Association of Coffee Consumption With Overall and Cause-Specific Mortality in a Large US Prospective Cohort Study

Abstract

Concerns about high caffeine intake and coffee as a vehicle for added fat and sugar have raised questions about the net impact of coffee on health. Although inverse associations have been observed for overall mortality, data for cause-specific mortality are sparse. Additionally, few studies have considered exclusively decaffeinated coffee intake or use of coffee additives. Coffee intake was assessed at baseline by self-report in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial. Hazard ratios were estimated using Cox proportional hazards models. Among 90,317 US adults without cancer at study baseline (1998-2001) or history of cardiovascular disease at study enrollment (1993-2001), 8,718 deaths occurred during 805,644 person-years of follow-up from 1998 through 2009. Following adjustment for smoking and other potential confounders, coffee drinkers, as compared with nondrinkers, had lower hazard ratios for overall mortality (<1 cup/day: hazard ratio (HR) = 0.99 (95% confidence interval (CI): 0.92, 1.07); 1 cup/day: HR = 0.94 (95% CI: 0.87, 1.02); 2-3 cups/day: HR = 0.82 (95% CI: 0.77, 0.88); 4-5 cups/day: HR = 0.79 (95% CI: 0.72, 0.86); ≥6 cups/day: HR = 0.84 (95% CI: 0.75, 0.95)). Similar findings were observed for decaffeinated coffee and coffee additives. Inverse associations were observed for deaths from heart disease, chronic respiratory diseases, diabetes, pneumonia and influenza, and intentional self-harm, but not cancer. Coffee may reduce mortality risk by favorably affecting inflammation, lung function, insulin sensitivity, and depression.

Keywords: additives; caffeine; cause-specific mortality; coffee; mortality.

Figure 1.

Associations of daily coffee intake (total, caffeinated only, and decaffeinated only) with overall and cause-specific mortality among participants who drank ≥4 cups/day as compared with non–coffee drinkers in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, 1998–2009. Hazard ratios (HRs) were adjusted for age, sex, detailed smoking history, race/ethnicity, educational level, marital status, employment status, presence or absence of diabetes, body mass index (weight (kg)/height (m)²), any supplemental vitamin use in the previous 12 months, regular ibuprofen use in the previous 12 months, regular aspirin use in the previous 12 months, receipt of menopausal hormone therapy (women only), alcohol consumption, total daily energy intake, and quintile of intake of daily red and processed meat, white meat (i.e., poultry and fish), saturated fat, fruits, and vegetables. HRs for cancer mortality were additionally adjusted for history of cancer (other than nonmelanoma skin cancer) in a first-degree relative. HRs for diabetes mortality were not adjusted for the presence or absence of self-reported diabetes. Because there were no deaths from kidney disease in the highest category (≥4-cups/day) of decaffeinated coffee consumption, the association for 2–3 cups/day (as compared with non–coffee drinking) is shown. Statistical tests were 2-sided, and P < 0.05 was interpreted as statistically significant. Bars, CI, confidence intervals (CIs).

Figure 2.

Associations of daily coffee intake with overall mortality among subgroups of participants (determined by important baseline factors) who drank ≥4 cups/day as compared with non–coffee drinkers in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, 1998–2009. Where applicable, hazard ratios (HRs) were adjusted for age, sex, detailed smoking history, race/ethnicity, educational level, marital status, employment status, presence or absence of diabetes, body mass index (weight (kg)/height (m)²), any supplemental vitamin use in the previous 12 months, regular ibuprofen use in the previous 12 months, regular aspirin use in the previous 12 months, receipt of menopausal hormone therapy (MHT; women only), alcohol consumption, total daily energy intake, and quintile of daily intake of red and processed meat, white meat (i.e., poultry and fish), saturated fat, fruits, and vegetables. Risk estimates for cancer mortality were additionally adjusted for history of cancer (other than nonmelanoma skin cancer) in a first-degree relative. Statistical tests were 2-sided, and P < 0.05 was interpreted as statistically significant. Nonsteroidal antiinflammatory drug (NSAID) use was defined as use of any product containing aspirin or ibuprofen. Bars, CI, confidence intervals (CIs).