potential confounders of models 1 and 2 plus total daily intake of energy, carbohydrates, fat, and protein.
      The associations between coffee consumption and caffeine intake are presented in Figure 2. Although daily caffeine intake increased with coffee consumption, about 90 mg/d of caffeine intake was found in the participants who consumed no coffee or less than 1 cup of coffee per day. The proportion of participants with a caffeine intake of less than 10 mg/d was only 30% to 35% in those who consumed no coffee or less than 1 cup of coffee per day. In addition, although daily caffeine intake increased with decaffeinated coffee consumption, participants who consumed 2 or more cups of decaffeinated coffee per day had a caffeine intake of 100 mg/d or more. Only about 40% of participants who consumed 1 or less cups of decaffeinated coffee per day had a caffeine intake of less than 10 mg/d. To exclude the potential effects of coffee components other than caffeine, the association between caffeine intake and all-cause mortality was assessed in the participants who consumed less than 1 cup of coffee per week (Figure 3). Similar associations between caffeine intake and mortality were observed in those participants. Compared with participants with a caffeine intake of <10 mg/day, the adjusted HRs for all-cause mortality in multivariate model 3 was the lowest and significantly lower in those with a caffeine intake of 100 to 199 mg/day (10-99 mg/d: HR, 1.06; 95% CI, 0.58-1.94; P=.84; 100-199 mg/d: HR, 0.46; 95% CI, 0.22-0.93; P=.03: and ≥200 mg/d: HR, 1.45; 95% CI, 0.45-4.65; P=.51).
      Figure thumbnail gr2
      Figure 2Association between coffee consumption and caffeine intake. Caffeine intake (A) and proportion of caffeine intake of less than 10 mg/d (B) according to coffee consumption. Caffeine intake (C) and proportion of caffeine intake of less than 10 mg/d (D) according to decaffeinated coffee consumption.
      Figure thumbnail gr3
      Figure 3Hazard ratios for all-cause mortality according to daily caffeine intake in participants who consumed less than 1 cup of coffee per week. The reference line at 1.0 corresponds to the reference group with a caffeine intake of less than 10 mg/d.

      Discussion

      Our analyses of nationally representative data collected from the NHANES indicated that caffeine intake was associated with a decreased risk of all-cause mortality, particularly noncardiovascular mortality. Meanwhile, there was no association between caffeine intake and cardiovascular mortality. These results indicated that competing risk might not be an explanation for decreased noncardiovascular mortality. The HR for all-cause mortality was lowest in the participants with moderate caffeine intake of 100 to 199 mg/d, and the HRs for all-cause and cause-specific mortality were lower for women than men. Furthermore, a similar association between caffeine intake and all-cause mortality was observed in the participants who consumed less than 1 cup of coffee per week.
      Many studies have suggested that coffee consumption is inversely associated with all-cause and cause-specific mortality.
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      • Loftfield E.
      • Freedman N.D.
      • Graubard B.I.
      • et al.
      Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
      • Ding M.
      • Satija A.
      • Bhupathiraju S.N.
      • et al.
      Association of coffee consumption with total and cause-specific mortality in 3 large prospective cohorts.
      • Saito E.
      • Inoue M.
      • Sawada N.
      • et al.
      Association of coffee intake with total and cause-specific mortality in a Japanese population: the Japan Public Health Center-based Prospective Study.
      Coffee is a complex beverage that contains more than 1000 compounds.
      • O'Keefe J.H.
      • Bhatti S.K.
      • Patil H.R.
      • DiNicolantonio J.J.
      • Lucan S.C.
      • Lavie C.J.
      Effects of habitual coffee consumption on cardiometabolic disease, cardiovascular health, and all-cause mortality.
      Several studies have reported the beneficial effects of several coffee components, including caffeic acid,
      • Olthof M.R.
      • Hollman P.C.
      • Zock P.L.
      • Katan M.B.
      Consumption of high doses of chlorogenic acid, present in coffee, or of black tea increases plasma total homocysteine concentrations in humans.
      chlorogenic acid,
      • Lee W.J.
      • Zhu B.T.
      Inhibition of DNA methylation by caffeic acid and chlorogenic acid, two common catechol-containing coffee polyphenols.
      and diterpenoids.
      • Jayasuriya H.
      • Herath K.B.
      • Ondeyka J.G.
      • et al.
      Diterpenoid, steroid, and triterpenoid agonists of liver X receptors from diversified terrestrial plants and marine sources.
      However, although decaffeinated coffee may also be inversely associated with a small reduction in all-cause mortality,
      • Loftfield E.
      • Freedman N.D.
      • Graubard B.I.
      • et al.
      Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
      • Ding M.
      • Satija A.
      • Bhupathiraju S.N.
      • et al.
      Association of coffee consumption with total and cause-specific mortality in 3 large prospective cohorts.
      • Lopez-Garcia E.
      • van Dam R.M.
      • Li T.Y.
      • Rodriguez-Artalejo F.
      • Hu F.B.
      The relationship of coffee consumption with mortality.
      even decaffeinated coffee still contains caffeine in varying amounts (typical range of 5-15 mg per 8 oz).
      • O'Keefe J.H.
      • Bhatti S.K.
      • Patil H.R.
      • DiNicolantonio J.J.
      • Lucan S.C.
      • Lavie C.J.
      Effects of habitual coffee consumption on cardiometabolic disease, cardiovascular health, and all-cause mortality.
      • Harland B.F.
      Caffeine and nutrition.
      Actually, the results of the present study revealed that the participants who drank no coffee or decaffeinated coffee consumed at least some caffeine, and caffeine intake increased with decaffeinated coffee consumption. Therefore, the results of decaffeinated coffee should not be interpreted as meaning that caffeine intake has no beneficial effects on mortality. In the present study, moderate caffeine intake was associated with a decreased risk of all-cause and noncardiovascular mortality, and an inverse association was observed in those who consumed no coffee. The mechanism of this association is currently unclear and may reflect mere chance or residual confounding. Possible explanations are that caffeine reduces the risk of depression,
      • Lucas M.
      • Mirzaei F.
      • Pan A.
      • et al.
      Coffee, caffeine, and risk of depression among women.
      stimulates the metabolic rate and has beneficial effects on weight control,
      • Acheson K.J.
      • Zahorska-Markiewicz B.
      • Pittet P.
      • Anantharaman K.
      • Jéquier E.
      Caffeine and coffee: their influence on metabolic rate and substrate utilization in normal weight and obese individuals.
      is a methylxanthine bronchodilator,
      • Gong Jr., H.
      • Simmons M.S.
      • Tashkin D.P.
      • Hui K.K.
      • Lee E.Y.
      Bronchodilator effects of caffeine in coffee: a dose-response study of asthmatic subjects.
      enhances performance in sustained high-intensity exercise,
      • Wiles J.D.
      • Bird S.R.
      • Hopkins J.
      • Riley M.
      Effect of caffeinated coffee on running speed, respiratory factors, blood lactate and perceived exertion during 1500-m treadmill running.
      and is associated with protective effects against some infectious
      • Matheson E.M.
      • Mainous III, A.G.
      • Everett C.J.
      • King D.E.
      Tea and coffee consumption and MRSA nasal carriage.
      and malignant
      • Bøhn S.K.
      • Blomhoff R.
      • Paur I.
      Coffee and cancer risk, epidemiological evidence, and molecular mechanisms.
      diseases. Conversely, the present analyses found no beneficial effects of caffeine intake on cardiovascular mortality, although several studies and a recent meta-analysis found an inverse association between coffee consumption and cardiovascular mortality.
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      Caffeine intake can stimulate sympathetic activity and increase catecholamine levels,
      • Corti R.
      • Binggeli C.
      • Sudano I.
      • et al.
      Coffee acutely increases sympathetic nerve activity and blood pressure independently of caffeine content: role of habitual versus nonhabitual drinking.
      leading to significant cardiovascular stress.
      • Svatikova A.
      • Covassin N.
      • Somers K.R.
      • et al.
      A randomized trial of cardiovascular responses to energy drink consumption in healthy adults.
      However, a previous study reported that habitual coffee consumption may not lead to cardiovascular stress, in contrast to nonhabitual coffee consumption.
      • Corti R.
      • Binggeli C.
      • Sudano I.
      • et al.
      Coffee acutely increases sympathetic nerve activity and blood pressure independently of caffeine content: role of habitual versus nonhabitual drinking.
      Hence, further studies are needed to clarify how and in what form we should consume moderate amounts of caffeine.
      The present study found that the HRs for all-cause and cause-specific mortality were lower for women than men. Similar findings were reported in recent studies about the inverse association between coffee consumption and all-cause and cause-specific mortality.
      • Freedman N.D.
      • Park Y.
      • Abnet C.C.
      • Hollenbeck A.R.
      • Sinha R.
      Association of coffee drinking with total and cause-specific mortality.
      • Lopez-Garcia E.
      • van Dam R.M.
      • Li T.Y.
      • Rodriguez-Artalejo F.
      • Hu F.B.
      The relationship of coffee consumption with mortality.
      Possible reasons may include differences in mortality, causes of death, and levels of sex hormones, such as testosterone and estrogen, which may protect against caffeine-induced cardiac stress by attenuating vascular responses to intra-arterial norepinephrine.
      • Sudhir K.
      • Elser M.D.
      • Jennings G.L.
      • Komesaroff P.A.
      Estrogen supplementation decreases norepinephrine-induced vasoconstriction and total body norepinephrine spillover in perimenopausal women.
      Although all-cause mortality in the present study was not significantly associated with the interaction term for sex and caffeine intake (P=.11), further studies are warranted to elucidate differences in the effects of caffeine intake between men and women.
      This study had several limitations that should be addressed. First, this was an observational study, and thus it is not possible to conclude that the inverse relationship between moderate caffeine intake and mortality reflects cause and effect. Second, caffeine intake was self-reported and may not accurately reflect long-term patterns of caffeine intake. Notwithstanding, this is the first report of beneficial effects on all-cause and noncardiovascular mortality in response to moderate caffeine intake. The results of this study indicate that moderate caffeine intake conveys potential health benefits and may have a substantial impact on the management of various diseases.

      Conclusion

      In this study, moderate caffeine intake was associated with a decreased risk of all-cause mortality, particularly noncardiovascular mortality, regardless of the presence or absence of coffee consumption.

      Supplemental Online Material

      Supplemental Online Material

      Supplemental material can be found online at http://www.mayoclinicproceedings.org. Supplemental material attached to journal articles has not been edited, and the authors take responsibility for the accuracy of all data.

      References

        • Freedman N.D.
        • Park Y.
        • Abnet C.C.
        • Hollenbeck A.R.
        • Sinha R.
        Association of coffee drinking with total and cause-specific mortality.
        N Engl J Med. 2012; 366 ([published correction appears in N Engl J Med. 2012;367(3):285]): 1891-1904
        • Loftfield E.
        • Freedman N.D.
        • Graubard B.I.
        • et al.
        Association of coffee consumption with overall and cause-specific mortality in a large US prospective cohort Study.
        Am J Epidemiol. 2015; 182: 1010-1022
        • Ding M.
        • Satija A.
        • Bhupathiraju S.N.
        • et al.
        Association of coffee consumption with total and cause-specific mortality in 3 large prospective cohorts.
        Circulation. 2015; 132: 2305-2315
        • Saito E.
        • Inoue M.
        • Sawada N.
        • et al.
        Association of coffee intake with total and cause-specific mortality in a Japanese population: the Japan Public Health Center-based Prospective Study.
        Am J Clin Nutr. 2015; 101: 1029-1037
        • Guercio B.J.
        • Sato K.
        • Niedzwiecki D.
        • et al.
        Coffee intake, recurrence, and mortality in stage III colon cancer: results from CALGB 89803 (Alliance).
        J Clin Oncol. 2015; 33: 3598-3607
        • Frary C.D.
        • Johnson R.K.
        • Wang M.Q.
        Food sources and intakes of caffeine in the diets of persons in the United States.
        J Am Diet Assoc. 2005; 105 ([published correction appears in J Am Diet Assoc. 2008;108(4):727]): 110-113
        • Corti R.
        • Binggeli C.
        • Sudano I.
        • et al.
        Coffee acutely increases sympathetic nerve activity and blood pressure independently of caffeine content: role of habitual versus nonhabitual drinking.
        Circulation. 2002; 106: 2935-2940
        • Svatikova A.
        • Covassin N.
        • Somers K.R.
        • et al.
        A randomized trial of cardiovascular responses to energy drink consumption in healthy adults.
        JAMA. 2015; 314: 2079-2082
      1. Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey. http://www.cdc.gov/nchs/nhanes.htm. Updated January 25, 2017. Accessed June 30, 2015.

      2. Centers for Disease Control and Prevention. NCHS Research Ethics Review Board (ERB) Approval. http://www.cdc.gov/nchs/nhanes/irba98.htm. Updated November 5, 2012. Accessed June 30, 2015.

      3. Centers for Disease Control and Prevention. Public-use Linked Mortality File. http://www.cdc.gov/nchs/data/datalinkage/Public_use_linked_mortality_file_readme_text_1_2015.pdf. Updated February 2015. Accessed June 30, 2015.

      4. Centers for Disease Control and Prevention. National Health and Nutrition Examination Survey: measuring guides for the dietary recall interview. https://www.cdc.gov/nchs/nhanes/measuring_guides_dri/measuringguides.htm. Updated May 20, 2010. Accessed February 14, 2017.

        • Raper N.
        • Perloff B.
        • Ingwersen L.
        • Steinfeldt L.
        • Anand J.
        An overview of USDA's Dietary Intake Data System.
        J Food Comp Anal. 2004; 17: 545-555
      5. Agricultural Research Service, United States Department of Agriculture. AMPM—USDA Automated Multiple-Pass Method. http://www.ars.usda.gov/Services/docs.htm?docid=7710. Updated September 8, 2016. Accessed October 21, 2014.

        • Drewnowski A.
        • Rehm C.D.
        Sources of caffeine in diets of US children and adults: trends by beverage type and purchase location.
        Nutrients. 2016; 8: 154
        • Albright A.L.
        • Gregg E.W.
        Preventing type 2 diabetes in communities across the U.S.: the National Diabetes Prevention Program.
        Am J Prev Med. 2013; 44: S346-S351
      6. Centers for Disease Control and Prevention. NCHS surveys: 2011 linked mortality files; public-use data dictionary. http://www.cdc.gov/nchs/data/datalinkage/Public-use_Data_Dictionary.pdf. Updated April 16, 2015. Accessed June 30, 2015.

        • He Y.
        Missing data analysis using multiple imputation: getting to the heart of the matter.
        Circ Cardiovasc Qual Outcomes. 2010; 3: 98-105
        • Donnelly J.E.
        • Blair S.N.
        • Jakicic J.M.
        • Manore M.M.
        • Rankin J.W.
        • Smith B.K.
        • American College of Sports Medicine
        American College of Sports Medicine Position Stand: appropriate physical activity intervention strategies for weight loss and prevention of weight regain for adults.
        Med Sci Sports Exerc. 2009; 41 ([published correction appears in Med Sci Sports Exerc. 2009;41(7):1532]): 459-471
        • Hankinson A.L.
        • Daviglus M.L.
        • Bouchard C.
        • et al.
        Maintaining a high physical activity level over 20 years and weight gain.
        JAMA. 2010; 304 ([published correction appears in JAMA. 2011;305(2):150]): 2603-2610
        • O'Keefe J.H.
        • Bhatti S.K.
        • Patil H.R.
        • DiNicolantonio J.J.
        • Lucan S.C.
        • Lavie C.J.
        Effects of habitual coffee consumption on cardiometabolic disease, cardiovascular health, and all-cause mortality.
        J Am Coll Cardiol. 2013; 62: 1043-1051
        • Olthof M.R.
        • Hollman P.C.
        • Zock P.L.
        • Katan M.B.
        Consumption of high doses of chlorogenic acid, present in coffee, or of black tea increases plasma total homocysteine concentrations in humans.
        Am J Clin Nutr. 2001; 73: 532-538
        • Lee W.J.
        • Zhu B.T.
        Inhibition of DNA methylation by caffeic acid and chlorogenic acid, two common catechol-containing coffee polyphenols.
        Carcinogenesis. 2006; 27: 269-277
        • Jayasuriya H.
        • Herath K.B.
        • Ondeyka J.G.
        • et al.
        Diterpenoid, steroid, and triterpenoid agonists of liver X receptors from diversified terrestrial plants and marine sources.
        J Nat Prod. 2005; 68: 1247-1252
        • Lopez-Garcia E.
        • van Dam R.M.
        • Li T.Y.
        • Rodriguez-Artalejo F.
        • Hu F.B.
        The relationship of coffee consumption with mortality.
        Ann Intern Med. 2008; 148: 904-914
        • Harland B.F.
        Caffeine and nutrition.
        Nutrition. 2000; 16: 522-526
        • Lucas M.
        • Mirzaei F.
        • Pan A.
        • et al.
        Coffee, caffeine, and risk of depression among women.
        Arch Intern Med. 2011; 171: 1571-1578
        • Acheson K.J.
        • Zahorska-Markiewicz B.
        • Pittet P.
        • Anantharaman K.
        • Jéquier E.
        Caffeine and coffee: their influence on metabolic rate and substrate utilization in normal weight and obese individuals.
        Am J Clin Nutr. 1980; 33: 989-997
        • Gong Jr., H.
        • Simmons M.S.
        • Tashkin D.P.
        • Hui K.K.
        • Lee E.Y.
        Bronchodilator effects of caffeine in coffee: a dose-response study of asthmatic subjects.
        Chest. 1986; 89: 335-342
        • Wiles J.D.
        • Bird S.R.
        • Hopkins J.
        • Riley M.
        Effect of caffeinated coffee on running speed, respiratory factors, blood lactate and perceived exertion during 1500-m treadmill running.
        Br J Sports Med. 1992; 26: 116-120
        • Matheson E.M.
        • Mainous III, A.G.
        • Everett C.J.
        • King D.E.
        Tea and coffee consumption and MRSA nasal carriage.
        Ann Fam Med. 2011; 9: 299-304
        • Bøhn S.K.
        • Blomhoff R.
        • Paur I.
        Coffee and cancer risk, epidemiological evidence, and molecular mechanisms.
        Mol Nutr Food Res. 2014; 58: 915-930
        • Sudhir K.
        • Elser M.D.
        • Jennings G.L.
        • Komesaroff P.A.
        Estrogen supplementation decreases norepinephrine-induced vasoconstriction and total body norepinephrine spillover in perimenopausal women.
        Hypertension. 1997; 30: 1538-1543