Abstract
Background
Although coffee consumption has been associated with a variety of health benefits, it remains unclear whether the addition of sugar and saturated fat modifies these effects.
Objectives
This study aimed to examine the relationship between coffee intake, including the amount of added sugar and saturated fat, and mortality among United States adults.
Methods
We conducted a prospective cohort study using data from the National Health and Nutrition Examination Survey from 1999 to 2018, linked to National Death Index Mortality Data. The study included 46,222 adults aged 20 y and older who completed valid first-day 24-h dietary recalls. Coffee consumption was categorized by type (caffeinated or decaffeinated) and by sugar (<2.5 g per 8-oz), and saturated fat content (<1 g per 8-oz). Mortality outcomes included all-cause, cancer, and cardiovascular disease (CVD) mortality. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for mortality risk based on type, intake levels (nondrinkers, <1 cup/d, 1 to <2 cups/d, 2 to <3 cups/d, and ≥3 cups/d) and the amount of added sugar and saturated fat.
Results
Over a median follow-up of 9.3–11.3 y, a total of 7074 deaths occurred, including 1176 cancer-specific and 1089 CVD-specific deaths. Higher coffee consumption was associated with lower all-cause mortality [HR (95% CI): 0.89 (0.78, 1.02) for < 1 cup/d; 0.84 (0.77, 0.92) for 1 to <2 cups/d; 0.83 (0.75, 0.93) for 2 to <3 cups/d; and 0.85 (0.77, 0.95) for ≥3 cups/d; P-trend = 0.004]. However, the mortality benefits were restricted to black coffee [HR (95% CI): 0.86 (0.77, 0.97)] and coffee with low added sugar and saturated fat content [HR (95% CI): 0.86 (0.75, 0.99)].
Conclusions
The health benefits of coffee consumption may be diminished when sugar and saturated fat are added.
Keywords: coffee consumption, mortality, prospective cohort study, cardiovascular diseases, added sugar, saturated fat, NHANES, all-cause mortality
Introduction
Coffee is one of the world’s most commonly consumed beverages, with about half of the adults in the United States reported drinking coffee daily [1]. Meta-analyses of prospective cohort studies have reported an inverse association between coffee consumption and the risk of mortality, type 2 diabetes, cardiovascular disease (CVD), and certain cancers [[1], [2], [3], [4]]. The health benefits of coffee are potentially attributable to its bioactive compounds, such as caffeine, chlorogenic acid, and polyphenol, which have been associated with antioxidant, anti-inflammatory, and anticarcinogenic properties [5,6]. However, prior research has also revealed that flavored coffee and tea are the second-largest contributors to the highest deciles of added sugar consumption among United States adults aged 19 y and older, after sugar-sweetened sodas, and fruit drinks [7]. In addition, coffee drinkers consumed a higher amount of saturated fat than nondrinkers [8]. The addition of sugar and saturated fat to coffee can contribute to a high level of calories consumed from coffee, potentially counterbalancing coffee’s health benefits. Nevertheless, the extent to which adding sugar and saturated fat to coffee modifies its health benefits has not been extensively evaluated. Using a nationally representative sample of United States adults, our study aimed to assess whether the associations between coffee consumption and mortality risk differ by the amount of sugar and saturated fat added to coffee.
Methods
Study design and population
The study population included United States adults aged 20 y or older who participated in 9 cycles of the NHANES (1999–2000 to 2017–2018, the latest cycle linked to National Death Index data). NHANES is conducted biannually by the National Center for Health Statistics and is designed to assess the health and nutritional status of adults and children in the United States NHANES was approved by the research ethics review board at National Center for Health Statistics; all participants provided written informed consent. The Institutional Review Board of the Centers for Disease Prevention and Control approved the survey protocol [32]. For the current analysis, we excluded participants who were <20 y old (n = 46,235), had no valid dietary recall on the first day (n = 6345), and were pregnant or lactating women (n = 1551). We also excluded participants who had unreliable reporting of dietary intake, defined as log-transformed total energy intake outside 3 SDs of the mean (<472 or >6726 kcal/d; n = 333), participants who consumed excessively high levels of coffee, defined as log-transformed total coffee intake above 3 SDs of the mean (>20.1 cups/d; n = 93), and participants who had no linked mortality data (n = 79). In addition, we excluded participants who died within 12 months of dietary assessment (n = 458) to minimize the chance of reverse causation. These exclusions resulted in a total of 46,222 participants included in the current analysis (Figure 1).
FIGURE 1.
Study flowchart.
Assessment of coffee consumption
Coffee consumption was assessed using dietary intake data collected from 1 or 2 d of 24-h diet recalls by trained interviewers. The first diet recall was conducted in person at a Mobile Examination Center for participants who participated in the 1999–2002 cycles; from 2003 onward, a second recall was added, which was conducted over the phone ∼3–10 d after the first recall [33]. We included participants who provided valid dietary recall on the first day in this analysis. To ensure the representativeness of the United States population and appropriately scale each participant’s contribution within their respective group, first-day dietary data were weighted using the dietary day-one sample weight.
To estimate coffee consumption, we identified all food codes that indicate coffee-containing beverages (food codes from 92100000–92200000) in 2 categories: noncombination food and beverages with additions. For coffee consumed as noncombination foods, we first determined whether it was caffeinated compared with decaffeinated and black compared with nonblack based on the descriptions of each food code. For nonblack coffee, we further categorized it into one of the 4 types based on the grams of saturated fat and added sugars consumed per cup of coffee, adapting the recommended values proposed by the United States Food and Drug Administration for the use of the term “healthy” on food labeling: 1) coffee with low added sugar [contained < 5% daily value (DV) (2.5 g) per cup] and low saturated fat [<5% DV (1 g) per cup]; 2) coffee with low added sugar [contained <5% DV (2.5 g) per cup] and high saturated fat [≥5% DV (1 g) per cup]; 3) coffee with high added sugar [contained <5% DV (2.5 g) per cup] and low saturated fat [<5% DV (1 g) per cup]; and 4) coffee with high added sugar [contained ≥5% DV (2.5 g) per cup] and high saturated fat [≥5% DV (1 g) per cup]. For coffee consumed with additions (e.g., milk, cream, half and half, granulated sugar, honey, and syrup), the amount of saturated fat and added sugar consumed from these additions was added to that consumed from coffee to produce the total amount of saturated fat and added sugar. If multiple coffees were consumed by an individual in a single day, the consumption from the same type was combined to estimate the total consumption of each type. For participants who reported coffee consumption of the same type on both days, the daily consumption was averaged from 2 recalls. The consumption in grams per day was converted to fluid ounce (fl. oz.) equivalent per day using 1 fl. oz. = 30 g and further converted to cups per day using 1 cup = 8 fl. oz. (i.e., 1 cup = 240 g). The consumption was adjusted for total energy intake using the nutrient density method. The excluded participants were included in the analysis as a subsample with the use of DOMAIN statement to appropriately account for the survey weights in the NHANES analysis.
Assessment of mortality
All-cause mortality and cancer mortality outcomes were ascertained using the linkage to NHANES to death certificate records from the National Death Index, currently available from the CDC through 31 December 2019 [34]. The primary outcome was all-cause mortality, and the secondary outcomes were mortality due to CVD and cancer. On the basis of the International Classification of Diseases, 10th Revision, mortality due to cancer was defined as codes C00–C97 being the underlying cause of death. We defined mortality due to CVD mortality by International Classification of Diseases, 10th Revision codes for heart diseases (I00–I09, I11, I13, and I20–I51) and cerebrovascular diseases (I60–I69). However, CVD mortality data for linking the NHANES cycle 2015–2018 had not been published at the time of this analysis. Therefore, we only used the NHANES cycle 1999–2014 to assess CVD mortality. We excluded participants (n = 3754) with CVD (including congestive heart failure, coronary artery disease, myocardial infarction, or stroke) or cancer (n = 3614) at baseline when assessing the association of coffee consumption with CVD or cancer mortality. Other cause-specific mortality was not assessed due to a smaller number of deaths due to other specific causes as well as less plausible biological mechanisms for coffee and such causes. Follow-up length was defined as the interval from the interview date of the first 24-h diet recall to the date of death or 31 December 2019, in cases where the individual was censored.
Assessment of demographic, lifestyle, and comorbidity conditions
Demographic and lifestyle factors such as age, gender, race/ethnicity, education, family income, smoking status, and physical activity were collected during structured household interviews. Weight, height, and alcohol use were obtained during physical examinations and interviews at the Mobile Examination Center. Race/ethnicity was categorized as non-Hispanic White, non-Hispanic Black, Hispanic, and other. Family income level was based on the ratio of family income to the poverty threshold, adjusted for household size. Ever smokers were defined as individuals who had smoked 100+ cigarettes in their lifetime, and pack-years were calculated by multiplying the packs (1 pack = 20 cigarettes) of cigarettes smoked per day and years as a smoker. Drinkers were defined as participants who consumed 12+ alcoholic drinks each year. Moderate compared with heavy drinkers were defined as those who consumed <1 compared with 1+ drink per day (women) or <2 compared with 2+ drinks per day (men). Metabolic equivalent (MET)-hours per week for each moderate-to-vigorous leisure-time physical activity was calculated by multiplying the number of hours spent in the activity and its suggested MET scores. Physical activity was obtained by summing up the MET-hours per week of total leisure-time physical activities. BMI was calculated as weight (kilograms) divided by height (meters) squared. Healthy Eating Index-2015 (HEI-2020), which assesses adherence to the 2015 Dietary Guidelines for Americans, was calculated as a measure of overall diet quality, with a higher score corresponding to a better diet quality [35]. Tea consumption (g/d) was estimated using the dietary data collected using the 24-h diet recalls. Comorbid conditions (cancer, congestive heart failure, coronary artery disease, myocardial infarction, stroke, high cholesterol, hypertension, and diabetes) were defined if participants reported that they have ever been told by a healthcare professional that they had such conditions and/or to take prescribed medications because of these conditions, or they were currently taking medication for such a condition.
Statistical analysis
We assessed coffee consumption based on type, including caffeinated compared with decaffeinated coffee and black compared with nonblack coffee, with nonblack coffee being further categorized into 4 types: low added sugar and low saturated fat (LSLF); low added sugar and high saturated fat (LSHF); high added sugar and low saturated fat (HSLF); and high added sugar and high saturated fat (HSHF). For total, caffeinated, and decaffeinated coffee consumption, we further assessed the intake levels in 4 groups (<1 cup/d, 1 to <2 cups/d, 2 to <3 cups/d, and ≥3 cups/d). Those who reported no coffee consumption on both days were classified as nondrinkers.
We first compared the demographics, lifestyle factors, and comorbid conditions by coffee consumption using analysis of variance for continuous variables and chi-square tests for categorical variables. Cox proportional hazard models were used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) for coffee consumption by dose and type with mortality, with person-years as the underlying time metrics. Separate analyses were conducted for total coffee, caffeinated coffee, and decaffeinated coffee consumption. The analysis on caffeinated or decaffeinated coffee was restricted to those who reported consuming caffeinated or decaffeinated coffee on both days, respectively. Similarly, the analysis of coffee type by the amount of added sugar and saturated fat included participants who consumed the same type on both days. Those who consumed mixed coffee types on 2 d (e.g., individuals who reported drinking black coffee on day 1 and HSHF coffee on day 2), including 6515 who consumed mixed types of total coffee, 5304 who consumed mixed types of caffeinated coffee, and 487 who consumed mixed types of decaffeinated coffee, were excluded from the corresponding analysis.
The proportional hazard assumption was evaluated by adding interaction terms between follow-up time and coffee consumption; the lack of statistical significance of the interaction terms suggested the proportional hazard assumption was not violated. All analyses were adjusted for age, sex, race/ethnicity, education, family income, physical activity, BMI, comorbid conditions at baseline, cigarette smoking, alcohol intake, tea consumption, HEI2020, and total energy intake. The analysis of coffee type was additionally adjusted for the total number of cups of coffee consumed per day. Testing for linear trends was conducted by entering the median value of each category of coffee consumption as continuous variable in the model. We also applied restricted cubic splines with 3 knots to model the dose–response association between coffee consumption and risk of mortality. In addition, we examined whether the associations between coffee consumption and total mortality differed by age, sex, BMI, smoking status, and comorbidity status. We also conducted a sensitivity analysis excluding participants from the 2017–2018 cycle due to the short time interval between dietary data collection and mortality outcomes.
Sampling weights, stratification, and clustering were incorporated into all analyses to account for the complex sample design of NHANES and have findings that represent the national population. Analysis was conducted using SAS, version 9.4 (SAS Institute Inc). A P value <0.05 was considered statistically significant.
Results
More than half (53.4%) of the United States adults aged 20+ y in 1999–2018 reported drinking coffee daily. Among those who drank coffee, the mean (SE) consumption of total coffee was 2.84 (0.03) cups/d, caffeinated coffee was 2.57 (0.03) cups/d, and decaffeinated coffee was 0.27 (0.01) cups/d. For each 8-fluid oz. cup of coffee consumed by United States adults, the mean (SE) consumption of added sugar was 3.24 (0.08) g, and of saturated fat was 0.52 (0.01) g.
Individuals who drank more coffee were older in age, more likely to be female, non-Hispanic Whites, have higher education and income attainment, be nonsmokers or nondrinkers, and have lower levels of total energy intake but a higher BMI and lower levels of physical activity, and were less likely to drink tea. They were also more likely to have comorbid conditions such as cancer and cardiometabolic conditions (Table 1). Individuals who drank black coffee, compared with nondrinkers or those who drank other types of coffee, were older and more likely to be male, non-Hispanic Whites, smokers, heavy drinks, have lower levels of physical activity, and have cancer and cardiometabolic conditions except for diabetes. Individuals who drank LSLF coffee had higher education and income attainment, better diet quality, lower total energy intake, and diabetes. Individuals who drank HSHF coffee were less likely to be non-Hispanic White, had worse diet quality and high total energy intake, and were less likely to have cancer and cardiometabolic conditions. Individuals who drank LSHF coffee were more likely to be women. Individuals who drank HSLF coffee had lower levels of education but higher levels of physical activity and a lower BMI (Table 2).
TABLE 1.
Characteristics of United States adults 20+ y by amount of total coffee consumption, NHANES 1999–20181.
| Characteristics | Total | No coffee | Any coffee | P value2 | Total coffee consumption, 8-oz. cups/d |
P value2 | |||
|---|---|---|---|---|---|---|---|---|---|
| participants |
consumption |
consumption |
<1 cup |
1–2 cups |
2–3 cups |
≥3 cups |
|||
| n = 46,222 | N = 21,889 | N = 24,333 | n = 3691 | n = 8405 | n = 5146 | n = 7091 | |||
| Age, mean ± SE (y) | 47.3 ± 0.2 | 42.0 ± 0.2 | 52.0 ± 0.2 | <0.001 | 48.2 ± 0.4 | 51.2 ± 0.3 | 52.8 ± 0.3 | 54.0 ± 0.3 | <0.001 |
| Female, n (%) | 23,211 (51.0) | 11,000 (50.6) | 12,211 (51.2) | 0.08 | 1508 (41.4) | 4147 (50.1) | 2696 (52.4) | 3860 (55.6) | <0.001 |
| Race/ethnicity, n (%) | <0.001 | <0.001 | |||||||
| Non-Hispanic White | 20,724 (68.6) | 8322 (61.8) | 12,402 (74.6) | 1385 (61.5) | 3805 (69.8) | 2704 (76.7) | 4508 (83.5) | ||
| Non-Hispanic Black | 9731 (11.2) | 6430 (16.6) | 3301 (6.4) | 617 (9.4) | 1334 (7.8) | 671 (5.9) | 679 (4.1) | ||
| Hispanic | 11,818 (13.5) | 5045 (13.9) | 6773 (13.1) | 1331 (21.5) | 2545 (15.5) | 1401 (11.9) | 1496 (8.0) | ||
| Other | 3949 (6.7) | 2092 (7.7) | 1857 (5.9) | 358 (7.6) | 721 (7.0) | 370 (5.5) | 408 (4.4) | ||
| Education, n (%) | <0.001 | <0.001 | |||||||
| Grades 0–12 | 23,070 (41.1) | 10,768 (41.7) | 12,302 (40.6) | 1852 (42.3) | 4255 (40.2) | 2566 (39.1) | 3629 (41.4) | ||
| Some college | 13,133 (31.3) | 6590 (33.1) | 6543 (29.7) | 956 (28.4) | 2185 (28.9) | 1401 (29.2) | 2001 (31.3) | ||
| College graduate and above | 9965 (27.6) | 4502 (25.2) | 5463 (29.7) | 879 (29.3) | 1956 (30.9) | 1175 (31.7) | 1453 (27.4) | ||
| Family-income-to-poverty ratio3, n (%) | <0.001 | <0.001 | |||||||
| <1.30 | 12,755 (21.5) | 6709 (25.6) | 6046 (18.0) | 971 (21.5) | 2069 (18.5) | 1205 (15.4) | 1801 (17.7) | ||
| 1.30–2.99 | 13,575 (28.5) | 6405 (29.5) | 7170 (27.7) | 1089 (26.9) | 2508 (28.2) | 1511 (26.8) | 2062 (28.1) | ||
| 3.00–4.90 | 8408 (24.2) | 3840 (23.4) | 4568 (24.8) | 676 (23.2) | 1541 (24.1) | 1007 (26.2) | 1344 (25.4) | ||
| ≥5.00 | 7583 (25.8) | 3090 (21.5) | 4493 (29.5) | 636 (28.3) | 1538 (29.3) | 987 (31.6) | 1332 (28.8) | ||
| Cigarette smoking, n (%) | <0.001 | <0.001 | |||||||
| Never smokers | 24,902 (53.4) | 13,584 (62.5) | 11,318 (45.6) | 1958 (53.6) | 4297 (50.7) | 2424 (46.9) | 2639 (36.2) | ||
| Ever smokers | 21,285 (46.6) | 8286 (37.5) | 12,999 (54.4) | 1731 (46.4) | 4105 (49.3) | 2718 (53.1) | 4445 (63.8) | ||
| Pack-years for smokers, mean ± SE (y) | 21.3 ± 0.3 | 15.5 ± 0.4 | 24.7 ± 0.5 | <0.001 | 20.3 ± 1.2 | 21.0 ± 0.7 | 24.7 ± 0.8 | 28.7 ± 0.7 | <0.001 |
| Alcohol intake, n (%) | <0.001 | <0.001 | |||||||
| Nondrinkers | 16,702 (32.1) | 8122 (34.9) | 8580 (29.7) | 1166 (28.3) | 2859 (28.2) | 1857 (29.4) | 2698 (31.9) | ||
| Moderate drinkers | 22,591 (57.2) | 10,352 (55.4) | 12,239 (58.6) | 1958 (61.2) | 4286 (59.4) | 2538 (59.3) | 3457 (56.5) | ||
| Heavy drinkers | 3816 (10.7) | 1731 (9.6) | 2085 (11.7) | 312 (10.5) | 738 (12.4) | 434 (11.4) | 601 (11.6) | ||
| Tea consumption, mean ± SE (g/d) | 197.0 ± 5.1 | 230.8 ± 7.3 | 167.5 ± 5.4 | <0.001 | 190.5 ± 13.2 | 171.9 ± 8.5 | 164.3 ± 9.6 | 155.8 ± 8.5 | <0.001 |
| Healthy Eating Index-2020 score, mean ± SE | 52.5 ± 0.2 | 51.2 ± 0.2 | 53.7 ± 0.2 | <0.001 | 53.2 ± 0.4 | 54.0 ± 0.3 | 54.1 ± 0.3 | 53.2 ± 0.3 | <0.001 |
| Physical activity, mean ± SE, MET-h/wk | 17.0 ± 0.4 | 18.9 ± 0.5 | 15.3 ± 0.4 | <0.001 | 16.6 ± 0.8 | 16.2 ± 0.8 | 14.8 ± 0.6 | 14.2 ± 0.6 | <0.001 |
| BMI, mean ±SE, kg/m2 | 28.8 ± 0.1 | 29.1 ± 0.1 | 28.5 ± 0.1 | <0.001 | 28.4 ± 0.2 | 28.4 ± 0.1 | 28.6 ± 0.1 | 28.5 ± 0.1 | <0.001 |
| Weight status, n (%) | <0.001 | <0.001 | |||||||
| BMI <25 kg/m2 | 13,362 (31.1) | 6518 (31.9) | 6844 (30.4) | 1079 (31.0) | 2452 (32.4) | 1353 (28.9) | 1960 (29.1) | ||
| BMI 25–29.9 kg/m2 | 15,432 (33.3) | 6666 (29.9) | 8766 (36.3) | 1375 (36.3) | 2966 (35.1) | 1882 (37.1) | 2543 (36.8) | ||
| BMI ≥30 kg/m2 | 16,754 (35.6) | 8390 (38.2) | 8364 (33.3) | 1186 (32.6) | 2867 (32.5) | 1830 (33.9) | 2481 (34.0) | ||
| Total energy intake, mean ± SE, kcal/d | 2164.9 ± 6.8 | 2207.7 ± 9.4 | 2127.7 ± 9.1 | <0.001 | 2826.5 ± 27.4 | 2280.1 ± 13.5 | 2047.5 ± 18.7 | 1739.6 ± 12.9 | <0.001 |
| Comorbid conditions, n (%) | |||||||||
| Cancer | 3621 (7.8) | 1241 (5.8) | 2380 (9.5) | <0.001 | 255 (6.8) | 788 (9.2) | 543 (9.7) | 794 (10.7) | <0.001 |
| Congestive heart failure | 1299 (2.1) | 515 (1.7) | 784 (2.4) | <0.001 | 74 (1.6) | 260 (2.1) | 179 (2.3) | 271 (3.0) | <0.001 |
| Coronary artery disease | 1741 (3.1) | 583 (2.2) | 1158 (3.9) | <0.001 | 120 (3.2) | 364 (3.3) | 260 (3.7) | 414 (4.9) | <0.001 |
| Myocardial infarction | 1798 (3.1) | 608 (2.1) | 1190 (3.9) | <0.001 | 121 (3.0) | 347 (3.2) | 250 (3.4) | 472 (5.3) | <0.001 |
| Stroke | 1559 (2.5) | 600 (2.1) | 959 (2.9) | <0.001 | 106 (2.2) | 297 (2.5) | 204 (2.8) | 352 (3.6) | <0.001 |
| High cholesterol | 13,611 (31.4) | 4466 (24.7) | 9145 (36.0) | <0.001 | 1209 (32.5) | 3104 (35.2) | 1963 (37.2) | 2842 (39.6) | <0.001 |
| Hypertension | 20,003 (43.2) | 7571 (40.0) | 12,432 (45.5) | <0.001 | 1623 (43.4) | 4197 (45.3) | 2695 (46.7) | 3821 (49.1) | <0.001 |
| Diabetes | 5134 (8.9) | 1767 (7.6) | 3367 (9.8) | <0.001 | 390 (7.7) | 1119 (9.4) | 757 (9.8) | 1088 (11.1) | <0.001 |
Abbreviation: MET, metabolic equivalent.
Mean ± SE and percentage were adjusted for NHANES survey weights to generate nationally representative estimates.
P values were generated from analysis of variances for continuous variables and chi-square tests for categorical variables.
The family-income-to-poverty ratio is calculated as the ratio of a family’s total income to the poverty threshold established by the United States Department of Health and Human Services for that specific year, specific to family size, as well as the appropriate year and state.
TABLE 2.
Characteristics of United States adults 20+ y by type of total coffee consumption, NHANES 1999–20181.
| No coffee consumption |
Type of coffee consumption2 |
P value3 | |||||
|---|---|---|---|---|---|---|---|
| Black |
LSLF |
LSHF |
HSLF |
HSHF |
|||
| n = 21,889 | n = 5561 | n = 6607 | n = 1346 | n = 6830 | n = 2430 | ||
| Age, mean ± SE (y) | 42.0 ± 0.2 | 56.5 ± 0.3 | 53.7 ± 0.3 | 50.8 ± 0.6 | 48.2 ± 0.3 | 44.2 ± 0.4 | <0.001 |
| Female, n (%) | 11,000 (50.6) | 2362 (43.3) | 3597 (56.6) | 772 (61.5) | 3262 (48.1) | 1400 (58.2) | <0.001 |
| Race/ethnicity, n (%) | <0.001 | ||||||
| Non-Hispanic White | 8322 (61.8) | 4146 (88.1) | 3738 (80.7) | 564 (66.3) | 2448 (60.1) | 767 (57.1) | |
| Non-Hispanic Black | 6430 (16.6) | 313 (2.2) | 766 (4.7) | 165 (6.5) | 1441 (11.9) | 500 (12.3) | |
| Hispanic | 5045 (13.9) | 790 (5.6) | 1660 (9.9) | 470 (16.8) | 2383 (20.6) | 896 (21.9) | |
| Other | 2092 (7.7) | 312 (4.1) | 443 (4.8) | 147 (10.4) | 558 (7.4) | 267 (8.6) | |
| Education, n (%) | <0.001 | ||||||
| Grades 0–12 | 10,768 (41.7) | 2608 (38.4) | 3128 (36.9) | 667 (36.8) | 3861 (48.3) | 1322 (47.8) | |
| Some college | 6590 (33.1) | 1551 (30.2) | 1764 (28.2) | 325 (27.7) | 1833 (31.5) | 662 (29.4) | |
| College graduate and above | 4502 (25.2) | 1398 (31.4) | 1706 (34.9) | 353 (35.5) | 1128 (20.2) | 444 (22.8) | |
| Family-income-to-poverty ratio, n (%) | <0.001 | ||||||
| <1.30 | 6709 (25.6) | 1107 (13.9) | 1403 (14.0) | 327 (17.5) | 2088 (25.2) | 756 (26.6) | |
| 1.30–2.99 | 6405 (29.5) | 1682 (27.1) | 1897 (27.4) | 368 (26.7) | 2067 (29.8) | 728 (29.0) | |
| 3.00–4.90 | 3840 (23.4) | 1196 (28.2) | 1290 (23.9) | 246 (24.9) | 1123 (22.1) | 409 (23.4) | |
| ≥5.00 | 3090 (21.5) | 1168 (30.8) | 1463 (34.6) | 254 (30.9) | 948 (22.9) | 322 (21.0) | |
| Cigarette smoking, n (%) | <0.001 | ||||||
| Never smokers | 13,584 (62.5) | 2275 (41.0) | 3180 (47.9) | 753 (54.4) | 3041 (42.4) | 1302 (51.4) | |
| Ever smokers | 8286 (37.5) | 3283 (59.0) | 3421 (52.1) | 592 (45.6) | 3785 (57.6) | 1127 (48.6) | |
| Pack-years for smokers, mean (±SE) (y) | 15.5 ± 0.4 | 29.9 ± 0.8 | 24.4 ± 0.9 | 20.3 ± 1.6 | 22.7 ± 0.6 | 17.5 ± 0.9 | <0.001 |
| Alcohol intake, n (%) | <0.001 | ||||||
| Nondrinkers | 8122 (34.9) | 1903 (29.5) | 2376 (27.7) | 496 (30.3) | 2351 (30.6) | 896 (33.3) | |
| Moderate drinkers | 10,352 (55.4) | 2774 (56.5) | 3315 (60.1) | 664 (59.4) | 3490 (58.8) | 1200 (58.0) | |
| Heavy drinkers | 1731 (9.6) | 630 (14.1) | 576 (12.2) | 91 (10.3) | 552 (10.5) | 147 (8.7) | |
| Tea consumption, mean ± SE (g/d) | 230.8 ± 7.3 | 165.4 ± 9.0 | 176.8 ± 8.8 | 143.0 ± 14.2 | 184.4 ± 9.3 | 152.8 ± 12.3 | <0.001 |
| Healthy Eating Index-2015 score (mean ± SE) | 51.2 ± 0.2 | 55.4 ± 0.3 | 56.0 ± 0.2 | 53.7 ± 0.5 | 50.3 ± 0.3 | 48.9 ± 0.4 | <0.001 |
| Physical activity, mean ± SE, MET-(h/wk) | 18.9 ± 0.5 | 15.3 ± 0.7 | 17.0 ± 0.9 | 15.5 ± 1.3 | 13.5 ± 0.5 | 13.8 ± 0.9 | <0.001 |
| BMI, mean ± SE (kg/m2) | 29.1 ± 0.1 | 28.5 ± 0.1 | 28.7 ± 0.1 | 28.6 ± 0.3 | 28.2 ± 0.1 | 28.4 ± 0.2 | <0.001 |
| Weight status, n (%) | <0.001 | ||||||
| BMI <25 kg/m2 | 6518 (31.9) | 1575 (29.2) | 1667 (28.2) | 400 (33.8) | 2039 (32.7) | 718 (32.6) | |
| BMI 25–29.9 kg/m2 | 6666 (29.9) | 2062 (37.6) | 2365 (37.4) | 441 (30.7) | 2455 (35.1) | 864 (34.1) | |
| BMI ≥30 kg/m2 | 8390 (38.2) | 1826 (33.3) | 2457 (34.4) | 485 (35.5) | 2260 (32.1) | 821 (33.2) | |
| Total energy intake, mean ± SE (kcal/d) | 2207.7 ± 9.4 | 2085.3 ± 17.3 | 2007.8 ± 15.1 | 2097.9 ± 29.8 | 2252.5 ± 18.8 | 2265.0 ± 30.9 | <0.001 |
| Comorbid conditions, n (%) | |||||||
| Cancer | 1241 (5.8) | 821 (12.9) | 735 (10.5) | 97 (7.2) | 455 (6.5) | 132 (5.8) | <0.001 |
| Congestive heart failure | 515 (1.7) | 255 (3.3) | 253 (2.7) | 35 (1.6) | 149 (1.6) | 44 (1.3) | <0.001 |
| Coronary artery disease | 583 (2.2) | 419 (5.8) | 362 (4.2) | 57 (3.6) | 204 (2.6) | 58 (1.6) | <0.001 |
| Myocardial infarction | 608 (2.1) | 405 (5.5) | 358 (4.2) | 48 (3.0) | 246 (2.8) | 65 (2.1) | <0.001 |
| Stroke | 600 (2.1) | 288 (3.6) | 309 (3.2) | 45 (2.4) | 210 (2.3) | 64 (2.1) | <0.001 |
| High cholesterol | 5681 (24.9) | 2330 (41.2) | 2793 (39.4) | 480 (36.3) | 2216 (31.8) | 672 (25.6) | <0.001 |
| Hypertension | 9251 (39.7) | 3163 (53.3) | 3697 (48.9) | 623 (44.4) | 3074 (40.8) | 1004 (39.2) | <0.001 |
| Diabetes | 2306 (7.9) | 764 (11.1) | 1384 (13.5) | 265 (13.2) | 506 (5.1) | 210 (6.4) | <0.001 |
Abbreviations: DV, daily value; HSHF, high added sugar and high saturated fat; HSLF, low added sugar and high saturated fat; LSHF, low added sugar and high saturated fat; LSLF, low added sugar and low saturated fat.
Mean (SE) and percentage were adjusted for NHANES survey weights to generate nationally representative estimates.
LSLF coffee is coffee with low added sugar [contained <5% DV (2.5 g) per cup] and low saturated fat [<5% DV (1 g) per cup); LSHF coffee is coffee with low added sugar [contained <5% DV (2.5 g) per cup] and high saturated fat [≥5% DV (1 g) per cup); HSLF coffee is coffee with high added sugar [contained <5% DV (2.5 g) per cup] and low saturated fat [<5% DV (1 g) per cup); HSHF coffee is coffee with high added sugar [contained ≥5% DV (2.5 g) per cup] and high saturated fat [≥5% DV (1 g) per cup].
P values were generated from analysis of variances for continuous variables and chi-square tests for categorical variables.
During a median follow-up of 9.3–11.3 y, 7074 deaths occurred due to all causes, 1176 deaths due to cancer, and 1089 deaths due to CVD. Compared with noncoffee drinkers, total coffee consumption was associated with a lower risk of all-cause mortality [HR (95% CI): 0.89 (0.78, 1.02) for <1 cup/d; 0.84 (0.77, 0.92) for 1 to <2 cups/d; 0.83 (0.75, 0.93) for 2 to <3 cups/d; and 0.85 (0.77, 0.95) for ≥3 cups/d; P-trend= 0.004] (Table 3). In addition, total coffee consumption was associated with reduced CVD mortality [HR (95% CI): 0.90 (0.65, 1.25) for <1 cups/d; 0.71 (0.58, 0.85) for 1 to <2 cups/d; 0.67 (0.54, 0.84) for 2 to <3 cups/d; and 0.86 (0.70, 1.05) for ≥3 cups/d; P-trend = 0.12] (Table 3). In the cubic spline model (Figure 2), we saw a nonlinear inverse association between coffee consumption and both all-cause mortality (P-nonlinearity < 0.001) and CVD mortality (P-nonlinearity = 0.02). No significant associations were found between total coffee consumption and cancer mortality. Similar associations were observed for the consumption of caffeinated coffee, but no associations were found for the consumption of decaffeinated coffee.
TABLE 3.
Total, caffeinated, and decaffeinated coffee consumption and mortality among United States adults
| All-cause mortality1 |
Cancer mortality1 |
CVD mortality1 |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Person-years | No. of deaths | HR (95% CI)2 | Person-years | No. of deaths | HR (95% CI)3 | Person-years | No. of deaths | HR (95% CI)4 | |
| Total coffee | |||||||||
| No coffee | 217,513 | 2476 | 1.00 | 206,753 | 387 | 1.00 | 192,126 | 401 | 1.00 |
| <1 cup/d | 34,858 | 509 | 0.89 (0.78, 1.02) | 32,747 | 95 | 1.24 (0.89, 1.72) | 29,869 | 75 | 0.90 (0.65, 1.25) |
| 1 to <2 cups/d | 78,799 | 1516 | 0.84 (0.77, 0.92) | 72,395 | 246 | 0.96 (0.76, 1.21) | 66,989 | 234 | 0.71 (0.58, 0.85) |
| 2 to <3 cups/d | 48,732 | 1023 | 0.83 (0.75, 0.93) | 44,079 | 166 | 0.95 (0.73, 1.24) | 40,736 | 141 | 0.67 (0.54, 0.84) |
| ≥3 cups/d | 69,179 | 1550 | 0.85 (0.77, 0.95) | 61,981 | 282 | 1.09 (0.88, 1.35) | 57,473 | 238 | 0.86 (0.70, 1.05) |
| P-trend | 0.004 | 0.59 | 0.12 | ||||||
| Caffeinated coffee | |||||||||
| No coffee | 217,513 | 2476 | 1.00 | 206,753 | 387 | 1.00 | 192,126 | 401 | 1.00 |
| <1 cup/d | 29,564 | 374 | 0.87 (0.74, 1.01) | 27,951 | 78 | 1.22 (0.84, 1.76) | 25,368 | 59 | 1.01 (0.68, 1.48) |
| 1 to <2 cups/d | 66,523 | 1153 | 0.82 (0.74, 0.90) | 61,492 | 200 | 0.97 (0.76, 1.24) | 56,891 | 179 | 0.71 (0.58, 0.87) |
| 2 to <3 cups/d | 41,295 | 777 | 0.81 (0.72, 0.92) | 37,584 | 128 | 0.96 (0.72, 1.29) | 34,740 | 111 | 0.64 (0.50, 0.82) |
| ≥3 cups/d | 58,860 | 1227 | 0.87 (0.78, 0.97) | 53,209 | 225 | 1.09 (0.87, 1.36) | 49,258 | 194 | 0.87 (0.70, 1.10) |
| P-trend | 0.02 | 0.59 | 0.14 | ||||||
| Decaffeinated coffee | |||||||||
| No coffee | 217,513 | 2476 | 1.00 | 206,753 | 387 | 1.00 | 192,126 | 401 | 1.00 |
| <1 cup/d | 5259 | 134 | 0.97 (0.75, 1.26) | 4761 | 17 | 1.26 (0.63, 2.51) | 4466 | 16 | 0.65 (0.39, 1.11) |
| 1 to <2 cups/d | 11,502 | 345 | 0.93 (0.79, 1.09) | 10,221 | 44 | 0.89 (0.58, 1.38) | 9432 | 53 | 0.72 (0.47, 1.12) |
| 2 to <3 cups/d | 6243 | 217 | 0.97 (0.80, 1.18) | 5507 | 32 | 0.97 (0.61, 1.54) | 5034 | 27 | 0.86 (0.56, 1.31) |
| ≥3 cups/d | 7378 | 244 | 0.74 (0.62, 0.88) | 6322 | 41 | 0.97 (0.62, 1.50) | 5718 | 31 | 0.83 (0.54, 1.27) |
| P-trend | 0.003 | 0.80 | 0.16 | ||||||
Abbreviations: CI, confidence interval; CVD, cardiovascular disease; HR, hazard ratio; MET, metabolic equivalent.
All-cause mortality and CVD mortality were assessed using NHANES 1999–2018 linked mortality data; CVD mortality was analyzed using NHANES 1999–2014 linked mortality data.
The HRs and 95% CIs were adjusted for age (continuous), sex (male vs. female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and other), education (grades 0–12, some college, and college graduates and above), family income-to-poverty ratio (<1.30, 1.30–2.99, 3.00–4.90, and ≥5.00), ever smoker (yes vs. no), cigarette smoking (pack-years, continuous), alcohol intake (nondrinkers, moderate drinkers, and heavy drinkers), tea drinkers (yes vs. no), amount of tea consumption (g/d, continuous), Healthy Eating Index-2020 score (continuous), physical activity (MET-hours, continuous), BMI (kg/m2, continuous), total energy intake (kcal/d, continuous), presence of comorbid conditions at baseline (yes vs. no), and NHANES survey weights.
For analysis of cancer mortality, participants who reported cancer diagnosis at baseline were excluded.
For analysis of CVD mortality, participants who reported a diagnosis of congestive heart failure, coronary artery disease, myocardial infarction, or stroke at baseline were excluded.
FIGURE 2.
Dose–response associations of coffee consumption with all-cause, CVD, and cancer mortality. CVD, cardiovascular disease.
By coffee type, individuals who consumed black coffee [HR (95% CI): 0.86 (0.77, 0.97)] and LSLF coffee [HR (95% CI): 0.86 (0.75, 0.99)] had a lower risk of all-cause mortality, compared with noncoffee drinkers. However, for CVD mortality, black coffee consumption [HR (95% CI): 0.65 (0.50, 0.84)], LSLF coffee [HR (95% CI): 0.75 (0.57, 0.98)], and HSLF coffee [HR (95% CI): 0.71 (0.55,0.91)] were significantly associated with a reduced risk of CVD mortality. LSHF and HSHF coffee showed no significant associations with CVD mortality (Table 4). When the type of caffeinated compared with noncaffeinated coffee was further examined, the consumption of caffeinated black coffee was associated with a reduced risk of all-cause mortality [HR (95% CI): 0.88 (0.78, 0.99)] and CVD mortality [HR (95% CI): 0.64 (0.47, 0.88)], and the consumption of caffeinated LSLF coffee was associated with a reduced risk of all-cause mortality [HR (95% CI): 0.86 (0.75, 1.00)]. In contrast, drinking coffee with high levels of added sugar and saturated fat, whether caffeinated or decaffeinated coffee, was not associated with reduced mortality compared with noncoffee drinkers.
TABLE 4.
Coffee consumption and mortality among United States adults, by amount of added sugar and saturated fat1
| All-cause mortality2 |
Cancer mortality2 |
CVD mortality2 |
|||||||
|---|---|---|---|---|---|---|---|---|---|
| Person-years | No. of deaths | HR (95% CI)3 | Person-years | No. of deaths | HR (95% CI)4 | Person-years | No. of deaths | HR (95% CI)5 | |
| Total coffee | |||||||||
| No coffee | 217,513 | 2476 | 1.00 | 206,753 | 387 | 1.00 | 192,126 | 496 | 1.00 |
| Black | 53,090 | 1521 | 0.86 (0.77, 0.97) | 45,919 | 225 | 1.06 (0.81, 1.39) | 43,439 | 262 | 0.65 (0.50, 0.84) |
| LSLF | 62,119 | 1411 | 0.86 (0.75, 0.99) | 55,931 | 243 | 1.25 (0.96, 1.62) | 51,210 | 265 | 0.75 (0.57, 0.98) |
| LSHF | 13,206 | 219 | 0.85 (0.69, 1.06) | 12,371 | 28 | 1.03 (0.60, 1.79) | 11,422 | 41 | 0.63 (0.39, 1.00) |
| HSLF | 63,673 | 925 | 0.90 (0.79, 1.04) | 59,937 | 208 | 1.20 (0.88, 1.65) | 54,515 | 160 | 0.71 (0.55, 0.91) |
| HSHF | 24,338 | 284 | 0.89 (0.73, 1.08) | 23,189 | 44 | 0.77 (0.52, 1.13) | 21,477 | 69 | 1.02 (0.71, 1.46) |
| Caffeinated coffee | |||||||||
| No coffee | 217,513 | 2476 | 1.00 | 206,753 | 387 | 1.00 | 192,126 | 496 | 1.00 |
| Black | 44,806 | 1146 | 0.88 (0.78, 0.99) | 39,294 | 176 | 1.07 (0.81, 1.41) | 36,922 | 197 | 0.64 (0.47, 0.88) |
| LSLF | 50,768 | 1052 | 0.86 (0.75, 1.00) | 45,911 | 188 | 1.24 (0.94, 1.63) | 42,069 | 200 | 0.77 (0.57, 1.04) |
| LSHF | 11,053 | 167 | 0.93 (0.73, 1.19) | 10,421 | 21 | 1.11 (0.62, 1.98) | 9631 | 31 | 0.63 (0.35, 1.11) |
| HSLF | 56,064 | 768 | 0.98 (0.85, 1.13) | 52,976 | 179 | 1.23 (0.88, 1.72) | 48,213 | 139 | 0.80 (0.60, 1.06) |
| HSHF | 21,574 | 228 | 0.92 (0.74, 1.15) | 20,599 | 39 | 0.78 (0.52, 1.18) | 19,082 | 57 | 1.08 (0.73, 1.61) |
| Decaffeinated coffee | |||||||||
| No coffee | 217,513 | 2476 | 1.00 | 206,753 | 387 | 1.00 | 192,126 | 496 | 192,126 |
| Black | 7025 | 329 | 1.07 (0.89, 1.27) | 5669 | 38 | 0.77 (0.40, 1.47) | 5437 | 55 | 5437 |
| LSLF | 10,303 | 325 | 1.05 (0.85, 1.28) | 9135 | 51 | 1.08 (0.65, 1.77) | 8310 | 61 | 8310 |
| LSHF | 1967 | 47 | 0.78 (0.52, 1.18) | 1773 | 6 | 0.62 (0.20, 1.88) | 1637 | 9 | 1637 |
| HSLF | 7238 | 150 | 0.75 (0.59, 0.95) | 6659 | 25 | 0.72 (0.34, 1.52) | 5991 | 21 | 5991 |
| HSHF | 2621 | 53 | 0.97 (0.72, 1.32) | 2462 | 5 | 0.53 (0.17, 1.65) | 2260 | 12 | 2260 |
Abbreviations: CI, confidence interval; CVD, cardiovascular disease; DV, daily value; HR, hazard ratio; HSHF, high added sugar and high saturated fat; HSLF, low added sugar and high saturated fat; LSHF, low added sugar and high saturated fat; LSLF, low added sugar and low saturated fat; MET, metabolic equivalent.
LSLF coffee is coffee with low added sugar [contained < 5% DV (2.5 g) per cup] and low saturated fat [<5% DV (1 g) per cup]; LSHF coffee is coffee with low added sugar [contained <5% DV (2.5 g) per cup] and high saturated fat [≥5% DV (1 g) per cup]; HSLF coffee is coffee with high added sugar [contained <5% DV (2.5 g) per cup] and low saturated fat [<5% DV (1 g) per cup]; HSHF coffee is coffee with high added sugar [contained ≥5% DV (2.5 g) per cup] and high saturated fat [≥5% DV (1 g) per cup].
All-cause mortality and CVD mortality were assessed using NHANES 1999–2018 linked mortality data; CVD mortality was analyzed using NHANES 1999–2014 linked mortality data.
The HRs and 95% CIs were adjusted for age (continuous), sex (male vs. female), race/ethnicity (non-Hispanic White, non-Hispanic Black, Hispanic, and other), education (grades 0–12, some college, and college graduates and above), family income-to-poverty ratio (<1.30, 1.30–2.99, 3.00–4.90 and ≥5.00), ever smoker (yes vs. no), cigarette smoking (pack-years, continuous), alcohol intake (nondrinkers, moderate drinkers, and heavy drinkers), tea drinkers (yes vs. no), amount of tea consumption (g/d, continuous), Healthy Eating Index-2020 score (continuous), physical activity (MET-hours, continuous), BMI (kg/m2, continuous), total energy intake (kcal/d, continuous), presence of comorbid conditions at baseline (yes vs. no), and NHANES survey weights.
Analysis of cancer mortality excluded participants who reported cancer diagnosis at baseline.
Analysis of CVD mortality excluded participants who reported a diagnosis of congestive heart failure, coronary artery disease, myocardial infarction, or stroke at baseline.
The associations between coffee consumption and total mortality did not differ by sex, smoking status, weight status, and comorbidity status. However, the inverse association between coffee consumption and all-cause mortality was observed among participants who did not drink tea [HR (95% CI): 0.87 (0.75, 1.02) for <1 cup/d; 0.80 (0.72, 0.88) for 1 to <2 cups/d; 0.78 (0.69, 0.88) for 2 to <3 cups/d; and 0.80 (0.71, 0.89) for ≥3 cups/d] but not among tea drinkers [HR (95% CI): 1.05 (0.85, 1.31) for <1 cup/d; 0.93 (0.79, 1.09) for 1 to <2 cups/d; 0.93 (0.77, 1.14) for 2 to <3 cups/d; and 0.93 (0.77, 1.17) for ≥3 cups/d; P-interaction = 0.04]. In addition, participants aged 65 y and older showed a more pronounced relationship between sugar and cream added to coffee and mortality than other age groups [HR (95% CI): 0.88 (0.78, 1.00) for < black coffee; 0.86 (0.76, 0.99) for LSLF coffee; 0.77 (0.55, 1.09) for LSHF coffee; 0.83 (0.72, 0.96) for HSLF coffee; and 0.85 (0.65, 1.11) for HSHF coffee] (Supplemental Tables 1–6). The results from the sensitivity analysis, excluding participants from the 2017–2018 cycle, remained consistent with those obtained when this cycle was included (Supplemental Tables 7 and 8)
Discussion
In this nationally representative sample of United States adults, we found that low-to-moderate coffee consumption (1–3 cups/d) was associated with a lower risk of all-cause mortality and CVD mortality, and the associations differed by the amount of sugar and saturated fat added to the coffee. Consuming black coffee and coffee with low levels of added sugar and saturated fat was associated with a 14% lower risk of all-cause mortality compared with no coffee consumption, whereas this association was not observed for coffee with high levels of added sugar and saturated fat. Furthermore, the inverse associations between coffee consumption and mortality were largely confined to caffeinated coffee, with no significant associations with mortality found for decaffeinated coffee.
Previous studies have reported inconsistent findings on the optimal dose of coffee consumption for reducing mortality risk. For example, in the multiethnic cohort [13] and a pooled analysis of nurses and health professionals [14], the largest reduction in all-cause mortality was seen at 1–3 cups/d. In the NIH-American Association of Retired Persons (AARP) Diet and Health Study [15] and the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial study [16], the largest reduction in all-cause mortality occurred when the consumption was at 4–5 cups/d. In our study, a 16% lower risk of all-cause mortality was observed with coffee consumption above 1 cup/d. The greatest reduction (17%) in all-cause mortality occurred at 2–3 cups/d, whereas consumption beyond 3 cups/d was not associated with additional reductions in all-cause mortality. Similarly, the inverse association with CVD mortality became non-significant with consumption above 3 cups/d. These findings suggest that the benefits of coffee consumption on mortality are nonlinear and may diminish at higher levels of intake.
Few studies have examined how coffee additions may modify the association between coffee consumption and mortality risk. The 3 prior studies that investigated coffee additions found that the inverse association between coffee consumption and mortality risk persisted when milk or cream, coffee creamer, sugar, or sugar-free sweeteners were added to coffee [9,10,16]. However, these studies assessed coffee additions only as a binary variable (yes compared with no) and did not quantify the amount of additions. Our study was among the first to quantify the amount of added sugar and saturated fat in coffee. Our results align with the Dietary Guidelines for Americans, which recommend reducing added sugar and saturated fat consumption. Specifically, our findings suggest that adding sugar, saturated fat, or both to coffee beyond the 5% DV threshold diminishes the mortality benefits associated with coffee consumption.
Unlike some prior studies that reported the health benefits of both caffeinated and decaffeinated coffee [3,17], our study found that the inverse association between coffee consumption and all-cause mortality was largely limited to caffeinated coffee. The lack of significant associations between decaffeinated coffee and all-cause mortality could be due to the low decaffeinated coffee consumption (0.17 cup/d) among the United States population, resulting in limited statistical power to detect a significant association. It is also possible that the health benefits of coffee consumption are primarily due to its caffeine content. Previous studies have found that caffeine affects energy balance by increasing energy expenditure [18] and basal metabolic rate [11], and has an anti-inflammatory effect [36], which may prevent long-term weight gain [12] and insulin resistance [20,21]. Additionally, caffeine consumption itself has been associated with low mortality risks in several cohort studies [19,22,23]. The bioactive components in coffee beans can contribute to the health benefits of coffee consumption. For example, chlorogenic acid, the main phenolic component of coffee beans, has been shown to improve insulin sensitivity, inhibit hepatic triglyceride accumulation, and reduce blood pressure in both animal and human studies [24]. There are also indications that polyphenol and caffeine may positively influence gut microbiota diversity [[25], [26], [27], [28]]. Interestingly, we found that the inverse association between coffee consumption and all-cause mortality was confined to nontea drinkers. It is possible that tea consumption might alter the bioavailability of coffee’s bioactive compounds or compete with coffee in shared biological pathways, thereby reducing coffee’s protective effects. Future research is needed to further investigate the biological mechanisms underlying the health benefits of coffee consumption.
The strengths of our study include the use of a nationally representative sample of United States adults from NHANES, ensuring the generalizability of our findings. The detailed dietary intake data collected using 24-h dietary recalls allowed us to assess the type of coffee consumed (caffeinated compared with decaffeinated) as well as the amount of sugar and saturated fat added. This level of detail enabled us to evaluate the potentially heterogeneous effects of coffee consumption on mortality risk by coffee type and level of coffee additions. There are several limitations to note. First, one or 2 d of diet recalls may not fully capture typical intake due to day-to-day variations in food intake and are subject to measurement errors. However, dietary recalls allow for more detailed reporting, and total energy adjustments help reduce measurement errors by removing correlated errors. Second, coffee consumption is correlated with various demographic and lifestyle factors, including education, income, cigarette smoking, alcohol drinking, physical activity, diet quality, and weight status. Although our analysis carefully adjusted for these factors, the possibility of residual confounding cannot be entirely ruled out. Third, mortality outcomes were assessed through linkage to the National Death Index through a probabilistic match, which may lead to some misclassification. However, a prior validation study showed that the misclassification is <4% for deceased participants and <1% for living participants [29]. Fourth, information on coffee preparation methods, such as filtered compared with unfiltered coffee, was not available. Additionally, we did not categorize coffee based on sources of added sugar (granulated sugar compared with low-calorie sweeteners) or saturated fat (e.g., dairy compared with nondairy additions). Finally, coffee consumption may decrease the risk of liver and endometrial cancer, as reported by the World Cancer Research Fund 2018 Continuous Update Project [30,31]. Although we did not find an association between coffee consumption and overall cancer mortality, we were unable to distinguish between deaths from different types of cancers due to the relatively small number of cancer-related deaths in this study.
In conclusion, this United States representative prospective cohort study demonstrated that coffee consumption is associated with a lower risk of all-cause mortality. However, the inverse association was limited to the consumption of black coffee and coffee with low levels of added sugar and saturated fat. Although our findings support the health benefits of coffee consumption, the potential negative effects of adding excessive amounts of sugar or saturated fat to coffee warrant public attention.
Author contributions
The authors’ responsibilities were as follows – BZ, FFZ: conceptualized and designed the study; BZ, YP, LW: collected, cleaned, and processed the data; BZ: conducted the statistical analyses and drafted the manuscript; MR, LW, FFZ: critically revised the manuscript; and all authors: reviewed and approved the final version of the manuscript.
Data availability
The data described in the manuscript, along with the codebook and analytic code, will be made available upon request.
Funding
This study was supported by the National Institutes of Health (award number: R01MD011501). The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Conflict of interest
FFZ reports financial support was provided by National Institutes of Health. All other authors report no conflicts of interest.
Footnotes
Supplementary data to this article can be found online at https://doi.org/10.1016/j.tjnut.2025.05.004.
Appendix A. Supplementary data
The following is the Supplementary data to this article:
References
- 1.Zhao L.-G., Li Z.-Y., Feng G.-S., Ji X.-W., Tan Y.-T., Li H.-L., et al. Coffee drinking and cancer risk: an umbrella review of meta-analyses of observational studies. BMC Cancer. 2020;20:101. doi: 10.1186/s12885-020-6561-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Poole R., Kennedy O.J., Roderick P., Fallowfield J.A., Hayes P.C., Parkes J. Coffee consumption and health: umbrella review of meta-analyses of multiple health outcomes. BMJ. 2017;359 doi: 10.1136/bmj.j5024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Grosso G., Godos J., Galvano F., Giovannucci E.L. Coffee, caffeine, and health outcomes: an umbrella review. Annu. Rev. Nutr. 2017;37:131–156. doi: 10.1146/annurev-nutr-071816-064941. [DOI] [PubMed] [Google Scholar]
- 4.Kim Y., Je Y., Giovannucci E. Coffee consumption and all-cause and cause-specific mortality: a meta-analysis by potential modifiers. Eur. J. Epidemiol. 2019;34:731–752. doi: 10.1007/s10654-019-00524-3. [DOI] [PubMed] [Google Scholar]
- 5.Hu G.L., Wang X., Zhang L., Qiu M.H. The sources and mechanisms of bioactive ingredients in coffee. Food Funct. 2019;10:3113–3126. doi: 10.1039/c9fo00288j. [DOI] [PubMed] [Google Scholar]
- 6.Makiso M.U., Tola Y.B., Ogah O., Endale F.L. Bioactive compounds in coffee and their role in lowering the risk of major public health consequences: a review. Food Sci. Nutr. 2024;12:734–764. doi: 10.1002/fsn3.3848. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bailey R.L., Fulgoni V.L., Cowan A.E., Gaine P.C. Sources of added sugars in young children, adolescents, and adults with low and high intakes of added sugars. Nutrients. 2018;10:102. doi: 10.3390/nu10010102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Rehm C.D., Ratliff J.C., Riedt C.S., Drewnowski A. Coffee consumption among adults in the United States by demographic variables and purchase location: analyses of NHANES 2011–2016 data. Nutrients. 2020;12:2463. doi: 10.3390/nu12082463. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.van Dongen L.H., Mölenberg F.J., Soedamah-Muthu S.S., Kromhout D., Geleijnse J.M. Coffee consumption after myocardial infarction and risk of cardiovascular mortality: a prospective analysis in the Alpha Omega Cohort. Am. J. Clin. Nutr. 2017;106:1113–1120. doi: 10.3945/ajcn.117.153338. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Liu D., Li Z.-H., Shen D., Zhang P.-D., Song W.-Q., Zhang W.-T., et al. Association of sugar-sweetened, artificially sweetened, and unsweetened coffee consumption with all-cause and cause-specific mortality : a large prospective cohort study. Ann. Intern. Med. 2022;175:909–917. doi: 10.7326/M21-2977. [DOI] [PubMed] [Google Scholar]
- 11.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. doi: 10.1093/ajcn/33.5.989. [DOI] [PubMed] [Google Scholar]
- 12.Lopez-Garcia E., van Dam R.M., Rajpathak S., Willett W.C., Manson J.E., Hu F.B. Changes in caffeine intake and long-term weight change in men and women. Am. J. Clin. Nutr. 2006;83:674–680. doi: 10.1093/ajcn.83.3.674. [DOI] [PubMed] [Google Scholar]
- 13.Park S.-Y., Freedman N.D., Haiman C.A., Le Marchand L., Wilkens L.R., Setiawan V.W. Association of coffee consumption with total and cause-specific mortality among nonwhite populations. Ann. Intern. Med. 2017;167:228–235. doi: 10.7326/M16-2472. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Ding M., Satija A., Bhupathiraju S.N., Hu Y., Sun Q., Han J., et al. Association of coffee consumption with total and cause-specific mortality in 3 large prospective cohorts. Circulation. 2015;132:2305–2315. doi: 10.1161/CIRCULATIONAHA.115.017341. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.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;366:1891–1904. doi: 10.1056/NEJMoa1112010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Loftfield E., Freedman N.D., Graubard B.I., Guertin K.A., Black A., Huang W.-Y., 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. doi: 10.1093/aje/kwv146. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Li Q., Liu Y., Sun X., Yin Z., Li H., Cheng C., et al. Caffeinated and decaffeinated coffee consumption and risk of all-cause mortality: a dose–response meta-analysis of cohort studies. J. Hum. Nutr. Diet. 2019;32:279–287. doi: 10.1111/jhn.12633. [DOI] [PubMed] [Google Scholar]
- 18.Harpaz E., Tamir S., Weinstein A., Weinstein Y. The effect of caffeine on energy balance. J. Basic Clin. Physiol. Pharmacol. 2017;28:1–10. doi: 10.1515/jbcpp-2016-0090. [DOI] [PubMed] [Google Scholar]
- 19.Feng J., Wang J., Jose M., Seo Y., Feng L., Ge S. Association between caffeine intake and all-cause and cause-specific mortality: an analysis of the National Health and Nutrition Examination Survey (NHANES) 1999–2014 database. Nurs. Rep. 2021;11:901–912. doi: 10.3390/nursrep11040083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Yao W., Luo J., Dong X., Li Z., Zhang D. Exploration of association of the different types of coffee and caffeine intake with glucose metabolism markers among adults. J. Funct. Foods. 2021;85 [Google Scholar]
- 21.Lee S., Min J.-Y., Min K.-B. Caffeine and caffeine metabolites in relation to insulin resistance and beta cell function in U.S. adults. Nutrients. 2020;12:1783. doi: 10.3390/nu12061783. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Paganini-Hill A., Kawas C.H., Corrada M.M. Non-alcoholic beverage and caffeine consumption and mortality: the Leisure World Cohort Study. Prev. Med. 2007;44:305–310. doi: 10.1016/j.ypmed.2006.12.011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Chen S., Li J., Gao M., Li D., Shen R., Lyu L., et al. Association of caffeine intake with all-cause and cardiovascular mortality in elderly patients with hypertension. Front. Nutr. 2022;9 doi: 10.3389/fnut.2022.1023345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Tajik N., Tajik M., Mack I., Enck P. The potential effects of chlorogenic acid, the main phenolic components in coffee, on health: a comprehensive review of the literature. Eur. J. Nutr. 2017;56:2215–2244. doi: 10.1007/s00394-017-1379-1. [DOI] [PubMed] [Google Scholar]
- 25.González S., Salazar N., Ruiz-Saavedra S., Gómez-Martín M., de los Reyes-Gavilán C.G., Gueimonde M. Long-term coffee consumption is associated with fecal microbial composition in humans. Nutrients. 2020;12:1287. doi: 10.3390/nu12051287. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Dai A., Hoffman K., Xu A.A., Gurwara S., White D.L., Kanwal F., et al. The association between caffeine intake and the colonic mucosa-associated gut microbiota in humans—a preliminary investigation. Nutrients. 2023;15:1747. doi: 10.3390/nu15071747. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Hussain S.K., Dong T.S., Agopian V., Pisegna J.R., Durazo F.A., Enayati P., et al. Dietary protein, fiber and coffee are associated with small intestine microbiome composition and diversity in patients with liver cirrhosis. Nutrients. 2020;12:1395. doi: 10.3390/nu12051395. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Mansour A., Mohajeri-Tehrani M.R., Karimi S., Sanginabadi M., Poustchi H., Enayati S., et al. Short term effects of coffee components consumption on gut microbiota in patients with non-alcoholic fatty liver and diabetes: a pilot randomized placebo-controlled, clinical trial. EXCLI J. 2020;19:241–250. doi: 10.17179/excli2019-2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Menke A., Muntner P., Batuman V., Silbergeld E.K., Guallar E. Blood lead below 0.48 micromol/L (10 microg/dL) and mortality among US adults. Circulation. 2006;114:1388–1394. doi: 10.1161/CIRCULATIONAHA.106.628321. [DOI] [PubMed] [Google Scholar]
- 30.World Cancer Research Fund/American Institute for Cancer Research, Diet, nutrition, physical activity and liver cancer. https://www.wcrf.org/wp-content/uploads/2021/02/liver-cancer-report.pdf [Internet]. [10/02/2024]. Available from:
- 31.World Cancer Research Fund/American Institute for Cancer Research, Diet, nutrition, physical activity and endometrial cancer. https://www.wcrf.org/wp-content/uploads/2021/02/Endometrial-cancer-report.pdf [Internet]. [10/02/2024]. Available from:
- 32.National Center for Health Statistics. National Health and Nutrition Examination Survey [Internet]. [11/10/2024]. Available from: https://www.cdc.gov/nchs/nhanes/index.html
- 33.National Center for Health Statistics. NHANES Dietary Analyses [Internet]. [11/10/2024]. Available from: https://wwwn.cdc.gov/nchs/nhanes/tutorials/dietaryanalyses.aspx
- 34.Centers for Disease Control and Prevention. 2019 Public-Use Linked Mortality Files [Internet]. [11/10/2024]. Available from: https://www.cdc.gov/nchs/data-linkage/mortality-public.htm
- 35.National Cancer Institute. Developing the Health Eating Index [Internet]. [11/10/2024]. Available from: https://epi.grants.cancer.gov/hei/developing.html#f1b
- 36.T. Al Reef, E. Ghanem, Caffeine: Well-known as psychotropic substance, but little as immunomodulator. Immunobiology 2018;223:818–825. [DOI] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data described in the manuscript, along with the codebook and analytic code, will be made available upon request.


