Habitual Coffee and Tea Consumption and Cardiometabolic Biomarkers in the UK Biobank: The Role of Beverage Types and Genetic Variation

ABSTRACT

Background

Mechanisms linking habitual consumption of coffee and tea to the development of type 2 diabetes and cardiovascular diseases remain unclear.

Objectives

We leveraged dietary, genetic, and biomarker data collected from the UK Biobank to investigate the role of different varieties of coffee and tea in cardiometabolic health.

Methods

We included data from ≤447,794 participants aged 37–73 y in 2006–2010 who provided a blood sample and completed questionnaires regarding sociodemographic factors, medical history, diet, and lifestyle. Multivariable linear regression was used to examine the association between coffee or tea consumption and blood concentrations of glycated hemoglobin, fasting glucose, total cholesterol, HDL cholesterol, LDL cholesterol, fasting triglycerides (TGs), apoA-1, apoB, lipoprotein-a, and C-reactive protein (CRP). Lifestyle and genetic factors affecting caffeine metabolism, responses, or intake were tested for interactions with beverage intake in relation to biomarker concentrations.

Results

Compared with coffee nonconsumers, each additional cup of coffee was significantly associated with higher total cholesterol, HDL-cholesterol, and LDL-cholesterol concentrations and lower TG and CRP concentrations in both men and women (P-trend < 0.002). Higher consumption of espresso coffee (≥2 compared with 0 cups/d) was associated with higher LDL cholesterol in men (β: 0.110 mmol/L; 95% CI: 0.058, 0.163 mmol/L) and women (β: 0.161 mmol/L; 95% CI: 0.088, 0.234 mmol/L), whereas no substantial association was observed for instant coffee. Compared with tea nonconsumers, higher tea consumption was associated with lower total and LDL cholesterol and apoB and higher HDL cholesterol (P-trend < 0.002); these associations were similar for black and green tea. Associations were not modified by genetics.

Conclusions

In the UK Biobank, consumption of certain coffee brews such as espresso had unfavorable associations with blood lipids, whereas consumption of tea had favorable associations. Findings were not modified by genetic variants affecting caffeine metabolism, suggesting a role of noncaffeine constituents of these beverages in cardiometabolic health.

Introduction

Coffee and tea are among the most widely consumed beverages in the world. In prospective epidemiological studies, habitual coffee and tea consumption has consistently been associated with a lower risk of type 2 diabetes (T2D) and cardiovascular diseases (CVDs) (1–3). However, coffee and tea consumption did not substantially improve concentrations of classic biomarkers of these diseases in randomized controlled trials (RCTs), which provided limited insight into potential mechanisms that may underlie the observed inverse association between coffee or tea consumption and risk of T2D and CVD (4–7). This might be interpreted as evidence for noncausal effects, but small sample sizes, short trial duration, the type of coffee or tea, and treatment-response variability may, in part, explain inconsistent or lack of change in some biomarkers.

The evaluation of gene × diet interactions is a promising approach to gain mechanistic insights into the relation between dietary factors and disease outcomes. Coffee and tea are major dietary sources of caffeine, which has well-characterized physiological effects that might vary as a result of genetic variation in caffeine-related pathways (8). These beverages also contain multiple other components including polyphenols and minerals which might potentially affect pathways related to T2D and CVD (9). Thus, leveraging known genetic variation in caffeine pathways may assist with separating the effects of caffeine from the effects of these other phytochemicals. In addition, it is important to consider coffee brew type and the type of tea (green compared with black tea) because these differ substantially in their content of relevant phytochemicals (10). For example, the content of the cholesterol-raising diterpene cafestol differs by type of coffee brew with high concentrations in Turkish and French Press (cafetiere) coffee, low concentrations in drip-filtered and instant coffee, and intermediate concentrations in espresso-based coffees (11, 12). The UK Biobank is a large population cohort of adults who underwent medical, sociodemographic, and lifestyle assessment and also provided biological samples. Coffee and tea consumption were assessed using both FFQs and 24-h recalls, with the latter providing more detailed information on different types of coffee and tea. The current study uses this valuable resource to investigate the impact of habitual coffee and tea consumption on biomarkers of CVD and T2D (herein referred to as “cardiometabolic” biomarkers) while in addition accounting for genetic variation in caffeine-related pathways.

Methods

UK Biobank

In 2006–2010, the UK Biobank recruited >502,633 participants aged 37–73 y at 22 centers across England, Wales, and Scotland (13). Participants completed touchscreen questionnaires on sociodemographic factors, lifestyle, and medical history followed by an in-person questionnaire, a physical assessment, and biospecimen collection. The Supplemental Methods and Supplemental Table 1 provide additional study details. This study was covered by the generic ethical approval for UK Biobank studies from the National Research Ethics Service Committee North West–Haydock (approval letter dated 17 June, 2011, Ref: 11/NW/0382), and all study procedures were performed in accordance with the World Medical Association Declaration of Helsinki ethical principles for medical research.

Coffee and tea assessment

The touchscreen questionnaire included an FFQ assessing consumption of commonly used food and beverage items (14). For coffee intake, participants were asked, “How many cups of coffee do you drink each DAY (include decaffeinated coffee).” Participants either selected the number of cups, “less than one,” “do not know,” or “prefer not to answer.” Coffee drinkers were then asked about the type of coffee they usually consume and selected 1 of “decaffeinated (any type),” “instant coffee,” “ground coffee (include espresso, filter etc.),” or “other.” A similar question was asked about tea (“include black and green tea”). Other dietary sources of caffeine were not captured by the FFQ. We estimated total caffeine (mg/d) from regular coffee and tea by assigning each cup 75 mg and 40 mg caffeine, respectively. A subset of 122,283 participants also completed ≥2 of five 24-h diet recall questionnaires in 2009–2012 (14). In this subset of participants, the correlation coefficients (r) between their FFQ and mean dietary-recall coffee and tea intakes were 0.82 and 0.81, respectively. The 24-h diet recalls collected additional information on coffee (cups per type), tea (cups per type), milk/sweeteners added to beverages, and consumption of other foods and beverages (see Supplemental Methods). Data collected via 24-h diet recalls also allowed derivation of the energy content (kJ) and macronutrient composition of the diet. Because the FFQ items pertaining to type of coffee or tea consumed were limited, we used the dietary-recall data for our secondary analysis as detailed below.

Recent caffeine intake and other covariates

Participants were not asked to fast before the assessment visit. During the physical assessment period participants completed a screening questionnaire for spirometry testing that also asked whether they had consumed a caffeine-containing beverage (yes/no) within the last hour. A blood sample was collected ∼15 min after spirometry testing and at that time participants were asked to provide “Time since last meal or drink (except plain water)” (i.e., fasting time). Self-reported information on several covariates was also collected during the UK Biobank assessment as described in detail previously (13, 15). We considered baseline smoking status; Townsend deprivation index (a composite socioeconomic status metric based on employment status, car or home ownership, and household crowding) (16); education level; income level; home ownership; physical activity; ethnicity; employment status; self-rated health; alcohol intake; consumption of fish, red meat, fruit, and vegetables; diabetes (diagnosis or insulin use); CVD (myocardial infarction, angina, or stroke); aspirin use; and cholesterol-lowering or antihypertensive medication use.

Serum cardiometabolic biomarkers

Nonfasting venous blood sampling was conducted using standardized collection procedures (17). Blood samples were stored at −80°C at a central laboratory until analysis. Biomarker measures were performed using enzymatic [glucose, total cholesterol (tChol), triglycerides (TGs)], immuno-turbidimetric {apoA-1, apoB, high-sensitivity C-reactive protein [CRP], lipoprotein (a) [Lp(a)]}, enzyme immuno-inhibition (HDL cholesterol), or enzymatic selective protection (LDL cholesterol) methodology on the Beckman Coulter AU5800 platform. Three levels of internal quality control (QC) were used for each assay: low, medium, and high concentration. CVs across these 3 levels ranged from 1.4% (tChol) to 6.1% [Lp(a)] with most in the 1.0%–2.0% range. RBC glycated hemoglobin (HbA1c) was measured by HPLC on a Bio-Rad Variant II Turbo. Two levels of internal QC, low and high concentration, were used for this assay and the CVs were 2.1% and 1.5%, respectively. All biomarker values deemed invalid by the UK Biobank Biomarker Working Groups were set to missing. Further details can be found online (18). We set biomarker values ≥±4 SD from the mean to missing which applied to <1.5% of the data. For glucose and TGs we only considered data from individuals fasting for ≥6 h before blood draw (∼10% of the cohort).

Genetic data

All UK Biobank participants were genotyped using genome-wide arrays as detailed previously (19). QC and imputation to the HRC v1.1 and UK10K reference panels were performed by the Wellcome Trust Centre for Human Genetics (19). We excluded sample outliers based on heterozygosity and missingness, participants with sex discrepancies between the self-reported and X-chromosome heterozygosity, and those potentially related to other participants, based on estimated kinship coefficients for all pairs of samples. Genetic analyses performed in the current study were limited to unrelated individuals who self-reported as white British and who had very similar ancestral backgrounds based on the results of principal component analysis (19). From the genetic data we selected genome-wide association study (GWAS)-confirmed single-nucleotide polymorphisms (SNPs) for caffeine-related traits with plausible links to caffeine metabolism/response and that have minor allele frequencies ≥0.01 (Supplemental Table 2) (8, 20–24). We supplemented this list with rs762551 (CYP1A2, encoding cytochrome P450 1A2) and rs5751876 (ADORA2A, encoding adenosine receptor A_2A), candidate SNPs from the literature (24, 25), and rs1260326 (GCKR, encoding glucokinase regulator) and rs7800944 (MLXIPL, encoding MLX interacting protein like) because they associate with both coffee consumption behavior and cardiometabolic variables of interest (i.e., tChol, glucose, CRP) in GWASs (26). We did not combine SNPs for a “genetic score” because each locus captures a unique potential mechanism linking caffeine intakes to health outcomes.

Statistical analysis

Information on habitual coffee and tea intake, recent caffeine intake, and data for ≥1 biomarker at baseline were available for 447,794 participants and 370,193 were included in the genetic analysis. The number of participants varied across biomarkers (see Supplemental Table 1 for detailed sample sizes). CRP, Lp(a), and fasting TG values were log transformed before analysis because they were not distributed normally and their geometric means and 95% CIs are presented. We conducted the main analysis separately for men and women. ANOVA and chi-square tests were used as appropriate for comparisons of means and proportions across categories of coffee and tea intake.

We examined the association between habitual coffee consumption (cups/d, with intakes >15 cups/d recoded to 15 cups/d) and each biomarker using linear regression adjusting for age, ethnicity, assessment center, date of blood draw (deciles), and fasting time (model 1). In a second multivariable regression model, we further adjusted for baseline smoking (never, past, current: <10, 10–19, ≥20 cigarettes/d), Townsend deprivation index (quartiles), education (college or university degree, A levels/AS levels or equivalent, O levels/General Certificates of Secondary Education or equivalent, Certificates of Secondary Education or equivalent, National Vocational Qualification or Higher National Diploma or Higher National Certificate or equivalent, or other professional qualifications), income (4 levels), home ownership (yes, no), physical activity (quartiles of moderate/vigorous activity; min/wk), employment status (employed, retired, other), waist-to-hip ratio (WHR; quartiles), BMI (in kg/m²; <23, 23 <25, 25 <30, ≥30), oral contraceptive use (women only), postmenopausal hormone use (women only), self-rated health (excellent, good, fair, poor), aspirin use (yes, no), cholesterol-lowering medication use (yes, no), antihypertension medication use (yes, no), diabetes (yes, no), CVD (yes, no), recent caffeine intake (yes or no), and habitual intakes of water, alcohol, fish, red meat, fruits, and vegetables (quartiles of servings/d) as well as tea (categories) (model 2). Missing indicator variables, applicable to <5% of the sample, were constructed to maximize sample size (see Supplemental Table 1). In a separate model, coffee was modeled categorically: none (referent), <1, 1, 2–3, 4–5, 6–7, and ≥8 cups/d. The same statistical analyses described for coffee were also applied to the analysis of tea. Statistical significance was defined as P < 0.0025, after applying multiple testing correction for 10 biomarkers and 2 primary beverage traits [P = 0.05/(10 biomarkers × 2 exposures)]. To determine the clinical relevance of significant linear associations, we examined the risk of abnormal biomarker concentrations with beverage intake by applying clinical cutoffs to each biomarker and performing multivariable logistic regression analysis. To evaluate sensitivity to the coffee/tea assessment method, separate models were run using dietary data collected by multiple 24-h recalls in a subset of the same participants (n = 112,521). These diet recalls included more details on the type of coffee and tea consumed which allowed us to further explore our beverage–biomarker associations derived from FFQ data. We employed the same linear statistical models as aforementioned but in addition adjusted for total energy intake (kJ); intakes of protein (%kJ/d), total fat (%kJ/d), saturated fat (%kJ/d), milk (servings/d), and fizzy drinks (soda; servings/d); and the addition of milk and/or sugar to coffee and tea (yes/no) (see Supplemental Tables 3 and 4 for sample sizes).

In sensitivity analysis we excluded individuals reporting diabetes, CVD, or the use of cholesterol-lowering or antihypertension medication. We screened interactions with age (<55 or ≥55 y of age), smoking (nonsmoker, current smoker), BMI (<25, ≥25), and WHR (≤0.85/0.9 and >0.85/0.9 for women and men, respectively) by including in multivariable regression models the cross-product term of beverage intake with the interacting variable. Interactions with SNPs listed in Supplemental Table 2 were also tested using the same models as aforementioned and assuming an additive allele model. Significant interactions were also defined as P < 0.0025, and the nature of these interactions was described by stratified analysis.

Results

Participant characteristics

Supplemental Tables 5–8 present descriptive characteristics across categories of habitual coffee and tea consumption. Compared with coffee abstainers, coffee consumers were more likely to be white and drink more alcohol, but less likely to be taking antihypertensive medications. Heavy coffee drinkers were younger and more likely to be current smokers, have diabetes and poor self-reported health, and have a higher BMI than light drinkers. Tea drinkers were more likely to have diabetes, consume more coffee, and have a higher BMI than tea abstainers. Heavy tea drinkers were more likely to be white and current smokers, have coronary heart disease and a lower income, and consume less alcohol and coffee than light tea drinkers. Among the subset with multiple 24-h diet recalls, higher coffee consumption was associated with higher total and saturated fat intake and lower milk and carbohydrate intake. Higher tea consumption was associated with lower fizzy drink and milk consumption.

Habitual coffee intake and cardiometabolic biomarkers

For both women and men, tChol, LDL-cholesterol, and apoB concentrations were higher with each additional cup of coffee (Table 1). Based on the more detailed data from the 24-h recalls, the largest effect sizes for LDL cholesterol and apoB were observed for “ground coffee” (filtered/americano/cafetiere) and espresso (Tables 2, 3). Higher consumption of espresso coffee was associated with higher LDL cholesterol in men (β: 0.058 mmol/L per cup; 95% CI: 0.037, 0.080 mmol/L per cup) and women (β: 0.078 mmol/L per cup; 95% CI: 0.047, 0.109 mmol/L per cup). In contrast, instant coffee consumption was not substantially associated with higher LDL-cholesterol concentrations. Coffee consumption was inversely associated with fasting TGs, which was observed for all of the most commonly consumed coffee types (i.e., regular, instant, and “ground”). Coffee consumption was also significantly associated with higher HDL-cholesterol and apoA-1 concentrations, but this was only observed for low to medium consumption and did not follow a dose-response pattern. Although direct associations with HDL cholesterol were observed for all commonly consumed coffee types, associations with apoA-1 were less consistent with inverse associations for instant and decaffeinated coffee in women.

TABLE 1.

Associations between coffee consumption and cardiometabolic biomarkers¹

	Women			Men
Cups/d	n	Model 12, β (95% CI)	Model 23, β (95% CI)	n	Model 12, β (95% CI)	Model 23, β (95% CI)
ApoA-1, g/L
0	49,664	Ref.	Ref.	35,923	Ref.	Ref.
<1	15,569	0.03 (0.02, 0.03)5	0.01 (0.006, 0.01)5	12,672	0.02 (0.01, 0.02)5	0.005 (0.001, 0.009)4
1	44,595	0.04 (0.03, 0.04)5	0.01 (0.009, 0.01)5	33,808	0.02 (0.02, 0.03)5	0.005 (0.002, 0.008)5
2–3	64,327	0.04 (0.04, 0.05)5	0.02 (0.01, 0.02)5	57,157	0.02 (0.02, 0.03)5	0.005 (0.002, 0.008)5
4–5	25,651	0.02 (0.02, 0.03)5	0.01 (0.007, 0.01)5	26,737	0.01 (0.008, 0.02)5	0.004 (0.000, 0.007)4
6–7	7763	0.002 (−0.004, 0.008)	0.01 (0.004, 0.02)5	8339	−0.007 (−0.01, −0.001)4	−0.002 (−0.007, 0.003)
≥8	3246	−0.03 (−0.04, −0.02)5	−0.001 (−0.01, 0.008)	4433	−0.02 (−0.03, −0.01)5	−0.002 (−0.008, 0.005)
Cups/d, trend		−0.0003 (−0.0009, 0.0003)	0.0007 (0.0002, 0.001)4		−0.002 (−0.002, −0.001)5	−0.0003 (−0.0008, 0.0002)
ApoB, g/L
0	54,999	Ref.	Ref.	38,773	Ref.	Ref.
<1	17,170	0.001 (−0.003, 0.005)	0.003 (−0.0004, 0.007)	13,731	0.005 (0.0002, 0.009)4	0.002 (−0.003, 0.006)
1	49,298	0.004 (0.001, 0.007)4	0.006 (0.004, 0.009)5	36,556	0.01 (0.007, 0.01)5	0.006 (0.003, 0.01)5
2–3	71,314	0.01 (0.01, 0.01)5	0.009 (0.006, 0.01)5	61,694	0.02 (0.01, 0.02)5	0.007 (0.004, 0.01)5
4–5	28,377	0.02 (0.02, 0.02)5	0.008 (0.005, 0.01)5	28,792	0.02 (0.02, 0.03)5	0.009 (0.005, 0.01)5
6–7	8558	0.02 (0.02, 0.03)5	0.003 (−0.002, 0.008)	8929	0.03 (0.03, 0.04)5	0.01 (0.007, 0.02)5
≥8	3616	0.03 (0.02, 0.04)5	0.006 (−0.001, 0.01)	4794	0.03 (0.02, 0.03)5	0.02 (0.009, 0.02)5
Cups/d, trend		0.004 (0.003, 0.004)5	0.001 (0.0005, 0.001)5		0.004 (0.003, 0.004)5	0.002 (0.001, 0.002)5
Lp(a), log nmol/L
0	50,019	Ref.	Ref.	35,936	Ref.	Ref.
<1	15,691	−0.007 (−0.03, 0.01)	−0.005 (−0.03, 0.02)	12,678	−0.008 (−0.03, 0.02)	−0.009 (−0.03, 0.02)
1	44,955	−0.008 (−0.02, 0.007)	−0.007 (−0.02, 0.008)	33,840	−0.02 (−0.03, 0.003)	−0.02 (−0.03, 0.003)
2–3	64,864	−0.01 (−0.03, 0.003)	−0.01 (−0.02, 0.004)	57,200	−0.02 (−0.04, −0.007)4	−0.02 (−0.04, −0.002)4
4–5	25,827	−0.01 (−0.03, 0.006)	−0.01 (−0.03, 0.008)	26,751	−0.02 (−0.04, 0.0004)	−0.008 (−0.03, 0.01)
6–7	7816	−0.009 (−0.04, 0.02)	−0.009 (−0.04, 0.02)	8343	−0.02 (−0.04, 0.01)	0.001 (−0.03, 0.03)
≥8	3265	−0.03 (−0.07, 0.009)	−0.03 (−0.08, 0.01)	4435	−0.05 (−0.09, −0.02)4	−0.04 (−0.07, 0.004)
Cups/d, trend		−0.002 (−0.005, 0.0007)	−0.002 (−0.005, 0.001)		−0.004 (−0.006, −0.0009)4	−0.002 (−0.005, 0.001)
tChol, mmol/L
0	55,081	Ref.	Ref.	38,985	Ref.	Ref.
<1	17,191	0.05 (0.03, 0.07)5	0.03 (0.01, 0.05)5	13,785	0.06 (0.04, 0.08)5	0.02 (0.002, 0.04)4
1	49,344	0.09 (0.07, 0.10)5	0.06 (0.05, 0.07)5	36,721	0.09 (0.07, 0.10)5	0.05 (0.03, 0.06)5
2–3	71,428	0.13 (0.12, 0.15)5	0.09 (0.08, 0.10)5	61,954	0.13 (0.11, 0.14)5	0.06 (0.05, 0.07)5
4–5	28,412	0.15 (0.13, 0.16)5	0.09 (0.08, 0.11)5	28,924	0.15 (0.13, 0.16)5	0.08 (0.06, 0.09)5
6–7	8573	0.12 (0.10, 0.15)5	0.08 (0.05, 0.10)5	8967	0.15 (0.13, 0.18)5	0.09 (0.07, 0.11)5
≥8	3610	0.14 (0.10, 0.17)5	0.09 (0.06, 0.13)5	4826	0.11 (0.08, 0.14)5	0.10 (0.07, 0.13)5
Cups/d, trend		0.02 (0.02, 0.02)5	0.01 (0.01, 0.02)5		0.02 (0.02, 0.02)5	0.01 (0.01, 0.01)5
HDL-C, mmol/L
0	44,522	Ref.	Ref.	31,191	Ref.	Ref.
<1	13,938	0.05 (0.04, 0.05)5	0.02 (0.01, 0.02)5	11,005	0.03 (0.02, 0.03)5	0.007 (0.002, 0.01)4
1	39,553	0.06 (0.05, 0.06)5	0.02 (0.01, 0.02)5	29,338	0.03 (0.03, 0.03)5	0.005 (0.001, 0.009)4
2–3	57,492	0.06 (0.06, 0.07)5	0.03 (0.02, 0.03)5	49,521	0.03 (0.03, 0.03)5	0.007 (0.003, 0.01)5
4–5	22,689	0.03 (0.03, 0.04)5	0.03 (0.02, 0.03)5	23,073	0.01 (0.008, 0.02)5	0.008 (0.004, 0.01)5
6–7	6844	0.008 (−0.001, 0.02)	0.03 (0.02, 0.04)5	7158	−0.009 (−0.02, −0.002)4	0.004 (−0.002, 0.01)
≥8	2898	−0.03 (−0.04, −0.01)5	0.02 (0.01, 0.04)5	3877	−0.03 (−0.04, −0.02)5	0.005 (−0.003, 0.01)
Cups/d, trend		0.0003 (−0.0005, 0.001)	0.004 (0.003, 0.005)5		−0.003 (−0.003, −0.002)5	0.0006 (−0.00002, 0.001)
LDL-C, mmol/L
0	54,700	Ref.	Ref.	38,711	Ref.	Ref.
<1	17,114	0.02 (0.003, 0.03)4	0.02 (0.004, 0.03)4	13,691	0.04 (0.02, 0.05)5	0.01 (0.0003, 0.03)4
1	49,114	0.04 (0.03, 0.05)5	0.04 (0.03, 0.05)5	36,519	0.06 (0.05, 0.07)5	0.04 (0.03, 0.05)5
2–3	71,089	0.07 (0.06, 0.08)5	0.06 (0.05, 0.07)5	61,665	0.09 (0.08, 0.10)5	0.05 (0.04, 0.06)5
4–5	28,241	0.10 (0.09, 0.11)5	0.07 (0.06, 0.08)5	28,758	0.12 (0.10, 0.13)5	0.07 (0.06, 0.08)5
6–7	8520	0.10 (0.08, 0.12)5	0.05 (0.04, 0.07)5	8900	0.13 (0.11, 0.15)5	0.08 (0.07, 0.10)5
≥8	3593	0.13 (0.10, 0.16)5	0.07 (0.05, 0.10)5	4790	0.11 (0.08, 0.13)5	0.10 (0.08, 0.12)5
Cups/d, trend		0.02 (0.02, 0.02)5	0.01 (0.008, 0.01)5		0.02 (0.01, 0.02)5	0.01 (0.01, 0.01)5
Fasting TGs, log mmol/L
0	6074	Ref.	Ref.	5500	Ref.	Ref.
<1	1676	−0.03 (−0.06, −0.009)4	−0.006 (−0.03, 0.02)	1755	−0.01 (−0.04, 0.01)	−0.003 (−0.03, 0.02)
1	4525	−0.04 (−0.05, −0.02)5	−0.01 (−0.03, 0.004)	4327	−0.01 (−0.03, 0.007)	−0.01 (−0.03, 0.008)
2–3	6620	−0.04 (−0.05, −0.02)5	−0.02 (−0.03, −0.004)4	7205	−0.02 (−0.04, −0.001)4	−0.02 (−0.04, −0.005)4
4–5	2752	−0.02 (−0.04, 0.003)	−0.03 (−0.05, −0.01)5	3485	−0.002 (−0.02, 0.02)	−0.02 (−0.04, −0.0001)4
6–7	931	0.03 (−0.003, 0.06)	−0.02 (−0.05, 0.004)	1220	0.01 (−0.02, 0.04)	−0.03 (−0.06, −0.0007)4
≥8	479	0.04 (−0.003, 0.08)	−0.04 (−0.08, −0.004)4	737	0.01 (−0.03, 0.05)	−0.03 (−0.07, 0.008)
Cups/d, trend		0.003 (−0.0003, 0.005)	−0.005 (−0.008, −0.002)5		0.002 (−0.001, 0.005)	−0.004 (−0.007, −0.001)4
CRP, log mg/L
0	54,417	Ref.	Ref.	38,524	Ref.	Ref.
<1	17,014	−0.11 (−0.13, −0.10)5	−0.05 (−0.06, −0.03)5	13,617	−0.09 (−0.10, −0.07)5	−0.04 (−0.06, −0.02)5
1	48,869	−0.14 (−0.15, −0.13)5	−0.05 (−0.06, −0.04)5	36,327	−0.11 (−0.12, −0.09)5	−0.04 (−0.05, −0.03)5
2–3	70,746	−0.15 (−0.16, −0.14)5	−0.09 (−0.10, −0.08)5	61,272	−0.10 (−0.12, −0.09)5	−0.05 (−0.06, −0.04)5
4–5	28,129	−0.10 (−0.11, −0.08)5	−0.12 (−0.14, −0.11)5	28,627	−0.04 (−0.06, −0.03)5	−0.05 (−0.07, −0.04)5
6–7	8477	−0.03 (−0.05, −0.005)4	−0.13 (−0.15, −0.11)5	8861	0.05 (0.02, 0.07)5	−0.04 (−0.06, −0.02)5
≥8	3578	−0.04 (−0.07, −0.004)4	−0.19 (−0.22, −0.16)5	4754	0.12 (0.09, 0.14)5	−0.04 (−0.07, −0.01)4
Cups/d, trend		−0.006 (−0.008, −0.004)5	−0.02 (−0.02, −0.02)5		0.01 (0.008, 0.01)5	−0.004 (−0.006, −0.002)5
Fasting glucose, mmol/L
0	5572	Ref.	Ref.	5122	Ref.	Ref.
<1	1527	−0.02 (−0.05, 0.02)	−0.009 (−0.04, 0.02)	1630	0.008 (−0.03, 0.05)	−0.002 (−0.04, 0.03)
1	4109	0.02 (−0.003, 0.04)	0.02 (−0.006, 0.04)	3974	0.009 (−0.02, 0.04)	−0.002 (−0.03, 0.03)
2–3	6062	0.01 (−0.009, 0.03)	0.005 (−0.02, 0.03)	6695	0.04 (0.009, 0.06)4	0.03 (0.004, 0.05)4
4–5	2528	−0.005 (−0.03, 0.02)	−0.005 (−0.03, 0.02)	3235	0.04 (0.003, 0.07)4	0.03 (−0.002, 0.06)
6–7	843	−0.009 (−0.05, 0.03)	−0.003 (−0.04, 0.04)	1150	0.03 (−0.02, 0.07)	0.03 (−0.02, 0.07)
≥8	434	−0.04 (−0.09, 0.02)	−0.01 (−0.07, 0.04)	689	−0.01 (−0.05, 0.04)	0.02 (−0.04, 0.07)
Cups/d, trend		−0.002 (−0.006, 0.001)	−0.002 (−0.006, 0.002)		0.002 (−0.002, 0.006)	0.004 (−0.0006, 0.008)
HbA1c, mmol/mol
0	54,084	Ref.	Ref.	38,257	Ref.	Ref.
<1	16,936	−0.13 (−0.20, −0.06)5	0.04 (−0.02, 0.10)	13,543	−0.20 (−0.30, −0.10)5	0.02 (−0.06, 0.11)
1	48,603	−0.24 (−0.29, −0.19)5	0.01 (−0.03, 0.06)	36,180	−0.23 (−0.31, −0.16)5	0.06 (0.003, 0.13)4
2–3	70,611	−0.29 (−0.33, −0.24)5	−0.04 (−0.09, −0.0004)4	61,094	−0.26 (−0.33, −0.19)5	0.04 (−0.01, 0.10)
4–5	28,054	−0.15 (−0.20, −0.09)5	−0.07 (−0.13, −0.01)4	28,474	0.07 (−0.007, 0.16)	0.11 (0.05, 0.18)5
6–7	8445	0.02 (−0.07, 0.11)	−0.08 (−0.17, 0.004)	8839	0.43 (0.30, 0.55)5	0.17 (0.07, 0.27)5
≥8	3566	0.38 (0.24, 0.52)5	−0.03 (−0.15, 0.10)	4703	0.89 (0.73, 1.05)5	0.24 (0.11, 0.37)5
Cups/d, trend		0.007 (−0.002, 0.02)	−0.02 (−0.03, −0.01)5		0.08 (0.07, 0.09)5	0.02 (0.01, 0.03)5

¹β coefficients represent change in serum biomarker concentration with increasing coffee intake relative to no coffee intake. CRP, Lp(a), and TG concentrations were log transformed before analysis and thus β coefficients represent percentage change. At least nominally significant tests for sex heterogeneity were observed for HDL (P = 0.003), CRP (P < 0.0001), and HbA1c (P < 0.0001). CRP, C-reactive protein; HbA1c, glycated hemoglobin; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; Lp(a), lipoprotein (a); tChol, total cholesterol; TG, triglyceride.

²Model 1: adjusted for age, race, date of blood draw, fasting time, and assessment center.

³Model 1 further adjusted for Townsend deprivation index; education; income; employment status; home ownership; smoking; physical activity; waist-to-hip ratio; BMI; oral contraceptive use (women only); postmenopausal hormone use (women only); self-rated health; aspirin use; cholesterol-lowering medication use; antihypertension medication use; history of diabetes; history of cardiovascular disease; intakes of water, alcohol, fish, red meat, fruits, and vegetables; recent caffeine intake; and habitual tea intake.

⁴0.0025 ≤ P < 0.05.

⁵ P < 0.0025.

TABLE 2.

Associations between coffee consumption (24-h diet recalls) and cardiometabolic biomarkers among women¹

Cups/d	Regular2	Decaf2	Instant	Filtered/americano/cafetiere2	Cappuccino2	Latte2	Espresso2
ApoA-1, g/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.012 (0.005, 0.019)4	−0.002 (−0.009, 0.006)	0.004 (−0.003, 0.010)	0.011 (0.006, 0.017)4	−0.002 (−0.009, 0.005)	−0.004 (−0.011, 0.003)	0.015 (0.002, 0.027)3
1	0.019 (0.011, 0.028)4	−0.0007 (−0.013, 0.011)	0.006 (−0.002, 0.014)	0.015 (0.007, 0.024)4	0.007 (−0.009, 0.023)	−0.007 (−0.024, 0.010)	0.023 (−0.007, 0.053)
2–3	0.019 (0.013, 0.026)4	−0.002 (−0.010, 0.005)	0.0003 (−0.006, 0.007)	0.025 (0.018, 0.032)4	0.014 (−0.004, 0.031)	−0.003 (−0.022, 0.016)	0.036 (0.011, 0.061)3
4–5	0.016 (0.006, 0.025)4	−0.034 (−0.049, −0.019)4	−0.008 (−0.017, 0.002)	0.029 (0.008, 0.049)3	—	—	—
≥6	0.004 (-0.016, 0.024)	—	−0.031 (−0.052, −0.010)3	—	—	—	—
Cups/d, trend	0.003 (0.0006, 0.005)3	−0.006 (−0.009, −0.003)4	−0.003 (−0.005, −0.001)3	0.011 (0.008, 0.014)4	0.006 (−0.002, 0.014)	−0.005 (−0.013, 0.002)	0.020 (0.009, 0.030)4
ApoB, g/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.004 (−0.002, 0.010)	0.0006 (−0.005, 0.007)	0.006 (0.001, 0.012)3	0.004 (−0.001, 0.009)	0.001 (−0.005, 0.007)	−0.005 (−0.011, 0.001)	0.014 (0.004, 0.024)3
1	0.0009 (−0.006, 0.008)	0.008 (−0.002, 0.018)	−0.001 (−0.008, 0.006)	0.004 (−0.003, 0.011)	0.005 (−0.008, 0.019)	−0.009 (−0.023, 0.004)	0.022 (−0.003, 0.047)
2–3	0.006 (0.0007, 0.011)3	0.005 (−0.0007, 0.012)	0.001 (−0.005, 0.006)	0.016 (0.010, 0.022)4	−0.001 (−0.015, 0.014)	0.006 (−0.010, 0.021)	0.040 (0.020, 0.061)4
4–5	−0.004 (−0.012, 0.004)	0.004 (−0.009, 0.016)	−0.015 (−0.023, −0.007)4	0.025 (0.009, 0.042)3	—	—	—
≥6	−0.002 (−0.018, 0.015)	—	−0.004 (−0.021, 0.014)	—	—	—	—
Cups/d, trend	−0.001 (−0.003, 0.0005)	0.0007 (−0.002, 0.003)	−0.003 (−0.005, −0.001)4	0.007 (0.004, 0.009)4	0.001 (−0.005, 0.007)	−0.001 (−0.007, 0.006)	0.018 (0.009, 0.027)4
Lp(a), log nmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.017 (−0.017, 0.051)	−0.020 (−0.056, 0.016)	0.016 (−0.016, 0.049)	0.007 (−0.022, 0.035)	−0.024 (−0.058, 0.010)	−0.027 (−0.062, 0.008)	−0.061 (−0.123, 0.0002)
1	0.015 (−0.025, 0.055)	0.021 (−0.037, 0.080)	−0.023 (−0.064, 0.019)	0.024 (−0.020, 0.067)	−0.025 (−0.103, 0.053)	0.009 (−0.073, 0.091)	−0.064 (−0.212, 0.084)
2–3	−0.0002 (−0.032, 0.031)	0.009 (−0.028, 0.046)	0.003 (−0.029, 0.034)	−0.004 (−0.039, 0.031)	−0.021 (−0.108, 0.067)	−0.013 (−0.106, 0.080)	−0.028 (−0.149, 0.093)
4–5	0.007 (−0.040, 0.054)	0.003 (−0.071, 0.078)	0.025 (−0.023, 0.072)	0.040 (−0.060, 0.140)	—	—	—
≥6	0.144 (0.046, 0.242)3	—	0.137 (0.033, 0.241)3	—	—	—	—
Cups/d, trend	0.006 (−0.004, 0.015)	0.005 (−0.009, 0.018)	0.008 (−0.002, 0.018)	0.003 (−0.012, 0.018)	−0.027 (−0.064, 0.010)	−0.017 (−0.056, 0.022)	−0.034 (−0.084, 0.017)
tChol, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.038 (0.011, 0.065)3	0.0007 (−0.027, 0.029)	0.032 (0.007, 0.057)3	0.050 (0.027, 0.072)4	0.007 (−0.019, 0.034)	−0.013 (−0.040, 0.014)	0.084 (0.036, 0.132)4
1	0.049 (0.018, 0.081)4	0.045 (−0.0007, 0.091)	0.010 (−0.022, 0.043)	0.063 (0.029, 0.097)4	0.039 (−0.023, 0.101)	−0.046 (−0.110, 0.018)	0.132 (0.017, 0.247)3
2–3	0.082 (0.057, 0.107)4	0.046 (0.017, 0.075)4	0.026 (0.001, 0.050)3	0.150 (0.123, 0.177)4	0.046 (−0.022, 0.115)	0.054 (−0.018, 0.126)	0.219 (0.125, 0.314)4
4–5	0.042 (0.005, 0.078)3	0.005 (−0.054, 0.063)	−0.043 (−0.080, −0.006)3	0.165 (0.087, 0.243)4	—	—	—
≥6	0.062 (−0.015, 0.138)	—	−0.007 (−0.087, 0.074)	—	—	—	—
Cups/d, trend	0.011 (0.003, 0.018)3	0.005 (−0.005, 0.016)	−0.008 (−0.015, −0.00007)3	0.062 (0.050, 0.073)4	0.029 (−0.0003, 0.058)	0.008 (−0.023, 0.038)	0.110 (0.070, 0.149)4
HDL-C, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.018 (0.009, 0.028)4	0.0004 (−0.009, 0.010)	0.011 (0.002, 0.019)3	0.026 (0.018, 0.034)4	−0.002 (−0.011, 0.008)	−0.005 (−0.014, 0.005)	0.021 (0.004, 0.038)3
1	0.033 (0.022, 0.044)4	0.003 (−0.012, 0.019)	0.018 (0.006, 0.029)4	0.028 (0.016, 0.040)4	0.018 (−0.004, 0.040)	−0.008 (−0.030, 0.015)	0.024 (−0.016, 0.064)
2–3	0.037 (0.028, 0.046)4	0.008 (−0.002, 0.018)	0.015 (0.007, 0.024)4	0.049 (0.039, 0.058)4	0.017 (−0.007, 0.041)	0.001 (−0.024, 0.026)	0.060 (0.027, 0.093)4
4–5	0.043 (0.030, 0.056)4	−0.023 (−0.044, −0.003)3	0.014 (0.002, 0.027)3	0.049 (0.022, 0.077)4	—	—	—
≥6	0.039 (0.012, 0.065)4	—	−0.004 (−0.032, 0.024)	—	—	—	—
Cups/d, trend	0.010 (0.007, 0.012)4	−0.002 (−0.006, 0.002)	0.003 (−0.0002, 0.005)	0.021 (0.017, 0.025)4	0.011 (0.001, 0.021)3	−0.005 (−0.015, 0.006)	0.030 (0.016, 0.044)4
LDL-C, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.020 (−0.0001, 0.041)	0.002 (−0.020, 0.023)	0.019 (−0.001, 0.038)	0.026 (0.009, 0.043)3	0.011 (−0.009, 0.032)	−0.008 (−0.029, 0.013)	0.059 (0.022, 0.096)4
1	0.020 (−0.004, 0.044)	0.038 (0.003, 0.074)3	0.001 (−0.024, 0.026)	0.032 (0.006, 0.058)3	0.027 (−0.020, 0.075)	−0.023 (−0.072, 0.027)	0.110 (0.021, 0.199)3
2–3	0.045 (0.026, 0.064)4	0.037 (0.014, 0.059)4	0.014 (−0.005, 0.033)	0.094 (0.073, 0.114)4	0.034 (−0.019, 0.086)	0.047 (−0.009, 0.102)	0.161 (0.088, 0.234)4
4–5	0.014 (−0.014, 0.042)	0.025 (−0.020, 0.070)	−0.038 (−0.067, −0.010)3	0.120 (0.060, 0.180)4	—	—	—
≥6	0.035 (−0.024, 0.094)	—	0.004 (−0.058, 0.066)	—	—	—	—
Cups/d, trend	0.004 (−0.002, 0.010)	0.008 (−0.0001, 0.016)	−0.006 (−0.012, −0.0005)3	0.039 (0.030, 0.048)4	0.022 (−0.0004, 0.045)	0.012 (−0.012, 0.035)	0.078 (0.047, 0.109)4
Fasting TGs, log mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	−0.017 (−0.054, 0.020)	0.007 (−0.033, 0.046)	−0.019 (−0.053, 0.016)	−0.034 (−0.065, −0.003)3	0.008 (−0.030, 0.046)	−0.031 (−0.069, 0.007)	−0.050 (−0.116, 0.016)
1	−0.008 (−0.050, 0.035)	0.003 (−0.070, 0.076)	−0.018 (−0.062, 0.026)	−0.036 (−0.082, 0.010)	−0.053 (−0.137, 0.031)	0.082 (−0.005, 0.168)	0.052 (−0.100, 0.204)
2–3	−0.025 (−0.059, 0.008)	−0.043 (−0.085, −0.002)3	−0.047 (−0.080, −0.014)3	−0.055 (−0.092, −0.018)3	−0.054 (−0.148, 0.039)	−0.012 (−0.115, 0.091)	0.062 (−0.061, 0.186)
4–5	−0.067 (−0.116, −0.018)3	−0.047 (−0.129, 0.034)	−0.060 (−0.111, −0.009)3	−0.018 (−0.116, 0.080)	—	—	—
≥6	−0.078 (−0.176, 0.020)	—	−0.075 (−0.176, 0.026)	—	—	—	—
Cups/d, trend	−0.014 (−0.024, −0.004)3	−0.014 (−0.028, 0.001)	−0.014 (−0.024, −0.003)3	−0.019 (−0.034, −0.003)3	−0.031 (−0.073, 0.011)	0.001 (−0.040, 0.043)	0.024 (−0.022, 0.071)
CRP, log mg/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	−0.058 (−0.082, −0.034)4	−0.023 (−0.048, 0.002)	−0.012 (−0.034, 0.011)	−0.069 (−0.089, −0.049)4	−0.037 (−0.061, −0.014)4	0.002 (−0.023, 0.026)	−0.079 (−0.122, −0.036)4
1	−0.073 (−0.102, −0.045)4	−0.058 (−0.099, −0.016)3	−0.030 (−0.059, −0.0005)3	−0.104 (−0.134, −0.074)4	−0.071 (−0.127, −0.015)3	−0.043 (−0.100, 0.014)	−0.066 (−0.170, 0.038)
2–3	−0.124 (−0.146, −0.102)4	−0.088 (−0.113, −0.062)4	−0.067 (−0.089, −0.044)4	−0.167 (−0.192, −0.143)4	−0.067 (−0.128, −0.005)3	−0.011 (−0.076, 0.054)	−0.121 (−0.206, −0.036)3
4–5	−0.178 (−0.211, −0.145)4	−0.155 (−0.207, −0.102)4	−0.121 (−0.154, −0.087)4	−0.256 (−0.326, −0.186)4	—	—	—
≥6	−0.261 (−0.330, −0.192)4	—	−0.178 (−0.251, −0.106)4	—	—	—	—
Cups/d, trend	−0.047 (−0.054, −0.040)4	−0.040 (−0.050, −0.031)4	−0.035 (−0.042, −0.028)4	−0.084 (−0.094, −0.073)4	−0.061 (−0.088, −0.035)4	−0.020 (−0.047, 0.007)	−0.099 (−0.135, −0.063)4
Fasting glucose, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.001 (−0.048, 0.050)	0.008 (−0.045, 0.062)	0.003 (−0.044, 0.049)	0.021 (−0.020, 0.063)	−0.005 (−0.057, 0.046)	−0.050 (−0.101, 0.001)	−0.035 (−0.124, 0.053)
1	0.003 (−0.055, 0.061)	−0.028 (−0.127, 0.072)	−0.029 (−0.088, 0.031)	0.012 (−0.050, 0.075)	−0.003 (−0.115, 0.108)	0.042 (−0.079, 0.162)	0.159 (−0.042, 0.360)
2–3	0.030 (−0.016, 0.075)	0.022 (−0.034, 0.078)	0.015 (−0.029, 0.060)	0.051 (0.001, 0.101)3	0.066 (−0.061, 0.193)	−0.094 (−0.230, 0.042)	−0.043 (−0.206, 0.119)
4–5	0.029 (−0.038, 0.096)	0.029 (−0.081, 0.138)	0.001 (−0.068, 0.070)	0.029 (−0.106, 0.164)	—	—	—
≥6	−0.015 (−0.145, 0.116)	—	−0.052 (−0.192, 0.087)	—	—	—	—
Cups/d, trend	0.004 (−0.010, 0.018)	0.006 (−0.014, 0.026)	0.002 (−0.012, 0.016)	0.023 (0.002, 0.044)3	0.012 (−0.045, 0.068)	−0.035 (−0.091, 0.021)	−0.001 (−0.063, 0.062)
HbA1c, mmol/mol
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	−0.032 (−0.128, 0.064)	0.026 (−0.074, 0.127)	0.038 (−0.053, 0.129)	−0.086 (−0.166, −0.006)3	0.040 (−0.055, 0.136)	0.017 (−0.080, 0.115)	−0.106 (−0.279, 0.068)
1	−0.099 (−0.212, 0.015)	0.029 (−0.137, 0.194)	−0.020 (−0.137, 0.097)	−0.043 (−0.165, 0.078)	−0.071 (−0.294, 0.151)	−0.102 (−0.330, 0.126)	0.157 (−0.257, 0.571)
2–3	−0.088 (−0.178, 0.001)	−0.018 (−0.122, 0.085)	−0.043 (−0.132, 0.045)	−0.088 (−0.185, 0.009)	0.064 (−0.182, 0.310)	0.037 (−0.223, 0.298)	−0.316 (−0.658, 0.025)
4–5	−0.160 (−0.291, −0.028)3	−0.121 (−0.330, 0.088)	−0.084 (−0.215, 0.048)	−0.312 (−0.593, −0.031)3	—	—	—
≥6	−0.106 (−0.382, 0.171)	—	−0.078 (−0.369, 0.213)	—	—	—	—
Cups/d, trend	−0.041 (−0.069, −0.014)3	−0.021 (−0.059, 0.018)	−0.029 (−0.056, −0.001)3	−0.056 (−0.098, −0.014)3	−0.015 (−0.120, 0.090)	−0.012 (−0.121, 0.097)	−0.133 (−0.276, 0.010)

¹Values are β (95% CI). Results from linear regression models adjusted for age; race; date of blood draw; fasting time; assessment center; Townsend deprivation index; education; income; employment status; home ownership; smoking; physical activity; waist-to-hip ratio; BMI; oral contraceptive use; postmenopausal hormone use; self-rated health; aspirin use; cholesterol-lowering medication use; antihypertension medication use; history of diabetes; history of cardiovascular disease; total energy intake; intakes of carbohydrates, fat, saturated fat, protein, fizzy drinks, milk, water, alcohol, fish, red meat, fruits, and vegetables; recent caffeine drinking; tea; and 1) Decaf (for Regular analysis), 2) Regular (for Decaf analysis), or 3) all other types of coffee (for separate analysis of Filtered, Instant, Latte, Cappuccino, and Espresso). CRP, C-reactive protein; HbA1c, glycated hemoglobin; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; Lp(a), lipoprotein (a); tChol, total cholesterol; TG, triglyceride.

²Regular: caffeine-containing filtered, instant, latte, cappuccino, and espresso coffee; Decaf: decaffeinated filtered, instant, latte, cappuccino, and espresso coffee. Maximum coffee intake categories for Decaf, Filtered, Latte, Espresso, and Cappuccino are ≥4, ≥4, ≥2, ≥2, and ≥2, respectively.

³0.0025 ≤ P < 0.05.

⁴ P < 0.0025.

TABLE 3.

Associations between coffee consumption (24-h diet recalls) and cardiometabolic biomarkers among men¹

Cups/d	Regular	Decaf	Instant	Filtered/americano/ cafetiere2	Cappuccino	Latte	Espresso
ApoA-1, g/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.014 (0.007, 0.021)4	−0.003 (−0.011, 0.005)	0.005 (−0.001, 0.011)	0.012 (0.006, 0.017)4	0.001 (−0.006, 0.008)	−0.014 (−0.021, −0.006)4	0.009 (−0.001, 0.018)
1	0.013 (0.005, 0.021)4	−0.0006 (−0.013, 0.012)	0.008 (0.0004, 0.016)3	0.013 (0.005, 0.021)4	0.009 (−0.008, 0.025)	−0.0004 (−0.017, 0.016)	0.012 (−0.008, 0.033)
2–3	0.019 (0.013, 0.025)4	−0.002 (−0.009, 0.006)	0.00006 (−0.006, 0.006)	0.022 (0.016, 0.029)4	0.008 (−0.010, 0.026)	−0.017 (−0.034, −0.0005)3	0.002 (−0.014, 0.017)
4–5	0.016 (0.008, 0.024)4	−0.003 (−0.019, 0.013)	0.001 (−0.008, 0.009)	0.025 (0.010, 0.041)4	—	—	—
≥6	0.021 (0.005, 0.037)3	—	−0.012 (−0.031, 0.006)	—	—	—	—
Cups/d, trend	0.002 (0.0007, 0.004)3	−0.003 (−0.005, 0.0003)	−0.001 (−0.003, 0.001)	0.010 (0.007, 0.012)4	0.003 (−0.004, 0.010)	−0.009 (−0.016, −0.001)3	0.006 (−0.00002, 0.012)
ApoB, g/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.007 (−0.0003, 0.014)	−0.002 (−0.010, 0.005)	−0.006 (−0.012, 0.001)	0.005 (−0.001, 0.011)	0.001 (−0.006, 0.009)	−0.007 (−0.015, −0.0001)3	0.017 (0.007, 0.026)4
1	0.009 (0.0006, 0.017)3	−0.001 (−0.013, 0.011)	−0.010 (−0.018, −0.002)3	0.007 (−0.001, 0.015)	0.010 (−0.007, 0.026)	0.006 (−0.011, 0.022)	0.007 (−0.013, 0.028)
2–3	0.010 (0.004, 0.016)4	0.003 (−0.005, 0.010)	−0.005 (−0.011, 0.001)	0.013 (0.006, 0.019)4	0.023 (0.005, 0.041)3	0.004 (−0.013, 0.021)	0.025 (0.010, 0.040)4
4–5	0.016 (0.007, 0.024)4	0.010 (−0.007, 0.025)	0.002 (−0.007, 0.011)	0.008 (−0.007, 0.024)	—	—	—
≥6	0.005 (−0.012, 0.021)	—	−0.006 (−0.025, 0.012)	—	—	—	—
Cups/d, trend	0.002 (0.0004, 0.004)3	0.001 (−0.002, 0.004)	−0.0001 (−0.002, 0.002)	0.006 (0.003, 0.008)4	0.008 (0.001, 0.016)3	−0.001 (−0.008, 0.007)	0.013 (0.007, 0.019)4
Lp(a), log nmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	−0.006 (−0.048, 0.037)	−0.014 (−0.060, 0.033)	0.006 (−0.032, 0.043)	0.004 (−0.030, 0.037)	0.013 (−0.032, 0.057)	−0.005 (−0.050, 0.039)	0.017 (−0.041, 0.075)
1	−0.015 (−0.064, 0.035)	0.004 (−0.070, 0.077)	0.001 (−0.046, 0.049)	−0.004 (−0.053, 0.044)	0.032 (−0.068, 0.132)	0.036 (−0.065, 0.136)	0.070 (−0.055, 0.195)
2–3	−0.007 (−0.045, 0.032)	0.022 (−0.024, 0.068)	−0.008 (−0.044, 0.027)	0.006 (−0.032, 0.044)	0.066 (−0.042, 0.173)	−0.065 (−0.168, 0.038)	0.074 (−0.016, 0.164)
4–5	0.039 (−0.012, 0.090)	0.053 (−0.044, 0.150)	0.057 (0.005, 0.110)3	−0.031 (−0.124, 0.061)	—	—	—
≥6	−0.089 (−0.187, 0.009)	—	−0.117 (−0.230, −0.005)3	—	—	—	—
Cups/d, trend	0.002 (−0.008, 0.013)	0.008 (−0.009, 0.025)	0.003 (−0.008, 0.013)	−0.002 (−0.018, 0.013)	0.018 (−0.026, 0.062)	−0.015 (−0.061, 0.031)	0.038 (0.001, 0.074)3
tChol, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.058 (0.026, 0.090)4	−0.025 (−0.060, 0.010)	−0.008 (−0.036, 0.020)	0.048 (0.023, 0.073)4	0.001 (−0.032, 0.034)	−0.037 (−0.070, −0.004)3	0.078 (0.035, 0.121)4
1	0.079 (0.042, 0.116)4	−0.018 (−0.073, 0.038)	−0.020 (−0.056, 0.015)	0.072 (0.036, 0.108)4	0.071 (−0.003, 0.146)	0.056 (−0.019, 0.130)	0.068 (−0.025, 0.162)
2–3	0.101 (0.073, 0.130)4	0.031 (−0.004, 0.065)3	−0.004 (−0.031, 0.022)	0.119 (0.091, 0.148)4	0.118 (0.037, 0.200)3	0.013 (−0.064, 0.091)	0.146 (0.077, 0.215)4
4–5	0.125 (0.087, 0.163)4	0.056 (−0.017, 0.129)	0.032 (−0.007, 0.071)	0.138 (0.069, 0.207)4	—	—	—
≥6	0.099 (0.026, 0.171)3	—	0.012 (−0.072, 0.096)	—	—	—	—
Cups/d, trend	0.022 (0.014, 0.030)4	0.008 (−0.005, 0.021)	0.005 (−0.003, 0.013)	0.054 (0.042, 0.066)4	0.042 (0.009, 0.075)3	0.004 (−0.030, 0.038)	0.078 (0.050, 0.106)4
HDL-C, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.019 (0.009, 0.028)4	−0.001 (−0.011, 0.009)	0.007 (−0.001, 0.016)	0.017 (0.009, 0.024)4	−0.0001 (−0.010, 0.010)	−0.014 (−0.023, −0.004)3	0.013 (0.0002, 0.025)3
1	0.017 (0.006, 0.028)3	0.004 (−0.012, 0.021)	0.011 (0.001, 0.021)3	0.018 (0.008, 0.029)4	0.017 (−0.005, 0.039)	0.005 (−0.017, 0.026)	0.014 (−0.013, 0.041)
2–3	0.028 (0.020, 0.036)4	0.003 (−0.007, 0.013)	0.004 (−0.003, 0.012)	0.032 (0.024, 0.040)4	0.019 (−0.005, 0.042)	−0.018 (−0.040, 0.004)	0.002 (−0.018, 0.022)
4–5	0.027 (0.016, 0.038)4	0.004 (−0.017, 0.025)	0.010 (−0.001, 0.022)	0.039 (0.019, 0.059)4	—	—	—
≥6	0.033 (0.012, 0.054)4	—	0.0004 (−0.024, 0.024)	—	—	—	—
Cups/d, trend	0.005 (0.003, 0.007)4	−0.0006 (−0.004, 0.003)	0.001 (−0.001, 0.004)	0.014 (0.011, 0.017)4	0.007 (−0.003, 0.016)	−0.008 (−0.018, 0.001)	0.008 (0.000, 0.016)
LDL-C, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.035 (0.011, 0.059)3	−0.009 (−0.035, 0.018)	−0.011 (−0.033, 0.010)	0.030 (0.011, 0.049)4	0.011 (−0.014, 0.036)	−0.023 (−0.048, 0.003)	0.059 (0.027, 0.092)4
1	0.048 (0.020, 0.076)4	−0.009 (−0.051, 0.033)	−0.023 (−0.050, 0.004)	0.047 (0.019, 0.074)4	0.062 (0.005, 0.119)3	0.053 (−0.004, 0.110)	0.047 (−0.024, 0.118)
2–3	0.067 (0.045, 0.089)4	0.028 (0.002, 0.055)3	−0.003 (−0.023, 0.018)	0.082 (0.061, 0.104)4	0.096 (0.034, 0.158)3	0.038 (−0.021, 0.097)	0.110 (0.058, 0.163)4
4–5	0.097 (0.068, 0.126)4	0.059 (0.003, 0.115)3	0.033 (0.003, 0.062)3	0.098 (0.045, 0.150)4	—	—	—
≥6	0.073 (0.017, 0.128)3	—	0.019 (−0.045, 0.083)	—	—	—	—
Cups/d, trend	0.018 (0.012, 0.024)4	0.011 (0.001, 0.021)3	0.006 (0.0002, 0.013)3	0.038 (0.029, 0.047)4	0.042 (0.017, 0.067)4	0.015 (−0.011, 0.041)	0.058 (0.037, 0.080)4
Fasting TGs, log mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.004 (−0.045, 0.053)	0.017 (−0.043, 0.077)	−0.009 (−0.053, 0.035)	−0.028 (−0.068, 0.012)	0.035 (−0.020, 0.090)	0.036 (−0.019, 0.091)	−0.033 (−0.107, 0.040)
1	−0.020 (−0.078, 0.037)	−0.028 (−0.118, 0.062)	0.023 (−0.035, 0.081)	−0.045 (−0.102, 0.013)	0.008 (−0.111, 0.127)	0.066 (−0.060, 0.191)	−0.115 (−0.264, 0.035)
2–3	−0.043 (−0.088, 0.002)	0.010 (−0.047, 0.066)	−0.044 (−0.086, −0.003)3	−0.046 (−0.092, 0.0004)	−0.026 (−0.157, 0.105)	−0.037 (−0.153, 0.080)	0.127 (0.025, 0.229)3
4–5	−0.069 (−0.129, −0.009)3	−0.043 (−0.156, 0.071)	−0.107 (−0.168, −0.046)4	−0.140 (−0.249, −0.032)3	—	—	—
≥6	−0.083 (−0.193, 0.026)	—	−0.039 (−0.160, 0.082)	—	—	—	—
Cups/d, trend	−0.019 (−0.031, −0.007)4	−0.004 (−0.025, 0.016)	−0.018 (−0.031, −0.006)3	−0.025 (−0.044, −0.007)3	0.006 (−0.044, 0.056)	0.011 (−0.043, 0.065)	0.028 (−0.015, 0.072)
CRP, log mg/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	−0.008 (−0.037, 0.021)	−0.043 (−0.074, −0.011)3	−0.005 (−0.031, 0.021)	−0.013 (−0.036, 0.010)	−0.007 (−0.037, 0.023)	0.019 (−0.011, 0.049)	0.002 (−0.037, 0.042)
1	−0.025 (−0.058, 0.009)	−0.082 (−0.133, −0.032)4	−0.026 (−0.058, 0.006)	−0.042 (−0.075, −0.009)3	−0.022 (−0.090, 0.047)	0.010 (−0.059, 0.078)	−0.016 (−0.101, 0.070)
2–3	−0.067 (−0.093, −0.041)4	−0.042 (−0.074, −0.011)3	−0.032 (−0.056, −0.008)3	−0.089 (−0.115, −0.063)4	−0.011 (−0.085, 0.063)	−0.040 (−0.111, 0.030)	−0.046 (−0.108, 0.017)
4–5	−0.077 (−0.112, −0.042)4	0.016 (−0.050, 0.083)	−0.044 (−0.080, −0.009)3	−0.107 (−0.170, −0.044)4	—	—	—
≥6	−0.091 (−0.157, −0.025)3	—	−0.081 (−0.158, −0.005)3	—	—	—	—
Cups/d, trend	−0.019 (−0.026, −0.012)4	−0.011 (−0.023, 0.0002)	−0.012 (−0.019, −0.004)4	−0.037 (−0.048, −0.027)4	−0.012 (−0.042, 0.018)	−0.009 (−0.040, 0.023)	−0.019 (−0.045, 0.006)
Fasting glucose, mmol/L
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	0.003 (−0.065, 0.071)	−0.014 (−0.097, 0.069)	0.019 (−0.042, 0.080)	−0.005 (−0.061, 0.051)	−0.048 (−0.124, 0.027)	−0.040 (−0.115, 0.036)	0.014 (−0.085, 0.114)
1	−0.003 (−0.082, 0.076)	−0.019 (−0.143, 0.106)	0.020 (−0.059, 0.100)	0.023 (−0.056, 0.101)	0.0004 (−0.160, 0.161)	0.014 (−0.154, 0.182)	−0.106 (−0.307, 0.096)
2–3	0.004 (−0.058, 0.067)	0.027 (−0.051, 0.106)	0.034 (−0.024, 0.091)	0.013 (−0.052, 0.077)	0.030 (−0.150, 0.210)	−0.018 (−0.177, 0.140)	−0.011 (−0.152, 0.129)
4–5	0.054 (−0.029, 0.136)	−0.109 (−0.266, 0.047)	0.073 (−0.010, 0.157)	0.015 (−0.135, 0.164)	—	—	—
≥6	−0.050 (−0.201, 0.102)	—	−0.064 (−0.232, 0.103)	—	—	—	—
Cups/d, trend	0.011 (−0.006, 0.027)	−0.004 (−0.031, 0.024)	0.010 (−0.007, 0.028)	0.008 (−0.017, 0.034)	0.020 (−0.047, 0.088)	−0.009 (−0.083, 0.064)	−0.015 (−0.075, 0.045)
HbA1c, mmol/mol
0	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.	Ref.
<1	−0.028 (−0.166, 0.110)	−0.182 (−0.332, −0.031)3	−0.021 (−0.144, 0.102)	−0.051 (−0.159, 0.057)	0.046 (−0.098, 0.190)	−0.053 (−0.197, 0.090)	−0.071 (−0.258, 0.116)
1	0.080 (−0.080, 0.239)	−0.301 (−0.540, −0.062)3	−0.009 (−0.162, 0.145)	−0.073 (−0.229, 0.084)	−0.070 (−0.395, 0.256)	−0.048 (−0.372, 0.275)	0.259 (−0.148, 0.666)
2–3	−0.084 (−0.207, 0.040)	−0.124 (−0.273, 0.026)	−0.091 (−0.205, 0.024)	−0.109 (−0.232, 0.014)	0.059 (−0.292, 0.410)	−0.287 (−0.623, 0.049)	−0.282 (−0.579, 0.014)
4–5	0.033 (−0.132, 0.197)	−0.350 (−0.668, −0.032)3	−0.056 (−0.224, 0.112)	−0.068 (−0.367, 0.232)	—	—	—
≥6	0.072 (−0.243, 0.387)	—	0.037 (−0.326, 0.400)	—	—	—	—
Cups/d, trend	−0.004 (−0.038, 0.029)	−0.085 (−0.140, −0.029)4	−0.010 (−0.045, 0.026)	−0.028 (−0.079, 0.022)	0.007 (−0.136, 0.151)	−0.122 (−0.269, 0.026)	−0.079 (−0.200, 0.042)

¹Values are β (95% CI). Results from linear regression models adjusted for age; race; date of blood draw; fasting time; assessment center; Townsend deprivation index; education; income; employment status; home ownership; smoking; physical activity; waist-to-hip ratio; BMI; self-rated health; aspirin use; cholesterol-lowering medication use; antihypertension medication use; history of diabetes; history of cardiovascular disease; total energy intake; intakes of carbohydrates, fat, saturated fat, protein, fizzy drinks, milk, water, alcohol, fish, red meat, fruits, and vegetables; recent caffeine drinking; tea; and 1) Decaf (for Regular analysis), 2) Regular (for Decaf analysis), or 3) all other types of coffee (for separate analysis of Filtered, Instant, Latte, Cappuccino, and Espresso). CRP, C-reactive protein; HbA1c, glycated hemoglobin; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; Lp(a), lipoprotein (a); tChol, total cholesterol; TG, triglyceride.

²Regular: caffeine-containing filtered, instant, latte, cappuccino, and espresso coffee; Decaf: decaffeinated filtered, instant, latte, cappuccino, and espresso coffee. Maximum coffee intake categories for Decaf, Filtered, Latte, Espresso, and Cappuccino are ≥4, ≥4, ≥2, ≥2, and ≥2, respectively.

³0.0025 ≤ P < 0.05.

⁴ P < 0.0025.

CRP concentrations were lower with each additional cup of coffee and similar associations with CRP were observed for different types of coffee. Among women, any amount of coffee intake was associated with lower HbA1c than no coffee intake. Among men, consumption of ≥4 cups/d of coffee was associated with higher HbA1c than no coffee, but these results were not supported by results from the 24-h diet-recall analysis which tended toward the results observed for women. Coffee consumption was not substantially associated with Lp(a) or fasting glucose concentrations. Similar results were observed when excluding participants reporting diabetes, CVD, and use of related medication, with the exception of HbA1c among men: no significant direct association was observed between total coffee intake and HbA1c.

Coffee–biomarker associations stratified by smoking status are presented in Supplemental Table 9 and support several coffee × smoking interactions. Generally, the favorable associations between total coffee and biomarkers (i.e., higher HDL-C and apoA-1 and lower TGs, CRP, and HbA1c) were stronger in or restricted to nonsmokers, whereas nonfavorable associations (i.e., higher apoB, LDL cholesterol, and HbA1c) were stronger in or restricted to smokers. Clear crossover interactions were observed for HDL cholesterol and apoA-1: coffee intake was significantly associated with higher HDL cholesterol and apoA-1 in nonsmokers and significantly associated with lower HDL cholesterol and apoA-1 among smokers.

Select coffee–biomarker associations were significantly modified by measures of adiposity, age, and race (see the Supplemental Notes). For example, coffee intake was associated with lower HbA1c concentrations only in overweight women (BMI ≥25) (P < 0.0001 for interaction), whereas the direct association between coffee intake and HbA1c among men (Table 1) was most apparent among leaner men (BMI <25) (P = 0.0007 for interaction).

Supplemental Table 10 presents associations between total coffee intake and the prevalence of abnormal biomarker concentrations for multivariable models. For both men and women, coffee consumption was associated with a significantly higher likelihood of abnormal apoB, tChol, and LDL cholesterol and a lower likelihood of abnormal HDL cholesterol and TGs, with a 1%–4% higher or lower OR per cup (all P < 0.0003). For abnormal LDL cholesterol (≥3.00 mmol/L) the OR was 1.09 (95% CI: 1.05, 1.12) for 1 cup, 1.17 (95% CI: 1.14, 1.21) for 2–3 cups, 1.18 (95% CI: 1.14, 1.23) for 4–5 cups, 1.12 (95% CI: 1.05, 1.19) for 6–7 cups, and 1.23 (95% CI: 1.13, 1.35) for ≥8 cups/d as compared with no coffee consumption in women. In men, these ORs were 1.11 (95% CI: 1.07, 1.15), 1.13 (95% CI: 1.09, 1.17), 1.19 (95% CI: 1.14, 1.24), 1.26 (95% CI: 1.18, 1.34), and 1.26 (95% CI: 1.16, 1.36), respectively.

For women, the OR for abnormal CRP (>3 mg/L) was 0.89 (95% CI: 0.87, 0.92) for 1 cup, 0.84 (95% CI: 0.81, 0.86) for 2–3 cups, 0.79 (95% CI: 0.76, 0.82) for 4–5 cups, 0.80 (95% CI: 0.75, 0.85) for 6–7 cups, and 0.70 (95% CI: 0.64, 0.77) for ≥8 cups/d as compared with no consumption. In men, the OR for consuming 1 cup/d compared with no consumption was 0.90 (95% CI: 0.87, 0.94), but there was no trend for lower odds with higher consumption. Similarly, coffee consumption was associated with substantially lower ORs of abnormal HbA1c concentrations in women, but not in men. These risk estimates do not account for type of coffee brew or the interactions with smoking or adiposity identified above, which would be expected to strengthen or weaken estimates among subgroups defined by these factors.

Habitual tea intake and cardiometabolic biomarkers

Among both men and women, habitual total tea (including both black and green tea) consumption as assessed by FFQ was inversely associated with tChol, LDL cholesterol, apoB, and fasting TGs and directly associated with HDL cholesterol (Table 4). Total, black, and green tea were significantly inversely associated with CRP concentrations among women, but only a nominally significant inverse association with black tea was observed among men (P < 0.05) (Table 5). Higher tea consumption was associated with higher HbA1c following a dose-response pattern among both men and women, and this association appeared to be restricted to black tea. Tea consumption was not substantially associated with Lp(a) or fasting glucose concentrations. Overall, similar results were observed after excluding participants reporting diabetes, CVD, and use of related medications.

TABLE 4.

Associations between tea consumption and cardiometabolic biomarkers¹

	Women			Men
Cups/d	n	Model 12, β (95% CI)	Model 23, β (95% CI)	n	Model 12, β (95% CI)	Model 23, β (95% CI)
ApoA-1, g/L
0	32,064	Ref.	Ref.	25,162	Ref.	Ref.
<1	6211	0.02 (0.01, 0.03)5	0.002 (−0.004, 0.009)	5906	0.02 (0.02, 0.03)5	0.006 (0.0007, 0.01)4
1	17,195	0.03 (0.03, 0.04)5	0.006 (0.001, 0.01)4	16,004	0.02 (0.02, 0.03)5	0.003 (−0.001, 0.007)
2–3	61,981	0.03 (0.03, 0.03)5	0.005 (0.001, 0.008)4	53,092	0.02 (0.02, 0.03)5	0.004 (0.0008, 0.007)4
4–5	54,466	0.02 (0.02, 0.02)5	0.004 (−0.00006, 0.007)4	44,637	0.02 (0.01, 0.02)5	0.002 (−0.001, 0.005)
6–7	25,377	0.01 (0.006, 0.01)5	0.006 (0.002, 0.01)4	20,475	0.006 (0.001, 0.010)	0.0002 (−0.004, 0.004)
≥8	13,521	−0.005 (−0.01, 0.0005)	0.006 (0.001, 0.01)4	13,793	−0.003 (−0.007, 0.002)	0.004 (−0.0008, 0.008)
Cups/d, trend		−0.001 (−0.002, −0.0008)5	0.0002 (−0.0002, 0.0007)		−0.001 (−0.002, −0.001)5	−0.0001 (−0.0005, 0.0003)
ApoB, g/L
0	35,415	Ref.	Ref.	27,084	Ref.	Ref.
<1	6876	−0.005 (−0.01, 0.001)	−0.003 (−0.008, 0.003)	6368	0.005 (−0.002, 0.01)	0.001 (−0.005, 0.007)
1	19,076	−0.007 (−0.01, −0.003)5	−0.003 (−0.007, 0.0008)	17,234	−0.002 (−0.007, 0.002)	−0.005 (−0.009, −0.0004)4
2–3	68,630	−0.01 (−0.02, −0.01)5	−0.009 (−0.01, −0.006)5	57,319	−0.006 (−0.009, −0.002)5	−0.008 (−0.01, −0.004)5
4–5	60,370	−0.02 (−0.02, −0.02)5	−0.01 (−0.02, −0.01)5	48,211	−0.009 (−0.01, −0.006)5	−0.01 (−0.01, −0.008)5
6–7	28,016	−0.03 (−0.03, −0.02)5	−0.02 (−0.03, −0.02)5	22,139	−0.01 (−0.02, −0.009)5	−0.02 (−0.02, −0.01)5
≥8	14,949	−0.02 (−0.03, −0.02)5	−0.02 (−0.03, −0.02)5	14,914	−0.02 (−0.02, −0.01)5	−0.02 (−0.02, −0.01)5
Cups/d, trend		−0.003 (−0.003, −0.002)5	−0.003 (−0.003, −0.003)5		−0.002 (−0.002, −0.001)5	−0.002 (−0.002, −0.001)5
Lp(a), log nmol/L
0	28,390	Ref.	Ref.	21,733	Ref.	Ref.
<1	5538	−0.01 (−0.04, 0.02)	−0.007 (−0.04, 0.03)	5107	−0.005 (−0.04, 0.03)	−0.006 (−0.04, 0.03)
1	15,407	−0.0009 (−0.02, 0.02)	0.0003 (−0.02, 0.02)	13,867	0.02 (−0.007, 0.04)	0.01 (−0.01, 0.04)
2–3	55,391	0.002 (−0.01, 0.02)	0.003 (−0.01, 0.02)	46,106	0.02 (0.002, 0.04)4	0.01 (−0.004, 0.03)
4–5	48,581	0.007 (−0.009, 0.02)	0.006 (−0.01, 0.02)	38,624	0.03 (0.01, 0.05)5	0.02 (0.005, 0.04)4
6–7	22,602	0.009 (−0.01, 0.03)	0.006 (−0.01, 0.03)	17,826	0.04 (0.02, 0.06)5	0.03 (0.009, 0.06)4
≥8	12,027	0.02 (−0.004, 0.04)	0.02 (−0.008, 0.04)	11,900	0.03 (0.008, 0.06)4	0.03 (−0.001, 0.05)
Cups/d, trend		0.002 (0.0005, 0.004)4	0.002 (−0.00008, 0.004)		0.004 (0.002, 0.006)5	0.003 (0.0005, 0.005)4
tChol, mmol/L
0	35,513	Ref.	Ref.	27,313	Ref.	Ref.
<1	6897	0.02 (−0.01, 0.04)	−0.007 (−0.03, 0.02)	6417	0.06 (0.03, 0.09)5	0.02 (−0.002, 0.05)
1	19,127	0.01 (−0.008, 0.03)	−0.009 (−0.03, 0.009)	17,351	0.02 (−0.00002, 0.04)4	−0.02 (−0.03, 0.003)
2–3	68,821	−0.00005 (−0.01, 0.01)	−0.02 (−0.03, −0.005)4	57,678	0.02 (0.006, 0.04)4	−0.02 (−0.03, −0.001)4
4–5	60,548	−0.03 (−0.04, −0.02)5	−0.03 (−0.05, −0.02)5	48,525	−0.002 (−0.02, 0.01)	−0.03 (−0.04, −0.01)5
6–7	28,100	−0.06 (−0.08, −0.05)5	−0.06 (−0.07, −0.04)5	22,272	−0.04 (−0.06, −0.02)5	−0.05 (−0.07, −0.03)5
≥8	14,987	−0.07 (−0.09, −0.05)5	−0.05 (−0.07, −0.03)5	15,017	−0.06 (−0.08, −0.04)5	−0.05 (−0.07, −0.03)5
Cups/d, trend		−0.009 (−0.01, −0.008)5	−0.007 (−0.009, −0.006)5		−0.008 (−0.01, −0.007)5	−0.006 (−0.008, −0.005)5
HDL-C, mmol/L
0	32,325	Ref.	Ref.	25,186	Ref.	Ref.
<1	6266	0.03 (0.02, 0.04)5	0.003 (−0.006, 0.01)	5910	0.03 (0.02, 0.04)5	0.007 (−0.0001, 0.01)4
1	17,332	0.04 (0.04, 0.05)5	0.006 (−0.0005, 0.01)	16,020	0.03 (0.02, 0.04)5	0.003 (−0.002, 0.008)
2–3	62,495	0.05 (0.04, 0.05)5	0.008 (0.003, 0.01)5	53,128	0.04 (0.03, 0.04)5	0.006 (0.001, 0.01)4
4–5	54,865	0.04 (0.03, 0.04)5	0.01 (0.007, 0.02)5	44,661	0.03 (0.02, 0.03)5	0.007 (0.003, 0.01)5
6–7	25,549	0.03 (0.02, 0.04)5	0.02 (0.02, 0.03)5	20,485	0.02 (0.01, 0.02)5	0.006 (0.001, 0.01)4
≥8	13,605	0.01 (0.004, 0.02)5	0.03 (0.02, 0.03)5	13,793	0.009 (0.003, 0.02)4	0.01 (0.008, 0.02)5
Cups/d, trend		0.0003 (−0.0003, 0.0009)	0.003 (0.002, 0.003)5		−0.0005 (−0.001, 0.00004)	0.0009 (0.0005, 0.001)5
LDL-C, mmol/L
0	35,463	Ref.	Ref.	27,247	Ref.	Ref.
<1	6892	−0.009 (−0.03, 0.01)	−0.01 (−0.03, 0.008)	6403	0.04 (0.02, 0.06)5	0.02 (−0.003, 0.04)
1	19,100	−0.02 (−0.03, −0.003)4	−0.01 (−0.03, 0.0002)4	17,326	0.005 (−0.01, 0.02)	−0.01 (−0.03, 0.002)
2–3	68,722	−0.03 (−0.04, −0.02)5	−0.02 (−0.03, −0.01)5	57,565	0.007 (−0.006, 0.02)	−0.01 (−0.02, 0.0001)4
4–5	60,458	−0.04 (−0.05, −0.03)5	−0.03 (−0.05, −0.02)5	48,415	−0.005 (−0.02, 0.008)	−0.02 (−0.03, −0.006)5
6–7	28,042	−0.07 (−0.08, −0.05)5	−0.06 (−0.07, −0.05)5	22,229	−0.02 (−0.04, −0.007)4	−0.03 (−0.05, −0.02)5
≥8	14,962	−0.06 (−0.07, −0.04)5	−0.05 (−0.07, −0.04)5	14,977	−0.04 (−0.05, −0.02)5	−0.03 (−0.04, −0.01)5
Cups/d, trend		−0.007 (−0.008, −0.006)5	−0.007 (−0.008, −0.006)5		−0.005 (−0.006, −0.003)5	−0.004 (−0.005, −0.003)5
Fasting TGs, log mmol/L
0	4347	Ref.	Ref.	4235	Ref.	Ref.
<1	791	−0.03 (−0.07, −0.001)	−0.006 (−0.04, 0.02)	909	−0.05 (−0.09, −0.02)4	−0.03 (−0.06, 0.007)
1	2100	−0.03 (−0.05, −0.004)4	−0.005 (−0.03, 0.02)	2442	−0.009 (−0.04, 0.02)	0.006 (−0.02, 0.03)
2–3	6701	−0.04 (−0.06, −0.02)5	−0.01 (−0.03, 0.003)	6898	−0.02 (−0.04, −0.003)4	−0.0002 (−0.02, 0.02)
4–5	5207	−0.05 (−0.07, −0.03)5	−0.02 (−0.04, −0.003)4	5403	−0.03 (−0.05, −0.01)5	−0.02 (−0.04, 0.004)
6–7	2424	−0.05 (−0.07, −0.03)5	−0.04 (−0.06, −0.02)5	2504	−0.04 (−0.06, −0.01)4	−0.02 (−0.05, 0.002)
≥8	1487	−0.04 (−0.06, −0.01)4	−0.06 (−0.08, −0.03)5	1838	−0.06 (−0.08, −0.03)5	−0.05 (−0.07, −0.02)5
Cups/d, trend		−0.004 (−0.006, −0.002)5	−0.006 (−0.008, −0.003)5		−0.004 (−0.007, −0.002)5	−0.004 (−0.006, −0.002)5
CRP, log mg/L
0	35,033	Ref.	Ref.	26,937	Ref.	Ref.
<1	6810	−0.07 (−0.10, −0.04)5	−0.006 (−0.03, 0.02)	6327	−0.09 (−0.11, −0.06)5	−0.02 (−0.05, 0.0004)4
1	18,914	−0.10 (−0.12, −0.08)5	−0.009 (−0.02, 0.007)	17,110	−0.10 (−0.12, −0.08)5	−0.02 (−0.03, 0.0002)4
2–3	68,042	−0.10 (−0.11, −0.09)5	−0.007 (−0.02, 0.005)	56,994	−0.11 (−0.12, −0.10)5	−0.01 (−0.03, −0.0005)4
4–5	59,859	−0.09 (−0.10, −0.07)5	−0.02 (−0.03, −0.005)4	47,871	−0.08 (−0.10, −0.07)5	−0.002 (−0.02, 0.01)
6–7	27,783	−0.06 (−0.07, −0.04)5	−0.03 (−0.04, −0.01)5	21,975	−0.04 (−0.05, −0.02)5	0.01 (−0.005, 0.03)
≥8	14,789	−0.002 (−0.02, 0.02)	−0.03 (−0.05, −0.01)5	14,768	0.04 (0.02, 0.06)5	0.02 (0.0007, 0.04)4
Cups/d, trend		0.001 (−0.0003, 0.003)	−0.003 (−0.005, −0.002)5		0.007 (0.006, 0.009)5	0.004 (0.002, 0.005)5
Fasting glucose, mmol/L
0	3996	Ref.	Ref.	3938	Ref.	Ref.
<1	733	0.02 (−0.02, 0.06)	0.03 (−0.01, 0.07)	845	0.007 (−0.05, 0.06)	0.01 (−0.04, 0.06)
1	1919	0.01 (−0.02, 0.05)	0.01 (−0.02, 0.04)	2251	−0.05 (−0.09, −0.02)4	−0.04 (−0.07, −0.006)4
2–3	6107	0.02 (−0.006, 0.04)	0.02 (−0.007, 0.04)	6418	−0.03 (−0.06, −0.001)	−0.004 (−0.03, 0.02)
4–5	4740	−0.001 (−0.02, 0.02)	0.005 (−0.02, 0.03)	5021	−0.04 (−0.07, −0.01)4	−0.01 (−0.04, 0.02)
6–7	2215	−0.007 (−0.04, 0.02)	−0.001 (−0.03, 0.03)	2304	−0.03 (−0.06, 0.009)	0.02 (−0.02, 0.05)
≥8	1365	−0.04 (−0.08, −0.01)4	−0.03 (−0.06, 0.009)	1718	−0.09 (−0.13, −0.05)5	−0.03 (−0.07, 0.007)
Cups/d, trend		−0.004 (−0.007, −0.001)4	−0.002 (−0.005, 0.0005)		−0.006 (−0.01, −0.003)5	−0.001 (−0.005, 0.002)
HbA1c, mmol/mol
0	34,852	Ref.	Ref.	26,722	Ref.	Ref.
<1	6794	−0.13 (−0.24, −0.02)4	0.06 (−0.03, 0.16)	6285	−0.23 (−0.37, −0.08)5	0.11 (−0.01, 0.22)
1	18,717	−0.19 (−0.26, −0.12)5	0.09 (0.02, 0.15)4	17,006	−0.43 (−0.53, −0.33)5	0.05 (−0.03, 0.14)
2–3	67,606	−0.15 (−0.20, −0.10)5	0.17 (0.12, 0.21)5	56,638	−0.50 (−0.58, −0.43)5	0.11 (0.04, 0.17)5
4–5	59,803	−0.11 (−0.16, −0.06)5	0.17 (0.12, 0.22)5	47,788	−0.40 (−0.48, −0.33)5	0.18 (0.12, 0.25)5
6–7	27,743	−0.04 (−0.11, 0.02)	0.16 (0.10, 0.22)5	21,950	−0.22 (−0.31, −0.12)5	0.26 (0.18, 0.34)5
≥8	14,784	0.15 (0.07, 0.23)5	0.14 (0.07, 0.22)5	14,701	0.11 (0.001, 0.21)4	0.29 (0.20, 0.38)5
Cups/d, trend		0.02 (0.01, 0.02)5	0.01 (0.008, 0.02)5		0.02 (0.01, 0.03)5	0.03 (0.02, 0.04)5

¹β Coefficients represent change in serum biomarker concentration with increasing tea intake relative to no tea intake. CRP, Lp(a), and TG concentrations were log transformed before analysis and thus β coefficients represent percentage change. At least nominally significant tests for sex heterogeneity were observed for apoB (P = 0.009), tChol (P < 0.0001), HDL (P = 0.005), LDL (P < 0.0001), and CRP (P = 0.0008). CRP, C-reactive protein; HbA1c, glycated hemoglobin; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; Lp(a), lipoprotein (a); tChol, total cholesterol; TG, triglyceride.

²Model 1 adjusted for age, race, date of blood draw, fasting time, and assessment center.

³Model 1 further adjusted for Townsend deprivation index; education; income; employment status; home ownership; smoking; physical activity; waist-to-hip ratio; BMI; oral contraceptive use (women only); postmenopausal hormone use (women only); self-rated health; aspirin use; cholesterol-lowering medication use; antihypertension medication use; history of diabetes; history of cardiovascular disease; intakes of water, alcohol, fish, red meat, fruits, and vegetables; recent caffeine intake; and habitual coffee intake.

⁴0.0025 ≤ P < 0.05.

⁵ P < 0.0025.

TABLE 5.

Associations between tea consumption (24-h diet recalls) and cardiometabolic biomarkers¹

	Women		Men
Cups/d	Black	Green2	Black	Green2
ApoA-1, g/L
0	Ref.		Ref.
<1	0.004 (−0.005, 0.014)	0.002 (−0.006, 0.010)	0.007 (−0.003, 0.016)	0.0009 (−0.008, 0.010)
1	0.004 (−0.008, 0.015)	−0.015 (−0.029, −0.0005)3	0.002 (−0.009, 0.012)	0.006 (−0.010, 0.022)
2–3	0.005 (−0.004, 0.013)	0.003 (−0.009, 0.015)	0.012 (0.004, 0.021)3	0.005 (−0.007, 0.018)
4–5	0.001 (−0.008, 0.010)	−0.008 (−0.031, 0.015)	0.009 (0.0001, 0.019)3	0.015 (−0.011, 0.041)
≥6	0.017 (0.005, 0.029)3	—	0.012 (0.0001, 0.023)3	—
Cups/d, trend	0.0003 (−0.001, 0.002)	−0.001 (−0.005, 0.003)	0.0009 (−0.0006, 0.002)	0.003 (−0.001, 0.007)
ApoB, g/L
0	Ref.		Ref.
<1	0.002 (−0.006, 0.010)	−0.001 (−0.008, 0.006)	0.004 (−0.005, 0.014)	0.003 (−0.006, 0.013)
1	−0.004 (−0.013, 0.006)	0.004 (−0.008, 0.015)	0.004 (−0.006, 0.015)	−0.002 (−0.018, 0.014)
2–3	−0.006 (−0.013, 0.001)	−0.009 (−0.019, 0.0012)	−0.005 (−0.013, 0.004)	−0.014 (−0.026, −0.0008)3
4–5	−0.015 (−0.023, −0.007)4	−0.019 (−0.038, −0.00003)3	−0.010 (−0.019, −0.0008)3	−0.005 (−0.031, 0.020)
≥6	−0.020 (−0.030, −0.010)4	—	−0.015 (−0.027, −0.003)3	—
Cups/d, trend	−0.005 (−0.006, −0.003)4	−0.004 (−0.007, −0.0006)3	−0.004 (−0.005, −0.002)4	−0.004 (−0.008, 0.0007)
Lp(a), log nmol/L
0	Ref.		Ref.
<1	−0.011 (−0.058, 0.036)	0.006 (−0.034, 0.046)	−0.009 (−0.068, 0.050)	−0.009 (−0.064, 0.045)
1	−0.023 (−0.078, 0.032)	0.031 (−0.039, 0.101)	−0.047 (−0.112, 0.017)	0.017 (−0.078, 0.113)
2–3	−0.014 (−0.055, 0.027)	0.097 (0.037, 0.157)4	−0.030 (−0.082, 0.021)	−0.0009 (−0.078, 0.076)
4–5	−0.027 (−0.071, 0.018)	−0.094 (−0.204, 0.017)	−0.006 (−0.062, 0.050)	0.125 (−0.032, 0.282)
≥6	−0.064 (−0.124, −0.005)3	—	−0.0003 (−0.070, 0.070)	—
Cups/d, trend	−0.002 (−0.010, 0.006)	0.011 (−0.008, 0.030)	0.007 (−0.003, 0.016)	0.014 (−0.012, 0.040)
tChol, mmol/L
0	Ref.		Ref.
<1	0.016 (−0.021, 0.053)	−0.010 (−0.042, 0.021)	0.011 (−0.033, 0.055)	0.026 (−0.015, 0.067)
1	0.005 (−0.038, 0.048)	−0.011 (−0.066, 0.043)	0.010 (−0.038, 0.059)	0.020 (−0.052, 0.092)
2–3	−0.004 (−0.037, 0.028)	−0.039 (−0.086, 0.008)	−0.004 (−0.043, 0.034)	−0.043 (−0.101, 0.014)
4–5	−0.032 (−0.067, 0.004)	−0.111 (−0.199, −0.024)3	−0.021 (−0.063, 0.021)	−0.061 (−0.178, 0.055)
≥6	−0.014 (−0.060, 0.032)	—	−0.039 (−0.091, 0.013)	—
Cups/d, trend	−0.010 (−0.016, −0.004)4	−0.021 (−0.036, −0.006)3	−0.010 (−0.017, −0.003)3	−0.014 (−0.033, 0.005)
HDL-C, mmol/L
0	Ref.		Ref.
<1	0.009 (−0.004, 0.022)	0.002 (−0.009, 0.013)	0.005 (−0.008, 0.018)	0.004 (−0.008, 0.016)
1	0.006 (−0.009, 0.021)	−0.014 (−0.033, 0.005)	0.003 (−0.012, 0.017)	0.012 (−0.009, 0.033)
2–3	0.013 (0.001, 0.024)3	0.013 (−0.004, 0.029)	0.014 (0.003, 0.025)3	0.013 (−0.003, 0.030)
4–5	0.017 (0.004, 0.029)3	−0.021 (−0.052, 0.009)	0.014 (0.001, 0.026)3	0.025 (−0.008, 0.059)
≥6	0.047 (0.030, 0.063)4	—	0.021 (0.006, 0.036)3	—
Cups/d, trend	0.005 (0.003, 0.007)4	−0.0005 (−0.006, 0.005)	0.003 (0.0009, 0.005)3	0.006 (0.0008, 0.012)3
LDL-C, mmol/L
0	Ref.		Ref.
<1	0.006 (−0.022, 0.034)	−0.003 (−0.027, 0.021)	0.012 (−0.021, 0.046)	0.018 (−0.013, 0.049)
1	−0.008 (−0.041, 0.026)	0.006 (−0.036, 0.048)	0.009 (−0.028, 0.045)	0.012 (−0.043, 0.067)
2–3	−0.013 (−0.037, 0.012)	−0.035 (−0.071, 0.0004)	−0.006 (−0.035, 0.023)	−0.041 (−0.085, 0.003)
4–5	−0.035 (−0.062, −0.008)3	−0.077 (−0.144, −0.010)3	−0.016 (−0.048, 0.016)	−0.032 (−0.121, 0.057)
≥6	−0.035 (−0.071, 0.0007)3	—	−0.027 (−0.067, 0.013)	—
Cups/d, trend	−0.010 (−0.015, −0.005)4	−0.015 (−0.027, −0.003)3	−0.007 (−0.013, −0.002)3	−0.011 (−0.026, 0.004)
Fasting TGs, log mmol/L
0	Ref.		Ref.
<1	−0.019 (−0.065, 0.027)	−0.041 (−0.085, 0.004)	0.006 (−0.059, 0.071)	−0.030 (−0.099, 0.038)
1	−0.012 (−0.070, 0.046)	−0.034 (−0.106, 0.039)	0.038 (−0.037, 0.112)	−0.050 (−0.166, 0.066)
2–3	−0.036 (−0.078, 0.007)	−0.027 (−0.090, 0.036)	−0.049 (−0.109, 0.012)	−0.013 (−0.095, 0.069)
4–5	−0.020 (−0.067, 0.027)	−0.020 (−0.130, 0.090)	−0.084 (−0.150, −0.017)3	−0.168 (−0.353, 0.017)
≥6	−0.031 (−0.098, 0.035)	—	−0.129 (−0.214, −0.045)3	—
Cups/d, trend	−0.002 (−0.011, 0.007)	−0.015 (−0.035, 0.005)	−0.025 (−0.036, −0.015)4	−0.021 (−0.049, 0.008)
CRP, log mg/L
0	Ref.		Ref.
<1	−0.025 (−0.059, 0.008)	−0.010 (−0.038, 0.019)	−0.008 (−0.049, 0.032)	−0.001 (−0.039, 0.036)
1	−0.018 (−0.056, 0.021)	0.008 (−0.041, 0.057)	0.011 (−0.033, 0.056)	0.022 (−0.044, 0.087)
2–3	−0.034 (−0.062, −0.005)3	−0.056 (−0.098, −0.014)3	−0.022 (−0.057, 0.013)	−0.029 (−0.082, 0.023)
4–5	−0.051 (−0.083, −0.019)4	−0.149 (−0.227, −0.070)4	−0.038 (−0.076, 0.0006)	0.058 (−0.048, 0.164)
≥6	−0.105 (−0.147, −0.063)4	—	−0.053 (−0.101, −0.005)3	—
Cups/d, trend	−0.016 (−0.022, −0.011)4	−0.038 (−0.052, −0.025)4	−0.007 (−0.013, −0.0008)3	0.0003 (−0.017, 0.018)
Fasting glucose, mmol/L
0	Ref.		Ref.
<1	0.006 (−0.057, 0.068)	−0.024 (−0.084, 0.036)	0.052 (−0.037, 0.142)	−0.060 (−0.153, 0.032)
1	0.078 (0.0003, 0.156)3	0.038 (−0.060, 0.135)	0.044 (−0.059, 0.146)	−0.117 (−0.278, 0.043)
2–3	0.048 (−0.010, 0.106)	0.030 (−0.056, 0.116)	0.037 (−0.046, 0.120)	−0.014 (−0.129, 0.101)
4–5	0.043 (−0.020, 0.107)	0.099 (−0.049, 0.246)	0.048 (−0.042, 0.139)	0.029 (−0.220, 0.278)
≥6	0.014 (−0.075, 0.103)	—	0.035 (−0.082, 0.152)	—
Cups/d, trend	−0.001 (−0.013, 0.011)	0.021 (−0.005, 0.048)	0.005 (−0.010, 0.020)	−0.002 (−0.041, 0.036)
HbA1c, mmol/mol
0	Ref.		Ref.
<1	0.158 (0.026, 0.291)3	0.026 (−0.088, 0.139)	0.114 (−0.076, 0.305)	0.117 (−0.061, 0.294)
1	0.186 (0.031, 0.341)3	0.086 (−0.111, 0.283)	0.099 (−0.111, 0.309)	−0.011 (−0.320, 0.297)
2–3	0.187 (0.072, 0.303)4	0.060 (−0.107, 0.228)	0.192 (0.026, 0.358)3	0.191 (−0.057, 0.438)
4–5	0.266 (0.139, 0.393)4	−0.115 (−0.430, 0.199)	0.182 (0.002, 0.363)3	−0.208 (−0.717, 0.302)
≥6	0.271 (0.104, 0.438)4	—	0.202 (−0.024, 0.428)	—
Cups/d, trend	0.038 (0.016, 0.061)4	0.009 (−0.045, 0.063)	0.028 (−0.001, 0.058)	0.030 (−0.053, 0.113)

¹Values are β (95% CI). Results from linear regression models adjusted for age; race; date of blood draw; fasting time; assessment center; Townsend deprivation index; education; income; employment status; home ownership; smoking; physical activity; waist-to-hip ratio; BMI; oral contraceptive use (women only); postmenopausal hormone use (women only); self-rated health; aspirin use; cholesterol-lowering medication use; antihypertension medication use; history of diabetes; history of cardiovascular disease; total energy intake; intakes of carbohydrates, fat, saturated fat, protein, fizzy drinks, milk, water, alcohol, fish, red meat, fruits, and vegetables; recent caffeine drinking; coffee; herbal tea; redbush/rooibos tea; other types of tea; and green (for black tea analysis) or black (for green tea analysis) tea. CRP, C-reactive protein; HbA1c, glycated hemoglobin; HDL-C, HDL cholesterol; LDL-C, LDL cholesterol; Lp(a), lipoprotein (a); tChol, total cholesterol; TG, triglyceride.

²The maximum tea intake category for green tea is ≥4.

³0.0025 ≤ P < 0.05.

⁴ P < 0.0025.

Smoking modified the associations of total tea consumption with HDL cholesterol and HbA1c (Supplemental Table 11). Tea consumption was associated with higher HDL cholesterol in nonsmokers and tended to be associated with lower HDL cholesterol in smokers. The direct association between tea consumption and HbA1c was stronger among smokers than among nonsmokers. In addition, select tea–biomarker associations were significantly modified by age and adiposity (see Supplemental Notes).

Each additional cup of tea was associated with significantly lower odds of abnormal apoB, tChol, HDL-cholesterol, LDL-cholesterol, and TG concentrations by 1%–4% (Supplemental Table 12).

SNP × coffee and SNP × tea interactions

In the current analysis, no statistically significant SNP × coffee (P > 0.003) or SNP × tea (P > 0.02) interactions were observed in relation to cardiometabolic biomarkers. Any nominally significant interactions were also sex-specific and aside from modest differences in effect size were difficult to interpret upon stratified analysis. Tests for SNP × coffee and SNP × tea interactions performed separately for smokers and nonsmokers were also nonsignificant, as were tests for SNP × caffeine (derived from tea and regular coffee) interactions.

Because rs1481012, rs1260326, and rs7800944 have each been associated with lipid concentrations and coffee consumption behavior in GWASs (Supplemental Table 2), they may thus act as confounders. However, additional adjustment for these SNPs did not appreciably change the results (data not shown).

Discussion

Coffee and tea are among the most widely consumed beverages in the world. In the UK Biobank, coffee consumption was associated with higher LDL-cholesterol and apoB concentrations, but this association was limited to specific brew types such as espresso and not observed for instant coffee. In contrast, tea consumption was associated with lower LDL-cholesterol and apoB concentrations. Both coffee and tea consumption were associated with lower fasting TG and CRP concentrations. None of the observed associations was modified by genetic factors related to caffeine metabolism. Results for HbA1c varied by sex and other lifestyle factors and warrant further investigation.

In the UK Biobank, the associations of coffee consumption with blood lipids and apos varied by type of coffee brew. Although caffeine per se is unlikely to increase lipids (27), boiled coffee contains relatively large amounts of cafestol and kahweol—diterpenes with known lipid-raising effects—and espresso coffee has intermediate concentrations of diterpenes (11, 12). Indeed, espresso coffee, the most concentrated source of diterpenes distinguished in the UK Biobank, had the strongest direct associations with tChol, LDL cholesterol, and apoB. Taking the mean of our estimates for men and women, a 3-cup/d higher consumption of espresso was associated with a 0.20-mmol/L (∼5.8%) higher LDL-cholesterol concentration, which could be associated with a 5% higher risk of major CVDs (28). In a randomized trial in young Italian men (29), 3 cups of espresso per day led to a nonsignificant increase in tChol of 0.10 mmol/L and LDL cholesterol of 0.15 mmol/L. Similarly, based on the diterpene content of 3 cups of espresso, a 0.09-mmol/L increase in tChol would be expected (12). Based on these data, a larger trial of the effects of espresso coffee on blood lipids is warranted.

Instant coffee, devoid of diterpenes and the most commonly consumed coffee in the United Kingdom, was only weakly associated with increased LDL cholesterol and only in men with a consumption of ≥4 cups/d. Filtered coffee also has a low diterpene content, but in the UK Biobank 24-h diet recalls it was grouped with diterpene-containing cafeteria and Americano brews; together these were associated with higher LDL-cholesterol and apoB concentrations. Cappuccinos and lattes, espresso- and milk-based coffees, contain lower concentrations of diterpenes and generally low amounts of consumption and combination with other potentially unique constituents may have contributed to their weaker associations with lipid profiles (11, 30). Consumption of instant coffee was also associated with at least nominally significant higher HDL-cholesterol and lower TG concentrations, suggesting that mechanisms unrelated to the cholesterol-raising diterpenes may also be at play. Trends for lower apoA-1 with higher instant coffee consumption were unexpected in light of the higher HDL-cholesterol concentrations. Our findings for TGs are in agreement with most other cross-sectional studies (31–34). Berndt et al. (32) suggested that the inverse association of coffee consumption with TGs may reflect a decrease in TG-rich lipoproteins (i.e., VLDL and chylomicrons), whereas the direct association with LDL cholesterol may result from an accelerated breakdown of these lipoproteins (32). Our coffee–blood lipid associations align only partly with results from RCTs, in which unfiltered coffee increased tChol, LDL-cholesterol, and TG concentrations and did not change HDL-cholesterol concentrations compared with control beverages (5). Disparities may be a result of the blood-sampling conditions in the UK Biobank, bias in our study, or real biological response differences with long-term compared with short-term coffee consumption.

Consumption of tea, which does not contain diterpenes, was also associated with higher HDL-cholesterol concentrations, but lower concentrations of other blood lipids. Results from RCTs (35) and cross-sectional studies (36–40) have suggested beneficial or neutral relations between green or black tea consumption amounts and blood lipid concentrations. Most of these studies have included Asian participants and thus our findings for a large British population with different cultural characteristics are an important addition to the literature. Catechins (in green tea), theaflavins (in black tea), and theanine are constituents unique to tea that may contribute to the benefits of tea drinking on lipid profiles (10).

In the UK Biobank, coffee intake was inversely associated with CRP, which is consistent with previous cross-sectional studies, but not RCTs, which report no effect of coffee on CRP concentrations (41, 42). We also observed a similar but weaker inverse association between tea consumption and CRP concentrations. RCTs do not support a beneficial effect of green tea on CRP concentrations and the few RCTs of black tea have included participants with T2D or CVD, yielding inconsistent results (43–49). Cross-sectional studies of tea and CRP have been few and inconsistent (43–47).

RCTs and cross-sectional studies provide no support for an effect of habitual coffee or tea intake on concentrations of HbA1c (4, 44, 50–53). In the current study, the impact of coffee on HbA1c concentrations differed by sex, smoking status, and BMI: any clinically relevant benefits of coffee on HbA1c were observed among women, nonsmokers, and those with a higher BMI. For men and women, slightly but significantly higher HbA1c concentrations were observed with higher intake of tea and particularly black tea. The direct association may be a result of confounding because a biological interpretation is difficult and clinical interventions of individual tea constituents suggest the opposite (54). Tea, black tea in particular, is a traditional beverage of the British and may be a marker of socioeconomic status (55), which is potentially a strong determinant of HbA1c (56). Taken together, our findings for HbA1c are novel but warrant confirmation in independent studies.

Coffee and tea were not associated with fasting glucose concentrations, which is consistent with most cross-sectional studies (57). Interestingly, several cross-sectional studies (52, 53, 58–60) and an RCT (61) have suggested a beneficial effect of coffee on only 2-h postload glucose.

We observed strong smoking × coffee interactions for all biomarkers with the exception of tChol, TGs, and fasting glucose. Generally, the favorable associations between coffee and biomarkers were more evident among nonsmokers, whereas nonfavorable associations were more evident among smokers. These patterns of associations may reflect biological interaction between coffee consumption and smoking, but we cannot discount the possibility of residual confounding by smoking behaviors, which was highly correlated with coffee consumption. A reduction in HDL cholesterol is the most widely documented lipid abnormality related to smoking (62). It is possible that measurement error in smoking was disproportionally distributed across amounts of coffee intake; the result of this confounding would have attenuated associations between coffee consumption and favorable concentrations of particular biomarkers.

The current study accounted for several lifestyle and genetic factors that may alter caffeine metabolism (8, 63). The analysis of potential interaction with genetic factors, in particular, may in addition offer insight into the role caffeine per se might play in associations. The few studies of caffeine-containing beverages and cardiometabolic biomarkers that have considered genetic variation in caffeine metabolism have focused on CYP1A2, included relatively small sample sizes, and together yielded inconclusive results (64–67). The current study considered a more comprehensive list of SNPs, none of which modified the associations between beverage consumption and cardiometabolic biomarkers, suggesting noncaffeine constituents of coffee and tea more likely underlie these associations. Zhou and Hyppönen (68) recently examined the association between habitual coffee consumption and risk of CVD in the UK Biobank. Compared with participants drinking 1–2 cups/d, the risk of CVD was elevated for nondrinkers, drinkers of decaffeinated coffee, and those who reported drinking >6 cups/d, regardless of genetic variation in caffeine metabolism (68). However, their analysis did not account for coffee brew type.

Strengths of the current study include the very large sample size, detailed coffee and tea consumption data, and a comprehensive list of potential confounders or modifiers. Using the same population sample, we recently reported a direct association between caffeine consumption close to the time of blood draw and fasting glucose concentrations (M Cornelis, unpublished results, 2020). Recent caffeine drinking was not associated with other cardiometabolic biomarkers. The associations we report in the current study were thus unlikely a result of the acute effects of caffeine, but instead likely a reflection of habitual beverage consumption. Nevertheless, several limitations need to be considered. Our study design was cross-sectional, thus limiting the ability to make causal inferences. In addition, our study was based on self-reported coffee and tea consumption and information on other beverages consumed was limited for the full sample. These factors taken together would have likely introduced measurement error that would have reduced the power of the study to detect true effects. Furthermore, given the observational nature of the study, we cannot exclude the possibility that the observed associations were affected by residual confounding. Finally, the UK Biobank is not representative of the source population, with evidence of a “healthy volunteer” selection bias, which may limit the generalizability of our findings (69).

In summary, our findings suggest that habitual consumption of coffee may have both beneficial and detrimental effects on blood lipids, but that these effects may differ substantially by brew type. For example, espresso coffee, but not instant coffee, was associated with substantially higher LDL-cholesterol and apoB concentrations. In addition, our findings support a beneficial effect of tea consumption on blood lipids and benefits for both coffee and tea consumption for chronic inflammation. Findings were not modified by genetic factors affecting caffeine metabolism, suggesting an important role of noncaffeine constituents of these beverages for cardiometabolic health.

Notes

Supported by National Institute on Aging grant K01AG053477 (to MCC).

Author disclosures: The authors report no conflicts of interest.

Any opinions, findings, conclusions, or recommendations expressed in this publication are those of the authors and do not necessarily reflect the view of the National Institute on Aging.

Supplemental Methods, Supplemental Notes, and Supplemental Tables 1–12 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.

Abbreviations used: CRP, C-reactive protein; CVD, cardiovascular disease; GWAS, genome-wide association study; HbA1c, glycated hemoglobin; Lp(a), lipoprotein (a); QC, quality control; RCT, randomized controlled trial; SNP, single-nucleotide polymorphism; tChol, total cholesterol; TG, triglyceride; T2D, type 2 diabetes; WHR, waist-to-hip ratio.