The authors have declared that no competing interests exist
Conceived and designed the experiments: SG CA VK SS SP. Performed the experiments: LI ADC. Analyzed the data: SG. Contributed reagents/materials/analysis tools: LI ADC. Wrote the paper: SG VK CA SS. Reviewed the manuscript for important intellectual content: VP RT GF FR PV GM CS PC AM.
The relationship between coffee consumption and coronary heart disease (CHD) has been investigated in several studies with discrepant results. We examined the association between Italian-style (espresso and mocha) coffee consumption and CHD risk.
We investigated 12,800 men and 30,449 women without history of cardiovascular disease recruited to the EPICOR prospective cohort study. Coffee consumption was assessed at baseline. In a random sub-cohort of 1472 subjects, plasma triglycerides, and total, LDL and HDL cholesterol were determined to investigate the effect of coffee consumption on plasma lipids.
After a mean follow up of 10.9 years, 804 cases of CHD (500 acute events, 56 fatal events and 248 revascularizations, all first events) were identified. Multivariable adjusted hazard ratios for CHD were: 1.18 (95% CI 0.87–1.60) for drinking 1–2 cups/day, 1.37 (95% CI 1.03–1.82) for >2–4 cups/day and 1.52 (95% CI 1.11–2.07) for over 4 cups/day (
Consumption of over 2 cups/day of Italian-style coffee is associated with increased CHD risk, but coffee consumption was not associated with plasma lipid changes, so the adverse effect of consumption appears unrelated to lipid profile.
Coffee, one of the most widely-consumed beverages in the world, is extracted from the roasted seeds (beans) of
The relation of coffee consumption to coronary heart disease (CHD) has been extensively studied. Some studies suggest that coffee consumption is not a major risk factor for CHD [
To our knowledge few case-control studies have investigated the effects of Italian-style coffee consumption on CHD risk. One study found increased risk of acute myocardial infarction for heavy consumers [
The aim of the present study was to prospectively investigate the relationship between the consumption of Italian-style coffee and the incidence of CHD in a large cohort of Italian men and women recruited in the EPICOR study [
EPICOR is a prospective cohort study to investigate the risk of cardiovascular disease in the five centers (Varese, Turin, Florence, Naples, Ragusa) of EPIC Italy [
Volunteers who did not complete dietary or lifestyle questionnaires (
A total of 43,249 volunteers age 35–74 years, mean follow-up 10.9 years (471,139 person years) was included in the analyses. In a subset of 1472 randomly selected for a case-cohort study nested in the EPIC population, Florence excluded, and for whom plasma lipids were available, we assessed the relation of plasma lipids to coffee consumption. The study complies with the declaration of Helsinki and the study protocol was approved by the ethics committee of the Human Genetics Foundation, Turin. Participants gave informed consent to use clinical data for research.
Participants completed a validated [
Information about medical history, physical activity, smoking, education, and other socioeconomic variables was collected using a standardized questionnaire. Height, weight and blood pressure were measured at enrollment using standardized procedures [
A 30ml fasting blood sample was collected using a standardized protocol, and stored in liquid nitrogen at 196°C [
During follow-up, electronic hospital discharge records and mortality files were linked to the study database to identify incident and fatal cases of CHD. Case definition was based on International Classification of Diseases (ICD) 9th edition codes (410–414) and procedure codes for coronary revascularization; each case was verified by checking that the diagnosis was consistent with examinations and procedures performed (particularly percutaneous transluminal coronary angioplasty and coronary artery bypass surgery). If records were not exhaustive or information was discordant, clinical records were accessed and checked; CHD was considered verified when acute myocardial infarction, acute coronary syndrome, or coronary revascularization was present. All cases were censored at the date of the first event and were cross-checked with mortality files to identify fatal and nonfatal cases (the later defined as alive 28 days after diagnosis).
CHD deaths were identified in the mortality database by ICD 10th edition codes: I20- I25, R96, and R99, and when codes E10-E14, I10-I13, I30, I31, I33-I38, I40, I42, I44-I51, I70-I74, and I77 were reported with I20-I25 as associated conditions. Fatal cases were confirmed by consulting death certificates.
Coffee consumption was categorized as <1 cup/day (non consumers and occasional drinkers), 1–2 cups/day (light drinkers), > 2–4 cups/day (moderate drinkers) and >4 cups/day (heavy drinkers). Baseline characteristics of subjects, according to categories of coffee intake, were summarized as means and standard deviations (continuous variables) or frequencies (categorical variables). Differences across categories were examined by ANOVA for continuous variables and χ2-test for categorical variables. Follow-up was time from enrolment to CHD diagnosis, death, loss to follow-up, or end of follow-up, whichever came first. End of follow-up was December 31 of the following years: 2006 for Varese and Naples, 2003 for Florence, 2008 for Turin, and 2007 for Ragusa. Multivariate Cox proportional hazard models were used to investigate the association between coffee consumption and CHD events, with age as primary time-dependent variable. Model 1 was the crude model adjusted for sex, age, and stratified by center. Model 2 was additionally adjusted for hypertension (yes/no), diabetes (yes/no), hyperlipidemia (yes/no), smoking status (current/former/never) and intensity (pack-years), education (≤8 years, >8 years), alcohol consumption (0, up to 12, >12 g/day women; 0, up to 24, >24 g/day men), tea consumption (0, up to 150 ml/day; ≥150 ml/day), BMI (<25, 25–30, >30 kg/m2), and physical activity (ordinal) as well as waist circumference, nonalcoholic energy intake, fruit and vegetables and saturated fatty acids intake (all continuous). HRs of CHD events, with 95% confidence intervals (CI), were estimated for categories of coffee consumption, with <1 cup/day as reference. Linear trends across categories were tested treating each category as a continuous variable in the Cox model. We also ran continuous models to investigate a dose-dependent effect for one cup increments. Interactions of coffee consumption with alcohol intake, smoking status, sex and hypertension were tested. Models with and without multiplicative interaction terms were compared using the likelihood ratio test. P values <0.05, 2-sided tests, were considered significant.
We also examined the association of coffee intake with plasma levels of total, LDL and HDL cholesterol and triglycerides in 1472 volunteers with data available. Differences in mean levels of plasma lipids between categories of coffee consumption were tested by analysis of variance after adjusting for the covariates included in model 2 (above). All analyses were performed using STATA version 11.2 (College Station, TX).
During a mean follow-up of 10.9 years, 804 cases of CHD (500 acute events, 56 fatal events and 248 revascularizations) were identified. Baseline characteristics of the cohort are presented by levels of daily coffee consumption in
<1 cup/day | 1–2 cups/day | >2–4 cups/day | >4 cups/day | |
---|---|---|---|---|
Participants (N) | 4073 | 10179 | 19612 | 9385 |
BMI <25 (%) | 52.1 | 46.1 | 44.6 | 41.1 |
Education >8 years (%) | 44.8 | 50.4 | 49.1 | 49.0 |
Alcohol (g/day) | 10.8 (16.9) | 12.9 (16.9) | 12.9 (16.3) | 11.4 (15.9) |
Hypertension, yes (%) | 38.1 | 42.7 | 39.0 | 33.9 |
Systolic blood pressure (mmHg) | 129.5 (18.4) | 130.9 (18.5) | 129.5 (18.0) | 127.4 (17.6) |
Diastolic blood pressure (mmHg) | 81.6 (9.9) | 82.4 (10.2) | 81.8 (10.1) | 81.0 (10.0) |
Smoking, yes (%) | 15.6 | 19.2 | 27.5 | 41.9 |
Diabetes, treated |
1.2 | 0.9 | 0.9 | 0.9 |
Hyperlipidemia, treated |
3.1 | 3.7 | 2.7 | 2.3 |
Non-alcohol energy (kcal/day) | 2111.6 (656.7) | 2121.5 (633.0) | 2234.6 (643.9) | 2419.0 (699.7) |
Age (years) | 50.3 (8.1) | 51.0 (7.9) | 50.3 (7.7) | 48.8 (7.4) |
Waist (cm) | 82.9 (12.0) | 84.3 (11.9) | 84.2 (11.7) | 84.5 (12.0) |
BMI (kg/m2) | 25.3 (4.1) | 25.9 (4.2) | 26.0 (4.0) | 26.3 (4.1) |
Inactive (%) | 28.3 | 33.1 | 28.4 | 27.3 |
Saturated fat (g/day) | 28.3 (11.2) | 28.6 (10.8) | 30.9 (11.1) | 34.5 (12.4) |
Tea (ml/day) | 86.0 (137.2) | 46.7 (88.5) | 34.2 (69.1) | 30.3 (70.1) |
Fruit and vegetables (g/day) | 547.0 (278.2) | 536.2 (255.9) | 540.4 (247.8) | 559.7 (264.3) |
Values are means with standard deviations in parentheses, unless otherwise indicated. All variables differ significantly (P <0.001) between categories of coffee consumption except diabetes.
* Self-reported on lifestyle questionnaire.
In the multivariable models (
Coffee intake | One cup/day increment | P trend | ||||
---|---|---|---|---|---|---|
Cups/day | <1 | 1–2 | >2–4 | >4 | ||
Cases/person-years | 55/44900 | 169/111648 | 374/212221 | 206/102370 | 804/471139 | |
HRs (95%CI) |
1.00 | 1.21 (0.89, 1.64) | 1.51 (1.14, 2.01) | 1.99 (1.48, 2.68) | 1.13 (1.08, 1.17) | <0.001 |
HRs (95%CI) |
1.00 | 1.18 (0.87, 1.60) | 1.37 (1.03, 1.82) | 1.52 (1.11, 2.07) | 1.06 (1.01, 1.10) | 0.002 |
* Model 1: adjusted for sex and age at recruitment, stratified by center.
† Model 2: model 1 with additional adjustments for non-alcohol energy intake, hypertension (yes/no), diabetes (yes/no), hyperlipidemia (yes/no), alcohol intake (0, up to 12, >12 for women; 0, up to 24, >24 for men), fruit and vegetables intake, tea consumption (0, up to 150 ml/day, >150 ml/day), saturated fatty acid intake, smoking status, smoking pack-years, education (≤8 years, >8 years), BMI (≤25, 25–30, >30), waist circumference (cm), and physical activity. One cup = 30 ml.
The association between coffee and CHD was independent of sex (P interaction 0.873), age (P interaction 0.058), smoking status (P interaction 0.596), hypertension (P interaction 0.897), BMI (P interaction 0.412) and alcohol consumption (P interaction 0.544). For the 1472 volunteers with laboratory data, triglycerides, and total, LDL and HDL cholesterol were unrelated to coffee consumption (
Total cholesterol, mg/dL | LDL mg/dL | HDL mg/dL | Triglycerides mg/dL | |
---|---|---|---|---|
<1 cup/day | 232.9 (4.3) | 145.5 (3.7) | 60.2 (1.2) | 135.9 (7.0) |
1–2 cups/day | 233.7 (2.6) | 145.3 (2.2) | 59.1 (0.7) | 147.0 (4.3) |
>2–4 cups/day | 234.9 (1.7) | 147.8 (1.5) | 58.2 (0.5) | 144.9 (2.8) |
>4 cups/day | 233.9 (2.5) | 148.1 (2.1) | 55.8 (0.7) | 149.7 (4.1) |
P value |
0.780 | 0.583 | 0.963 | 0.760 |
* Adjusted for sex, age at recruitment, non-alcohol energy intake, hypertension (yes/no), diabetes (yes/no), hyperlipidemia (yes/no), alcohol intake (0, up to 12, >12 for women; 0, up to 24, >24 for men), saturated fatty acid intake, smoking status, education (≤8 years, >8 years), BMI (≤25, 25–30, >30), waist circumference (cm), and physical activity.
† ANOVA.
In this study the risk of CHD was significantly greater than reference (<1 cup/day) for those whose intake of Italian-style coffee was greater than 2 cups per day. We also found, in a randomly selected sub-cohort, that coffee consumption was not significantly associated with plasma levels of total, LDL or HDL cholesterol, or triglycerides. These results are consistent with those of an Italian case-control study published in 2001 [
A meta-analysis of 13 case-control studies and 10 cohort studies found a significant association between high coffee consumption and CHD in the case-control studies, but no significant association between daily coffee consumption and CHD on long-term follow-up of cohort studies [
To our knowledge the present study is the first cohort study to specifically investigate the effects of Italian-style coffee on risk of CHD in a healthy population. Most previous studies have been concerned with filtered or boiled coffee, which are the most popular preparation methods in the USA and Scandinavian countries, respectively [
A randomized trial on the effect of Italian-style coffee on serum cholesterol in young men did not show any alteration of the cholesterol or lipoprotein profile [
A standard cup (30 ml) of espresso contains about 100 mg of caffeine [
Caffeine is metabolized by cytochrome P4501A2 (CYP1A2) in the liver. Individuals who are homozygous for the CYP1A2*1A allele are rapid caffeine metabolizers, while carriers of CYP1A2*1F are slow metabolizers. In a case-control study, high coffee intake was associated with increased risk of non-fatal myocardial infarction only in individuals with the CYP1A2*1F allele [
Other studies have found a direct relation between homocysteine levels and filtered and unfiltered coffee intake [
The detrimental effect of espresso coffee suggested by our study findings could, at least in part, be due to its caffeine content and rapid consumption, which results in a high peak plasma concentration of caffeine. By contrast, with filtered coffee, the caffeine dose is similar but it is diluted in 140–200 ml of liquid and the beverage is drunk over several minutes. The detrimental effect of espresso coffee is also likely to vary with CYP1A2 genotype.
Although smoking is an established risk factor for cardiovascular disease, the association of coffee with CHD remained significant when adjusted for smoking status and intensity. We analyzed the non-smoker sub-cohort (46.5%) and found that coffee consumption was significantly associated with CHD: HRs: 1, 1.21 (95% CI 0.83,1.75); 1.47 (95% CI 1.04,2.09); and 1.70 (95% CI 1.15,2.51); P trend = 0.002 (data not shown). These findings indicate that the coffee-CHD association cannot be explained by smoking.
A limitation of the study is that coffee consumption and CHD risk factors were only assessed at baseline. We could not analyze the effect of decaffeinated coffee because this information was not available for the Ragusa and Napoli centers, and decaffeinated consumption was very low in the other centers. Furthermore we have no information about other major sources of caffeine, family history of CHD and hypertension, or sleep patterns or other psychological factors, so we cannot exclude some residual confounding by these variables not included in the multivariate model. Study subjects were volunteers recruited mainly among blood donors, women attending screening programs and their spouses, therefore they may not be fully representative of general population. Moreover, plasma lipid profile was available only for a random subsample which may not be representative of the cohort as a whole. Strengths of our study are its prospective design, and the fact that Italian coffee drinkers almost always drink coffee prepared by the espresso or mocha methods, thereby reducing confounding due to variation in preparation method.
To conclude, we have found that consumption of more than 2 cups/day of Italian-style coffee is associated with significantly increased risk of CHD, but found no evidence that Italian-style coffee had any effect on plasma cholesterol or triglycerides. Further studies are required to determine whether drinking Italian-style coffee is causally associated with increased risk of CHD (and the association does arise, for example, because coffee consumption is marker of a lifestyle that increases that increases CHD risk); and also whether the risk is influenced by CYP1A2 genotype.
We thank A. Evangelista and D. Del Sette for technical support, and Don Ward for help with the English.