Light to moderate coffee consumption is associated with lower risk of death: a UK Biobank study

Aims To study the association of daily coffee consumption with all-cause and cardiovascular (CV) mortality and major CV outcomes. In a subgroup of participants who underwent cardiovascular magnetic resonance (CMR) imaging, we evaluated the association between regular coffee intake and cardiac structure and function. Methods


Introduction
Even though coffee is among the most consumed beverages in the world, little is known about the long-term impact of its regular consumption on cardiovascular (CV) health.Besides caffeine, it contains many bioactive components such as minerals and antioxidants (1,2).
Recent studies have shown that coffee plays a preventive role against cancer, obesity, type 2 diabetes mellitus, Parkinson's disease and dementia (3)(4)(5)(6)(7).However, there are inconsistent results regarding its CV effects.Although most studies have found no relationship between regular coffee intake and CV disease mortality, some have reported a moderate inverse association while others have found an increased risk (8)(9)(10)(11)(12).
Coffee is mainly consumed as ground or instant form, containing different chemical compounds.Instant coffee is reported to contain not only more caffeine and antioxidants as ground type coffee, but it also has twice as much acrylamide, which was shown to be neurotoxic and carcinogenic (13,14).While several studies have investigated the relationship between coffee type and cancer, the association between coffee type and subclinical cardiac alterations are unclear (15)(16)(17).
We studied, in the UK Biobank, the association of regular coffee consumption with all-cause mortality, CV mortality, and cardiovascular magnetic resonance (CMR) phenotypes.We limited to participants without clinically manifested heart disease at the time of recruitment.
In secondary analyses, we investigated the relationship between consumed coffee type and cardiac structural and functional alterations.

Study sample and outcomes
The UK Biobank is a prospective cohort study which collected questionnaire data, physical measurements and biological samples from over half a million 40-69 year-old individuals in the United Kingdom recruited between 2006 and 2010 (18).Baseline assessment of the participants included detailed assessment of medical history, lifestyle and nutritional habits, physical examination and blood sampling.Exclusion criteria of the current study were refusal to report coffee drinking habits, those who drink >25 cups of coffee per day, presence of heart failure, angina, prior myocardial infarction and stroke at the time of recruitment and refusal to consent.The endpoints of death, stroke and myocardial infarction were derived from Hospital Episode Statistics and death register data.CV mortality was defined as deaths where underlying (primary) causes were related to the circulatory system.Detailed list of causes with number of cases is reported in Supplementary table 1.

Patient and public involvement
No patients were involved in setting the research question or the outcome measures, nor were they involved in developing plans for design or implementation of the study.No patients were asked to advise on interpretation or writing up of results.There are no plans to disseminate the results of the research to study participants or the relevant patient community.Arterial stiffness index (ASI) was measured at the baseline visit with finger photoplethysmography with the PulseTrace PCA2 device (CareFusion, San Diego, CA, USA) while participants were seated.Readings were taken over 10-15 seconds.A detailed protocol has been described (https://biobank.ndph.ox.ac.uk/showcase/showcase/docs/Pulsewave.pdf).

Cardiovascular measurements
ASI provides a measure of large artery stiffness, an indicator of arterial health and ischemic risk (19).
In total, 100,000 participants were recalled to undergo comprehensive imaging of the brain, heart, whole body, carotid artery, bone and joints.Imaging of the heart was performed by CMR (20).The UK Biobank CMR protocol has been described in detail previously (21,22).Briefly, all examinations were performed on a 1.5 Tesla scanner (MAGNETOM Aera, Syngo Platform VD13A, Siemens Healthcare, Erlangen, Germany).For cardiac function, long-axis cines and a complete short-axis stack of balanced steady-state free precession (bSSFP) cines were acquired covering the left and right ventricle.The manual analysis dataset of 5,065 participants was used to develop an automated image analysis pipeline, which has been propagated to the first 32,000 CMR studies (23).Those with heart failure, prior myocardial infarction, stroke and angina at imaging visit were excluded from the CMR subanalysis.We examined left and right ventricular end-diastolic (LVEDV; RVEDV), end-systolic (LVESV; RVESV), and stroke volumes (LVSV; RVSV), left ventricular mass (LVM) and left and right ventricular ejection fractions (LVEF; RVEF).

Measurement of baseline covariates, potential confounders and coffee consumption
Data regarding coffee consumption were obtained from standardized and validated questionnaires filled in by the study participants at baseline visit.Participants were asked their average coffee intake in the last year: "how many cups of coffee do you drink each day", as well as the most common type of consumed coffee (decaffeinated, ground, instant or other Downloaded from https://academic.oup.com/eurjpc/advance-article/doi/10.1093/eurjpc/zwac008/6512055 by University of Southampton user on 15 February 2022 type).We selected confounders and possible mediators of the relationship between coffee consumption with CV health.These covariates were determined from participant interview or touchscreen questionnaires.The following variables were recorded: age, sex, non-European ethnicities, Townsend deprivation index (a socio-economic measure based on area of residence), weight and height, cardiometabolic comorbidities such as hypertension or diabetes.Moreover, lifestyle factors such as physical activity (expressed as metabolic equivalent (MET) minutes/week), fresh and dried fruit, raw or cooked vegetable intake (portions per day), tea intake (cups per day), alcohol intake frequency (never, special occasions only, 1-3 times per month, 1-2 times per week, 3-4 times per week, and daily or almost daily), meat intake frequency (never, less than once per week, once per week, 2-4 times per week, 5-6 times per week, once or more daily) and smoking status (never smoker and previous or current smoker) were also ascertained.Detailed questions of the UK Biobank questionnaires can be found in UK Biobank Data Showcase (https://www.ukbiobank.ac.uk/data-showcase/).Total cholesterol levels were determined with blood biochemistry.

Data analysis and statistics
Summary statistics for independent variables were calculated as means and standard deviation (SD) for continuous variables.Categorical variables were expressed as frequencies and percentages.Consistent with previous studies, regular coffee consumption was categorized into 3 groups: zero, light-to-moderate (0.5-3 cups/day), and high (>3 cups per day) with the lowest group (zero coffee consumption) used as the reference in the analyses (24).The eventfree survival rate was estimated using Kaplan-Meier method and log-rank test was applied for the comparisons between the various coffee intake groups.Cumulative event rates were calculated with event or censoring times measured from the date of baseline visit.For participants without outcome, time-to-event measures were censored at the latest UK Biobank censor dates giving a follow-up duration of 10 to 15 years.To assess the relationship between the amount of coffee intake and CV morbidity and mortality, uni-and multivariable Cox proportional hazard regression models were executed.
We considered two approaches to covariate adjustment.Model 1 was created as a "true confounder model".In this model, we adjusted for covariates that were considered confounders of the associations of interest, selected based on biological plausibility and existing evidence.The purpose of Model 1 was to quantify the magnitude of the exposureoutcomes associations.Model 2 included adjustment of both true confounders (all covariates in Model 1) and potential mediators.Mediators were considered covariates that may lie on the causal pathway for the association between coffee and CV health.The expectation is that in the presence of significant mediating effect from these covariates, their addition to the model (as in Model 2) would attenuate the magnitude of exposure-outcome associations observed in the true confounder model.The purpose of Model 2 was to explore potential mediating effect from selected cardiometabolic morbidities.Model 1 included coffee consumption categories and the following potential confounders: age, sex, non-European ethnicity, weight and height, smoking status, physical activity, Townsend deprivation index, regular alcohol and meat consumption, tea, cooked and raw fruit and vegetable intake.Model 2 included parameters of Model 1 and the following potential mediators: hypertension, diabetes and cholesterol level.
Associations between regular coffee intake and structural and functional CMR parameters were analyzed using unadjusted and multivariable linear regression analyses using the same models.Definitions for these covariates were previously described (25,26).In order to assess the association between most common type of consumed coffee and CMR parameters, we ran a subanalysis of decaffeinated, ground, instant or other coffee type.A p-value below 0.05 was

Results
After exclusion, 468,629 individuals were included in this study.At the time of recruitment their mean age was 56.2±8.1 years and 44.2% of the participants were male.Table 1 illustrates the characteristics of the study population.Among the 468,629 studied participants, 103,384 (22.1%) did not consume coffee on a regular basis, 274,088 (58.5%) consumed 0.5-3 cups per day and 91,157 (19.5%) >3 cups per day.

The association of regular coffee intake with cardiovascular outcomes and all-cause mortality
Median follow-up time was 11 (10-12) years.All-cause and CV mortality rates significantly differed among the various coffee consumption groups (all-cause mortality rate was 3.72%, 3.40% and 4.03% and CV mortality rate was 0.65%, 0.58% and 0.74% in zero, light-tomoderate and high coffee drinkers, respectively; both p<0.001).The univariate analysis showed that light-to-moderate coffee drinking was associated with decreased, while high coffee intake was linked with increased all-cause and CV mortality, as compared to zero coffee drinkers [HR of all-cause mortality 0.91 (95%CI=0.88-0.95) for light-to-moderate and .After adjustment for all potential confounders and mediators drinking 0.5-3 cups of coffee per day was linked with significantly decreased ASI [ß=-0.12(95%CI=-0.18to -0.06), p<0.001].Data of the association between regular coffee intake and measured arterial stiffness can be seen in Table 3.

The association of regular coffee consumption with cardiac structure and function
In order to evaluate the possible underlying mechanisms of the observed association between regular coffee consumption and health outcomes, CMR data of  4 and Figure 2.

The impact of coffee type on cardiovascular system
We additionally explored potential differential impact from different coffee type consumed.
We compared the differences between the various coffee types, as compared to zero coffee consumption category.Among those who drank coffee regularly, 71,404 (

Discussion
In this large cross-sectional population study of 468,629 individuals free from clinical cardiovascular disease, light-to-moderate coffee consumption was associated with decreased all-cause and CV mortality, and incident stroke.In comparison to zero coffee intake, lightmoderate and high coffee consumption was also associated with favorable cardiovascular phenotypes, both in terms of cardiac and arterial health.Importantly, these associations remained robust to adjustment for cardiometabolic morbidities (hypertension, high cholesterol, diabetes).

Comparison with other studies
Coffee is among the most widely consumed pharmacologically active beverages in the world.
In our study population, 77.9% consumed coffee daily.The CV effects of coffee consumption are a combination of favorable and unfavorable effects of caffeine and other drink components.Caffeine is by far the most studied compound of coffee.While caffeine has inotropic effect on the heart, induces high blood pressure, increases cholesterol level in Nordic countries but not in other populations, regular coffee intake was also associated with lower risk of type 2 diabetes mellitus, lower body weight and decreased platelet aggregation, inhibition of inflammation (28)(29)(30)(31)(32)(33).While coffee consumption has been associated with an acute increase in blood pressure in caffeine-naive individuals, this effect is negligible in habitual coffee drinkers, and many further studies came to the conclusion that long-term coffee consumption has no clinical importance on the risk of hypertension (34)(35)(36).In line with these findings, the prevalence of hypertension was not higher in those drinking >3 cups of coffee per day as compared to zero coffee drinkers in the present study.Moreover, in a subpopulation of 139,727 participants who underwent arterial stiffness measurement at baseline visit, light-to-moderate coffee consumption was associated with lower ASI, an Antioxidants in coffee have been reported to improve glucose metabolism and insulin sensitivity (37).Another study concluded that consumption of more than 5 cups of coffee per day increases adiponectin levels, and decrease insulin resistance (38).In our study population prevalence of diabetes mellitus was significantly lower in light-to-moderate coffee drinkers as compared to zero coffee intake, and similar in high and zero coffee consuming categories.
Interestingly, we did not observe a significant attenuation of associations with additional adjustment for diabetes, suggesting that associations reported in the present studies are mediated by alternative processes.
Several previous studies aimed to investigate the effect of regular coffee consumption on CV health.In a prospective study of two Spanish cohorts, 1-7 cups of caffeinated coffee per week was associated with lower risk of atrial fibrillation (39).Moreover, in an umbrella review of 201 meta-analyses, coffee consumption was more often linked with beneficial than harmful health outcomes including lower all-cause and CV mortality with the largest relative risk reduction for those consuming 3 to 4 cups/day versus zero (40).Consistent with these findings, in our study light-to-moderate coffee consumption defined as 0.5-3 cups per day was associated with lower risk of all-cause and CV mortality, as compared to zero coffee drinkers.
This favorable effect might be partly explained by lower arterial stiffness measures as well as by significantly increased stroke volume in both cardiac ventricles.

Clinical competencies and translational outlook
The UK Biobank offers a unique opportunity for the assessment of the potential differences between consumed coffee types.In our study population, ground and instant coffee were the two most commonly taken types.While ground coffee was associated with decreased allcause and CV mortality, we did not find statistically significant association between regular instant coffee consumption and health outcomes..The difference among the various coffee types may be explained by the differences in their production process, as they contain different chemicals.
In a recent study of 508,747 participants in the Norwegian cardiovascular surveys, unfiltered brew was linked with higher mortality than filtered brew.
Filtered brew was associated with lower mortality than no coffee consumption (41).
Interestingly, in our study population regularly consuming decaffeinated coffee was significantly linked with lower all-cause mortality as compared to zero coffee drinkers, suggesting that the observed associations can be explained only partly by caffeine itself.
However, further studies with more detailed information on consumed coffee type and preparation are needed to explain the underlying mechanisms.

Strengths and limitations of this study
The main strengths of our study are firstly, the large cohort of asymptomatic population with no prevalent CV disease where prospectively collected data, physical measurements and biological samples are available.Secondly, we used CMR which provides the most accurate and reproducible imaging modality in the assessment of cardiac structure and function.Our study's limitations are important to be acknowledged.First, data regarding coffee consumption was assessed by questionnaires, therefore recall bias may lead to inaccuracy.
Furthermore, the single snapshot of coffee consumption habits registered by UK Biobank might not accurately reflect the total lifetime coffee consumption, especially in older people.
The observed dose-response relationship between the amount of regular coffee consumption and cardiac changes indicates favorable effects on an epidemiological level despite the observed small effect sizes.Moreover, data regarding strength or size of the consumed coffee were not obtained from the participants, as questionnaire contained only data on cups of daily coffee intake and type of usually consumed coffee.Therefore, we could not calculate the caffeine content.In our study population, there were significant differences between the Downloaded from https://academic.oup.com/eurjpc/advance-article/doi/10.1093/eurjpc/zwac008/6512055 by University of Southampton user on 15 February 2022 coffee consuming categories in many dietary, sociological and lifestyle aspects, therefore the possibility that coffee consumption may be acting as a surrogate marker of some other CV risk factor cannot be fully excluded.

Conclusions
To our knowledge, this is the biggest study to systematically assess the CV effects of regular coffee consumption in a large asymptomatic population.Our results suggest that regular coffee consumption is safe, as even high daily coffee intake was not associated with adverse cardiovascular outcomes and all-cause mortality after a follow-up duration of 10 to 15 years.Moreover, 0.5-3 cups of coffee per day was independently associated with lower risk of allcause and CV mortality, and incident stroke.This favorable impact might be partly explained by lower ASI and subclinical beneficial alterations in cardiac structure and function.
This study was covered by the ethical approval for UK Biobank studies from the National Health Service (NHS) National Research Ethics Service on 17th June 2011 (Ref 11/NW/0382) and extended on 10 th May 2016 (Ref 16/NW/0274).

Table 2 .
Association of regular coffee consumption with all-cause and CV mortality, and CV disease incidence as compared with non-coffee drinkers

Table 3 .
Association between regular coffee consumption and ASI as compared to non-coffee drinkersConfounders: (baseline) age, sex, non-European ethnicities, body mass index, smoking, physical activity, Townsend deprivation index, alcohol, meat, tea, fruit and vegetable intake.

Table 4 .
Association between regular coffee consumption and cardiac structural and functional parameters