Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Caffeine intake reduces incident atrial fibrillation at a population level

Caffeine intake reduces incident atrial fibrillation at a population level Abstract Background The general belief is that caffeine increases the risk of hyperkinetic arrhythmias, including atrial fibrillation. The aim of this study is to investigate the effect of chronic caffeine intake on incident atrial fibrillation in general population. Design and methods A population cohort of 1475 unselected men and women observed for 12 years and left free to intake food or beverages containing caffeine was studied. Subjects were stratified into tertiles of caffeine intake both in the whole cohort and after genotyping for the –163C > A polymorphism of the CYP1A2 gene, regulating caffeine metabolism. Results In the whole cohort, the 12-year incidence of atrial fibrillation was significantly lower in the third (2.2%) than in the first (10.2%) or second (5.7%) tertile of caffeine intake (P < 0.001). The same trend was observed in all genotypes; the apparently steeper reduction of atrial fibrillation in slow caffeine metabolisers found at univariate analysis was proved wrong by multivariate Cox analysis. Age, chronic pulmonary disease, history of heart failure and of coronary artery disease, and systolic blood pressure − but not the genotype or the caffeine × CYP1A2 interaction term − were significant confounders of the association between incident atrial fibrillation and being in the third tertile of caffeine intake (hazard ratio 0.249, 95% confidence intervals 0.161–0.458, P < 0.01). Conclusions A higher caffeine intake (>165 mmol/day or > 320 mg/day) is associated with a lower incidence of atrial fibrillation in the 12-year epidemiological prospective setting based on the general population. Arrhythmias, atrial fibrillation, epidemiology, caffeine, genetics, CYP1A2, risk factors Introduction Due to its stimulating effects,1 the general belief is that caffeine intake is associated with hyperkinetic dysrhythmias,2 and individuals prone to tachycardia are usually discouraged from consuming caffeine-containing food or beverages. Nevertheless, this belief is not supported by experimental data. Caffeine exerts a chronotropic and bathmotropic effect,2 and has sometimes been associated with hyperkinetic arrhythmias.2,3 However, atrial fibrillation (AF) depends on reciprocating mechanisms rather than on increased bathmotropism.4 Some authors and current guidelines suggest that no clear association exists between AF and caffeine intake;5–8 others believe caffeine can trigger AF.9 The topic is of paramount importance as AF is a major cardiovascular risk factor10 and substances containing caffeine (such as coffee, tea, chocolate, cola and energy drinks) are diffused worldwide. Epidemiology has the means to find the association, if any, between AF and caffeine intake in the real world. In doing this, it is important to take into account the fate of caffeine in the human body, which is largely regulated by the –163C > A polymorphism of the CYP1A2 gene, codifying for an enzymatic protein metabolising caffeine.1 This polymorphism cannot be disregarded, because caffeine effects can be very different in the so-called ‘fast metabolisers’ (AA homozygous) and ‘slow metabolisers’ (carrying the C allele).11,12 This study is aimed at evaluating the effects of caffeine consumption on incident AF in a cohort of unselected men and women from the general population across the –163C > A polymorphism of CYP1A2 gene. Methods General protocol The present analysis is based on unselected men and women living in an area of about 550 km2 in northeast Italy and sharing a homogeneous lifestyle, randomly taken from the adult general population in the frame of an epidemiological study whose protocol has been described elsewhere.11,13 All individuals aged 18 years and over residing in the municipalities of Torrebelvicino and Valli del Pasubio were identified through the register office and were invited by letter and then by phone call to take part in the study, irrespective of any personal characteristic. The 1475 who accepted (73%), aged 60.0 ± 16.7 years (range 19.4–93.9), constituted the population-based cohort object of the present study. Recruitment began in 1999 and ended in 2003. Participants and non-participants did not differ as to demography (data not shown). All subjects underwent a medical examination with a cardiovascular and neuropsychological assessment, anthropometric measurements, blood test and an anamnestic questionnaire; a dietary diary was compiled in the week following the visit. Daily caffeine intake was calculated from the formula: caffeinemmol/day =coffeecups/day × 41.2 + teacups/day × 21.4 + coladrinks/day ×8.2 + chocolateportions/day × 8.1; and ethanol intake from: ethanolmmol/day = (wineml/day × 0.12 + beerml/day ×0.05 + liquorsml/day × 0.42 + aperitifsml/day × 0.11) × 0.85× 0.217 (0.85 is the density of ethanol, 0.217 the conversion factor from mg to mmol). Ethics The investigation, conforming to the Declaration of Helsinki, was approved by the ethics committees of the University of Padua, and of the 4th Local Health Unit of the Veneto region. Each subject gave and signed informed consent including treatment of genetic data. Follow-up Vital status and events were monitored for 12 years. Taking into account mortality, the follow-up was 5.6 ± 0.1 years (median 5.3, range 0.002–12 years). Based on World Health Organization International Statistical Classification of Diseases version 10, the incidence of AF was obtained from register offices and was double checked by referring to hospitals, retirement homes or physicians’ files considering subjects diagnosed by means of the codes I48.0, I48.1, I48.2 or I48.91. Cerebrovascular ischaemic events were assessed on the basis of codes I63, I64 or I65. Genotyping At screening, 6 ml of blood were collected in ethylenediamine tetraacetic acid tubes. DNA was extracted using MagNA Pure 96 DNA and viral NA small volume and large volume kits (Roche Diagnostics GmbH, Penzberg, Germany). Primers and probes for specific allelic discrimination analysis of CYP1A2 polymorphism were included in the polymerase chain reaction (PCR) assay. The forward primer was 5′–TTT CCA gCT CTC AgA TTC TgT gAT, the reverse primer was 5′–ggA TAC CAg AAA gAC TAA gCT CCA TC; CYP1A2*1F probe 5′–6FAM-TCT gTg ggC ACA ggA CgC ATg g, CYP1A2*1A probe 5′–HEX-CTC TgT ggg CCC Agg ACg CAT, as described by single nucleotide polymorphism database reference number (rs762551). Purified DNA was amplified in a real-time PCR reaction in a LightCycler 480 with Gene Scanning software version 1.5.1 (Roche). Positive controls were included in each run, together with a negative control containing no DNA template. TaqMan reactions were thermocycled as follows: 95℃ in pre-incubation, 45 cycles at 95℃ for primer-dependent amplification and 66.5℃ for annealing. Statistics Power analysis showed that 334 subjects per cell were sufficient to show effects (power 0.90, test level 0.10 for β error and 0.05 for α error), assuming a putative difference of 6% in AF incidence between the highest and lowest caffeine consumers. This difference was chosen a priori based on preliminary tests of our laboratory, as no data on the effects of caffeine intake on incident AF in the general population exist. Our 1475 subjects also appeared adequate after stratification into tertiles. Linearity assumption of continuous variables was ascertained by the residuals method and normality assumption by the Kolmogorov–Smirnov one-sample test. Continuous variables were expressed as mean ± standard deviation and compared with analysis of variance. Variables putatively not independent from each other were logarithmised. Categorical variables were compared with the χ2 test. A Cox proportional hazard model was used to find the variables having a prognostic role on AF incidence (<0.10 to enter and remove) and to calculate the hazard ratios (HRs) with 95% confidence interval (CI). An exploratory analysis of the full model demonstrated that age, sex, ethanol intake, smoking, blood pressure, heart rate, New York Heart Association class and history of heart failure, coronary or cerebral artery disease, chronic pulmonary disease, body mass index and width of the P wave at surface ECG were potential predictors of incident AF. They were therefore used as covariables in the Cox models, together with daily caffeine intake. In sensitivity analyses, physical activity was also added, and subjects with cardiovascular disease at the initial screening were excluded from the multivariate model. As the –163C > A polymorphism of the CYP1A2 gene controls the effects of caffeine,1 this polymorphism was introduced as a covariable in multiple analyses, as well as (in a separate model) the interaction term between caffeine intake and genotype. According to the current literature, subjects carrying the C allele (C carriers) and those carrying the A allele (A carriers) were also considered together, respectively. Results General statistics Table 1 shows the general characteristics of the population cohort also after stratification by tertiles of increasing caffeine consumption; 165 mmol/L (320 mg/dl) was the cut-off between the second and third tertile. Subjects in the highest tertile of caffeine intake were younger and healthier than those in the first tertile; to avoid any possible confounding effect of this difference in risk factors for AF, multivariable analyses were adjusted for age and comorbidities (see the Statistics section for the full list of covariables used). Twenty-nine subjects were abstaining from caffeine. The other 1446 consumed 240.3 ± 116.4 mg/day of caffeine (range 0.03–718.6, coefficient of variation 0.48), coming from coffee for 89.1%, from tea for 10.2% and from other sources for 0.7%; smokers (more than one cigarette/day) were 15.5% and drinkers (ethanol > 0 mmol/L) 70.9%. Table 1. General characteristics of the study population, also stratified by tertiles of caffeine intake. Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Mean ± standard deviation (95% confidence intervals in brackets) and rate (percent). * P < 0.01 vs. first tertile. LDL: low-density lipoprotein; NYHA: New York Heart Association; CAD: coronary artery disease; ECG: electrocardiographic. Open in new tab Table 1. General characteristics of the study population, also stratified by tertiles of caffeine intake. Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Mean ± standard deviation (95% confidence intervals in brackets) and rate (percent). * P < 0.01 vs. first tertile. LDL: low-density lipoprotein; NYHA: New York Heart Association; CAD: coronary artery disease; ECG: electrocardiographic. Open in new tab Univariate analysis During the follow-up there were 89 new cases of AF (6.0%), i.e. 50 (10.2%) in the first tertile of caffeine intake, 28 (5.7%) in the second and 11 (2.2%) in the third. The cumulative incidence of AF during the follow-up in the tertiles of caffeine intake is shown in Figure 1; a significant trend towards the reduction of incident AF with increasing caffeine intake was detected, with the lowest values corresponding to the highest intake. Figure 1. Open in new tabDownload slide Cumulative incidence of atrial fibrillation in the three tertiles of caffeine intake. Table 2 shows the trend of incidence of AF in AA, CC, A carrier and C carrier subjects. A significantly lower incidence was always detected in the third than in the first or second tertile. The reduction was present in all the genotypes, although mildly steeper in the CC homozygous or in the C carriers than in the others. When the difference between the third and the first tertile was considered, the difference in this difference between CC and AA was 6% in favour of the former (χ2 = 3.84, P = 0.05), and that between C carriers and A carriers was 4.7% in favour of the former (χ2 = 11.36, P = 0.001). On the contrary, no significant difference was detected either between CC and C carriers or between AA and A carriers. Table 2. Crude incidence of atrial fibrillation according to –163C > A polymorphism of CYP1A2 gene, after dividing each genotype into tertiles of caffeine intake. Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Open in new tab Table 2. Crude incidence of atrial fibrillation according to –163C > A polymorphism of CYP1A2 gene, after dividing each genotype into tertiles of caffeine intake. Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Open in new tab The incidence of ischaemic stroke decreased with increasing caffeine intake (8.5% in the first tertile, 6.5% in the second tertile, 3.0% in the third tertile; P for trend < 0.001). Multivariate analysis In the multivariate Cox model (Table 3), being in the highest tertile of caffeine intake had a protective effect against incident AF in comparison to the reference represented by the first tertile, despite significant correction for age, systolic blood pressure, history of heart failure and of coronary artery disease, and chronic pulmonary disease; HRs are shown in Figure 2. In different analyses, the –163C > A polymorphism and the caffeine × CYP1A2 interaction term were also rejected, both considering the three genotypes separately and using the categorisation AA = 1 and C carrier = 0 (estimate −0.014, P = 0.9). Table 3. Cox model for incident atrial fibrillation among 1475 unselected subjects from the general population according to caffeine intake (left panel) and in subjects without any cardiovascular (CV) disease at initial screening. Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 † Interaction term is that between caffeine intake and CYP1A2 –163C > A genotype. ‡Statistically significant. The model is adjusted for the independent variables listed in the table, logarithmised when proper (*). The first tertile of caffeine intake was considered as the reference. Open in new tab Table 3. Cox model for incident atrial fibrillation among 1475 unselected subjects from the general population according to caffeine intake (left panel) and in subjects without any cardiovascular (CV) disease at initial screening. Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 † Interaction term is that between caffeine intake and CYP1A2 –163C > A genotype. ‡Statistically significant. The model is adjusted for the independent variables listed in the table, logarithmised when proper (*). The first tertile of caffeine intake was considered as the reference. Open in new tab Figure 2. Open in new tabDownload slide Hazard ratio (with 95% confidence intervals) of incident atrial fibrillation in the three tertiles of caffeine intake (first tertile as reference). In two sensitivity analyses, physical activity (sedentary, moderately active, very active and trained) was rejected from the risk equation (estimate 0.098, standard error 0.171, 95% CI –0.238 to 0.434, P = 0.6). Selecting subjects free from cardiovascular disease at the initial screening did not alter the multivariate model (Table 3, right panels). Discussion Background Exploring the effects of caffeine in a series of papers,11,14–16 we observed an increasing interest about the possible arrhythmogenic action of caffeine. Unfortunately, the studies available in the literature on this topic are limited in number and contradictory in results. The papers showing a direct caffeine → AF association are anecdotal.3,17 Other studies emphasise the absence of any effect of caffeine on incident AF in non-population-based8 and population-based epidemiological setting18 and in short-term experiments on little numbers of subjects.19 A neutral role of caffeine emerges from some data pooling6 or meta-analysis.5 This study The data shown here demonstrate that, in a population-based epidemiological setting, the higher the intake of caffeine in normal life, the lower the incidence of AF. This is clarified both in univariate and multivariate statistics. In univariate analysis, the 12-year incidence of AF was significantly lower in the third tertile of caffeine intake (>165 mmol/day, corresponding to four cups of espresso coffee)14 than in the first and the second tertiles (–78% and –62%, respectively). In Cox analysis adjusted for confounders, being in the third tertile of caffeine intake significantly reduced the risk of incident AF with a HR of 0.25 independent of the adverse effect of age, systolic blood pressure, chronic pulmonary disease and history of heart failure, coronary artery disease and cerebrovascular disease. To our knowledge, no population-based epidemiological study has demonstrated up to now this inverse association, although other prospective or pooled data studies found similar results.6,8 Possible mechanisms Due to the epidemiologic nature of our study and to the lack of data in the literature, the reasons for this inverse association can only be the object of speculation. The possibility that caffeine directly favours the AF → sinus rhythm conversion has been suggested;20 in this model, like in our study, the beneficial effect of caffeine was partly contrasted by systolic hypertension. Furthermore, an electrophysiological study demonstrated that the cumulative window of vulnerability is significantly lower in subjects taking caffeine, leading to a reduced propensity to AF.21 Finally, stimulation of A2B receptors plays a role in the pathogenesis of cardiac fibrosis, inhibits cardiac fibroblast production of collagen in vitro and diminishes myocardial remodelling.22–24 These caffeine-induced changes in cardiac homeostasis, when translated to the general population, could play a role in shifting subjects from the category prone to AF to that resistant to AF, reducing the final incidence of AF. Unfortunately, these data are insufficient and can only serve as a guide for future experiments. Role of genetics Studies on caffeine must take into account the –163C > A polymorphism of the CYP1A2 gene,12 as this polymorphism controls the in vivo effects of caffeine and of its active metabolites.1 Actually, the effect of caffeine on cardiovascular, metabolic and cognitive patterns strictly depend on genotype.11,12,16 Univariate analysis (Table 2) seemed to suggest a protective role of the C allele, but a significant trend towards AF reduction was detected in all genotypes. In Cox models, the genotype or the caffeine × CYP1A2 interaction term were not accepted in the equation of risk nor changed the Cox model. The multivariate analysis adjusted for confounders therefore belies the simplistic and misleading univariate approach and indicates that – contrary to other effects of caffeine – the action of caffeine on AF, whatever it is, is independent of genetic control. Strengths of the study The strengths of our study are that it was population-based, that it took into consideration non-selected men and women deriving from a homogenous group of individuals living in comparable conditions in a limited geographical area, that it was performed by a sole research staff and that it took into account the –163C > A polymorphism of the CYP1A2 gene. The ecogenetic context was taken into consideration,25 not excluding a priori any role for the genetic pattern. Another strength is that a double check of caffeine intake was performed, using both a questionnaire and a dietary diary. Finally, a further confirmation of our results comes from the reduction of ischaemic cerebral events, a well-known consequence of AF, with increasing caffeine intake. Limitations of the study A limitation is represented by the fact that caffeine intake was assessed only at the initial screening; eventual changes in the dietary pattern compared to the baseline were not considered. Another limitation is that only one ECG rescreening was planned in the protocol during the follow-up, so that asymptomatic paroxysmal AF not leading to hospitalisation or to the attention of general practitioners might not have been detected. On the other hand, we followed in these aspects the protocols of the Danish Diet, Cancer, and Health Study8 and of the Framingham Study.18 In a population-based study, asymptomatic paroxysmal AF is, through force of circumstances, underestimated. Furthermore, being older and having more comorbidities, subjects in the third tertile of caffeine intake could have been under more medical supervision. The best way we found to avoid any bias was to consider tertiles rather than arbitrary classes. Unfortunately, the three tertiles were not homogeneous but we are confident that multivariate analysis was sufficient to avoid any interference. Finally, caffeine intake in this cohort was quite homogenous, a typical trait of the Italian general population. This finding could have reduced the apparent effect of different caffeine consumption, but – as demonstrated – it was not sufficient to nullify the beneficial effect of caffeine on incident AF. Conclusions A higher caffeine intake is associated with a lower incidence of AF in a 12-year epidemiological prospective setting based on the general population also after adjusting for possible confounders, and is not influenced by the slow or fast caffeine metaboliser genotype. Chronic cardiac and pulmonary disease and high systolic blood pressure26 contrast but do not nullify the beneficial effect of caffeine. Author contribution EC, VT, FA and PP contributed to the conception and design. EC, FA, FG, AM, MM, ED, MB and PS contributed to the acquisition, analysis, or interpretation of data for the work. EC, FA and PS drafted the manuscript. EC, VT, FA, FG, AM, MM, ED, MB and PP critically revised it. All authors gave final approval and agree to be accountable for all aspects of the work ensuring integrity and accuracy. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: this work was supported by Veneto region programmes RSF 811-98 and 178-04. References 1 Cappelletti S , Piacentino D, Sani Get al. Caffeine: cognitive and physical performance enhancer or psychoactive drug? Curr Neuropharmacol 2015 ; 13 : 71 – 88 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Joung B , Lin SF, Chen Zet al. Mechanisms of sinoatrial node dysfunction in a canine model of pacing-induced atrial fibrillation . Heart Rhythm 2010 ; 7 : 88 – 95 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Di Rocco JR , During A, Morelli PJet al. Atrial fibrillation in healthy adolescents after highly caffeinated beverage consumption: two case reports . J Med Case Rep 2011 ; 5 : 18 – 18 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Voigt N , Heijman J, Wang Qet al. Cellular and molecular mechanisms of atrial arrhythmogenesis in patients with paroxysmal atrial fibrillation . Circulation 2014 ; 129 : 145 – 156 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Caldeira D , Martins C, Alves LBet al. Caffeine does not increase the risk of atrial fibrillation: a systematic review and meta-analysis of observational studies . Heart 2013 ; 99 : 1383 – 1389 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Cheng M , Hu Z, Lu Xet al. Caffeine intake and atrial fibrillation incidence: dose response meta-analysis of prospective cohort studies . Can J Cardiol 2014 ; 30 : 448 – 454 . Google Scholar Crossref Search ADS PubMed WorldCat 7 EHRA Scientific Committee Task Force . European Heart Rhythm Association (EHRA)/European Association of Cardiovascular Prevention and Rehabilitation (EACPR) position paper on how to prevent atrial fibrillation endorsed by the Heart Rhythm Society (HRS) and Asia Pacific Heart Rhythm Society (APHRS) . Eur J Prev Cardiol 2017 ; 24 : 4 – 40 . Google Scholar OpenURL Placeholder Text WorldCat 8 Frost L , Vestergaard P. Caffeine and risk of atrial fibrillation or flutter: the Danish Diet, Cancer, and Health Study . Am J Clin Nutr 2005 ; 81 : 578 – 582 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Hansson A , Madsen-Härdig B, Bertil Olsson S. Arrhythmia-provoking factors and symptoms at the onset of paroxysmal atrial fibrillation: a study based on interviews with 100 patients seeking hospital assistance . BMC Cardiovasc Dis 2004 ; 4 : 13 – 13 . Google Scholar Crossref Search ADS WorldCat 10 Ruddox V , Sandven I, Munkhaugen Jet al. Atrial fibrillation and the risk for myocardial infarction, all-cause mortality and heart failure: a systematic review and meta-analysis . Eur J Prev Cardiol 2017 ; 24 : 1555 – 1566 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Casiglia E , Tikhonoff V, Albertini Fet al. Effects of caffeine and coffee on incident heart failure in general population. Role of the CYP1A2 –163C > A polymorphism . WJRR 2017 ; 5 : 42 – 48 . Google Scholar Crossref Search ADS WorldCat 12 Cornelis MC , El-Sohemy A, Kabagambe EKet al. Coffee, CYP1A2 genotype, and risk of myocardial infarction . JAMA 2006 ; 295 : 1135 – 1141 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Tikhonoff V , Casiglia E, Guidotti Fet al. Body fat and the cognitive pattern: a population-based study . Obesity 2015 ; 23 : 1502 – 1510 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Pizziol A , Tikhonoff V, Paleari CDet al. Effects of caffeine on glucose tolerance: a placebo-controlled study . Eur J Clin Nutr 1998 ; 52 : 846 – 849 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Casiglia E , Paleari CD, Petucco Set al. Haemodynamic effects of coffee and purified caffeine in normal volunteers: a placebo-controlled clinical study . J Hum Hypertens 1992 ; 6 : 95 – 99 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 16 Palatini P . Coffee consumption and risk of type 2 diabetes . Diabetologia 2015 ; 58 : 199 – 200 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Thyagarajan B , Alagusundaramoorthy SS, Agrawal A. Atrial fibrillation due to over the counter stimulant drugs in a young adult . J Clin Diagn Res 2015 ; 9 : OD05 – 07 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 18 Shen J , Johnson VM, Sullivan LMet al. Dietary factors and incident atrial fibrillation: the Framingham Heart Study . Am J Clin Nutr 2011 ; 93 : 261 – 266 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Lemery R , Pecarskie A, Bernick Jet al. A prospective placebo controlled randomized study of caffeine in patients with supraventricular tachycardia undergoing electrophysiologic testing . J Cardiovasc Electrophysiol 2015 ; 26 : 1 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 20 Mattioli AV , Farinetti A, Miloro Cet al. Influence of coffee and caffeine consumption on atrial fibrillation in hypertensive patients . Nutr Metab Cardiovasc Dis 2011 ; 21 : 412 – 417 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Rashid A , Hines M, Scherlag BJet al. The effects of caffeine on the inducibility of atrial fibrillation . J Electrocardiol 2006 ; 39 : 421 – 425 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Anty R , Marjoux S, Iannelli Aet al. Regular coffee but not espresso drinking is protective against fibrosis in a cohort mainly composed of morbidly obese European women with NAFLD undergoing bariatric surgery . J Hepatol 2012 ; 57 : 1090 – 1096 . Google Scholar Crossref Search ADS PubMed WorldCat 23 Dubey RK , Gillespie DG, Jackson EK. Adenosine inhibits collagen and protein synthesis in cardiac fibroblasts: role of A2B receptors . Hypertension 1998 ; 31 : 943 – 948 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Wakeno M , Minamino T, Seguchi Oet al. Long-term stimulation of adenosine A2b receptors begun after myocardial infarction prevents cardiac remodeling in rats . Circulation 2006 ; 114 : 1923 – 1932 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Stolarz K , Staessen JA, Kawecka-Jaszcz Ket al. Genetic variation in CYP11B2 and AT1R influences heart rate variability conditional on sodium excretion . Hypertension 2004 ; 44 : 156 – 162 . Google Scholar Crossref Search ADS PubMed WorldCat 26 Mourtzinis G , Schiöler L, Kahan Tet al. Antihypertensive control and new-onset atrial fibrillation: results from the Swedish Primary Care Cardiovascular Database (SPCCD) . Eur J Prev Cardiol 2017 ; 24 : 1206 – 1211 . Google Scholar Crossref Search ADS PubMed WorldCat © The European Society of Cardiology 2018 This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © The European Society of Cardiology 2018 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Journal of Preventive Cardiology Oxford University Press

Loading next page...
 
/lp/oxford-university-press/caffeine-intake-reduces-incident-atrial-fibrillation-at-a-population-b2pTmlxrUR

References (26)

Publisher
Oxford University Press
Copyright
Copyright © 2022 European Society of Cardiology
ISSN
2047-4873
eISSN
2047-4881
DOI
10.1177/2047487318772945
Publisher site
See Article on Publisher Site

Abstract

Abstract Background The general belief is that caffeine increases the risk of hyperkinetic arrhythmias, including atrial fibrillation. The aim of this study is to investigate the effect of chronic caffeine intake on incident atrial fibrillation in general population. Design and methods A population cohort of 1475 unselected men and women observed for 12 years and left free to intake food or beverages containing caffeine was studied. Subjects were stratified into tertiles of caffeine intake both in the whole cohort and after genotyping for the –163C > A polymorphism of the CYP1A2 gene, regulating caffeine metabolism. Results In the whole cohort, the 12-year incidence of atrial fibrillation was significantly lower in the third (2.2%) than in the first (10.2%) or second (5.7%) tertile of caffeine intake (P < 0.001). The same trend was observed in all genotypes; the apparently steeper reduction of atrial fibrillation in slow caffeine metabolisers found at univariate analysis was proved wrong by multivariate Cox analysis. Age, chronic pulmonary disease, history of heart failure and of coronary artery disease, and systolic blood pressure − but not the genotype or the caffeine × CYP1A2 interaction term − were significant confounders of the association between incident atrial fibrillation and being in the third tertile of caffeine intake (hazard ratio 0.249, 95% confidence intervals 0.161–0.458, P < 0.01). Conclusions A higher caffeine intake (>165 mmol/day or > 320 mg/day) is associated with a lower incidence of atrial fibrillation in the 12-year epidemiological prospective setting based on the general population. Arrhythmias, atrial fibrillation, epidemiology, caffeine, genetics, CYP1A2, risk factors Introduction Due to its stimulating effects,1 the general belief is that caffeine intake is associated with hyperkinetic dysrhythmias,2 and individuals prone to tachycardia are usually discouraged from consuming caffeine-containing food or beverages. Nevertheless, this belief is not supported by experimental data. Caffeine exerts a chronotropic and bathmotropic effect,2 and has sometimes been associated with hyperkinetic arrhythmias.2,3 However, atrial fibrillation (AF) depends on reciprocating mechanisms rather than on increased bathmotropism.4 Some authors and current guidelines suggest that no clear association exists between AF and caffeine intake;5–8 others believe caffeine can trigger AF.9 The topic is of paramount importance as AF is a major cardiovascular risk factor10 and substances containing caffeine (such as coffee, tea, chocolate, cola and energy drinks) are diffused worldwide. Epidemiology has the means to find the association, if any, between AF and caffeine intake in the real world. In doing this, it is important to take into account the fate of caffeine in the human body, which is largely regulated by the –163C > A polymorphism of the CYP1A2 gene, codifying for an enzymatic protein metabolising caffeine.1 This polymorphism cannot be disregarded, because caffeine effects can be very different in the so-called ‘fast metabolisers’ (AA homozygous) and ‘slow metabolisers’ (carrying the C allele).11,12 This study is aimed at evaluating the effects of caffeine consumption on incident AF in a cohort of unselected men and women from the general population across the –163C > A polymorphism of CYP1A2 gene. Methods General protocol The present analysis is based on unselected men and women living in an area of about 550 km2 in northeast Italy and sharing a homogeneous lifestyle, randomly taken from the adult general population in the frame of an epidemiological study whose protocol has been described elsewhere.11,13 All individuals aged 18 years and over residing in the municipalities of Torrebelvicino and Valli del Pasubio were identified through the register office and were invited by letter and then by phone call to take part in the study, irrespective of any personal characteristic. The 1475 who accepted (73%), aged 60.0 ± 16.7 years (range 19.4–93.9), constituted the population-based cohort object of the present study. Recruitment began in 1999 and ended in 2003. Participants and non-participants did not differ as to demography (data not shown). All subjects underwent a medical examination with a cardiovascular and neuropsychological assessment, anthropometric measurements, blood test and an anamnestic questionnaire; a dietary diary was compiled in the week following the visit. Daily caffeine intake was calculated from the formula: caffeinemmol/day =coffeecups/day × 41.2 + teacups/day × 21.4 + coladrinks/day ×8.2 + chocolateportions/day × 8.1; and ethanol intake from: ethanolmmol/day = (wineml/day × 0.12 + beerml/day ×0.05 + liquorsml/day × 0.42 + aperitifsml/day × 0.11) × 0.85× 0.217 (0.85 is the density of ethanol, 0.217 the conversion factor from mg to mmol). Ethics The investigation, conforming to the Declaration of Helsinki, was approved by the ethics committees of the University of Padua, and of the 4th Local Health Unit of the Veneto region. Each subject gave and signed informed consent including treatment of genetic data. Follow-up Vital status and events were monitored for 12 years. Taking into account mortality, the follow-up was 5.6 ± 0.1 years (median 5.3, range 0.002–12 years). Based on World Health Organization International Statistical Classification of Diseases version 10, the incidence of AF was obtained from register offices and was double checked by referring to hospitals, retirement homes or physicians’ files considering subjects diagnosed by means of the codes I48.0, I48.1, I48.2 or I48.91. Cerebrovascular ischaemic events were assessed on the basis of codes I63, I64 or I65. Genotyping At screening, 6 ml of blood were collected in ethylenediamine tetraacetic acid tubes. DNA was extracted using MagNA Pure 96 DNA and viral NA small volume and large volume kits (Roche Diagnostics GmbH, Penzberg, Germany). Primers and probes for specific allelic discrimination analysis of CYP1A2 polymorphism were included in the polymerase chain reaction (PCR) assay. The forward primer was 5′–TTT CCA gCT CTC AgA TTC TgT gAT, the reverse primer was 5′–ggA TAC CAg AAA gAC TAA gCT CCA TC; CYP1A2*1F probe 5′–6FAM-TCT gTg ggC ACA ggA CgC ATg g, CYP1A2*1A probe 5′–HEX-CTC TgT ggg CCC Agg ACg CAT, as described by single nucleotide polymorphism database reference number (rs762551). Purified DNA was amplified in a real-time PCR reaction in a LightCycler 480 with Gene Scanning software version 1.5.1 (Roche). Positive controls were included in each run, together with a negative control containing no DNA template. TaqMan reactions were thermocycled as follows: 95℃ in pre-incubation, 45 cycles at 95℃ for primer-dependent amplification and 66.5℃ for annealing. Statistics Power analysis showed that 334 subjects per cell were sufficient to show effects (power 0.90, test level 0.10 for β error and 0.05 for α error), assuming a putative difference of 6% in AF incidence between the highest and lowest caffeine consumers. This difference was chosen a priori based on preliminary tests of our laboratory, as no data on the effects of caffeine intake on incident AF in the general population exist. Our 1475 subjects also appeared adequate after stratification into tertiles. Linearity assumption of continuous variables was ascertained by the residuals method and normality assumption by the Kolmogorov–Smirnov one-sample test. Continuous variables were expressed as mean ± standard deviation and compared with analysis of variance. Variables putatively not independent from each other were logarithmised. Categorical variables were compared with the χ2 test. A Cox proportional hazard model was used to find the variables having a prognostic role on AF incidence (<0.10 to enter and remove) and to calculate the hazard ratios (HRs) with 95% confidence interval (CI). An exploratory analysis of the full model demonstrated that age, sex, ethanol intake, smoking, blood pressure, heart rate, New York Heart Association class and history of heart failure, coronary or cerebral artery disease, chronic pulmonary disease, body mass index and width of the P wave at surface ECG were potential predictors of incident AF. They were therefore used as covariables in the Cox models, together with daily caffeine intake. In sensitivity analyses, physical activity was also added, and subjects with cardiovascular disease at the initial screening were excluded from the multivariate model. As the –163C > A polymorphism of the CYP1A2 gene controls the effects of caffeine,1 this polymorphism was introduced as a covariable in multiple analyses, as well as (in a separate model) the interaction term between caffeine intake and genotype. According to the current literature, subjects carrying the C allele (C carriers) and those carrying the A allele (A carriers) were also considered together, respectively. Results General statistics Table 1 shows the general characteristics of the population cohort also after stratification by tertiles of increasing caffeine consumption; 165 mmol/L (320 mg/dl) was the cut-off between the second and third tertile. Subjects in the highest tertile of caffeine intake were younger and healthier than those in the first tertile; to avoid any possible confounding effect of this difference in risk factors for AF, multivariable analyses were adjusted for age and comorbidities (see the Statistics section for the full list of covariables used). Twenty-nine subjects were abstaining from caffeine. The other 1446 consumed 240.3 ± 116.4 mg/day of caffeine (range 0.03–718.6, coefficient of variation 0.48), coming from coffee for 89.1%, from tea for 10.2% and from other sources for 0.7%; smokers (more than one cigarette/day) were 15.5% and drinkers (ethanol > 0 mmol/L) 70.9%. Table 1. General characteristics of the study population, also stratified by tertiles of caffeine intake. Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Mean ± standard deviation (95% confidence intervals in brackets) and rate (percent). * P < 0.01 vs. first tertile. LDL: low-density lipoprotein; NYHA: New York Heart Association; CAD: coronary artery disease; ECG: electrocardiographic. Open in new tab Table 1. General characteristics of the study population, also stratified by tertiles of caffeine intake. Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Items . Whole cohort . Tertiles of caffeine intake . (n = 1475) . 1st (n = 491) . 2nd (n = 491) . 3rd (n = 492) . Caffeine intake (mmol/day) 123.8 ± 60.0 (120.8–127.0) 64.8 ± 32.2 (62.0–67.6) 118.3 ± 28.2 (115.8–120.8) 188.4 ± 35.7* (185.3–191.3) Age (years) 60.0 ± 16.7 (59.2–60.9) 62.8 ± 16.3 (61.3–64.2) 61.1 ± 17.2 (59.6–62.6) 56.1 ± 16.0* (54.7–57.6) Women (0)/men (1) 806/669 (54.6/43.4) 244/247 (49.7/50.3) 312/180 (63.4/36.6) 250/242 (50.8/49.2) Body mass index (kg/m2) 26.3 ± 4.3 (26.1–26.5) 26.6 ± 4.6 (26.2–27.0) 26.1 ± 4.1 (25.7–26.4) 26.2 ± 4.2* (25.8–26.5) Systolic blood pressure (mmHg) 156.1 ± 26.1 (154.7–157.4) 160.8 ± 25.9 (158.8–163.1) 158.0 ± 27.0 (155.6–160.4) 149.5 ± 23.9* (147.4–151.6) Diastolic blood pressure (mmHg) 88.3 ± 11.4 (87.7–88.9) 89.6 ± 11.9 (88.5–90.6) 88.7 ± 11.8 (87.6–89.7) 86.7 ± 10.4 (85.8–87.6) Heart rate (bpm) 69.4 ± 10.6 (68.9–69.9) 69.9 ± 11.0 (68.9–70.9) 70.3 ± 10.7 (69.4–71.3) 67.9 ± 10.0* (67.0–68.8) Serum triglycerides (mmol/L) 1.31 ± 0.81 (1.27–1.35) 1.35 ± 0.87 (1.27–1.43) 1.35 ± 0.87 (1.27–1.43) 1.24 ± 0.66* (1.18–1.30) LDL-cholesterol (mmol/L) 3.87 ± 0.99 (3.82–3.92) 3.81 ± 0.96 (3.72–3.89) 3.85 ± 1.03 (3.75–3.94) 3.94 ± 0.98* (3.86–4.03) Serum uric acid (µmol/L) 297.4 ± 83.3 (291.4–303.3) 309.3 ± 83.3 (303.3–315.2) 291.3 ± 83.3 (285.5–297.4) 291.4 ± 77.3* (285.5–397.4) Smoking (cigarettes/day) 1.6 ± 4.5 (1.3–1.8) 1.0 ± 3.3 (0.7–1.2) 1.0 ± 3.2 (0.7–1.2) 2.8 ± 6.1* (2.2–3.3) Ethanol intake (mmol/day) 58.6 ± 68.6 (55.1–62.1) 62.1 ± 7.2 (55.8–68.4) 52.8 ± 64.9 (46.9–58.4) 64.0 ± 69.0 (54.9–67.3) Pulmonary disease (0: no; 1: yes) 47 (3.2%) 24 (4.9%) 13 (2.6%) 10 (2.0%)* NYHA class 1.12 ± 0.45 (1.10–1.14) 1.15 ± 0.50 (1.11–1.20) 1.15 ± 0.51 (1.10–1.19) 1.06 ± 0.31 (1.04–1.09) History of heart failure (0: no; 1: yes) 38 (2.6%) 18 (3.7%) 14 (2.8%) 6 (1.2%)* History of CAD (0: no; 1: yes) 80 (5.4%) 44 (9.0) 25 (5.1%) 11 (2.2%) Width of ECG P wave (ms) 85.4 ± 15.2 (84.6–86.3) 85.2 ± 16.5 (83.6–86.9) 85.8 ± 14.4 (84.4–87.2) 85.2 ± 14.7 (83.8–86.7) Diabetes (0: no; 1: yes) 279 (18.9%) 129 (26.3%) 98 (19.9%) 52 (10.6%)* Mean ± standard deviation (95% confidence intervals in brackets) and rate (percent). * P < 0.01 vs. first tertile. LDL: low-density lipoprotein; NYHA: New York Heart Association; CAD: coronary artery disease; ECG: electrocardiographic. Open in new tab Univariate analysis During the follow-up there were 89 new cases of AF (6.0%), i.e. 50 (10.2%) in the first tertile of caffeine intake, 28 (5.7%) in the second and 11 (2.2%) in the third. The cumulative incidence of AF during the follow-up in the tertiles of caffeine intake is shown in Figure 1; a significant trend towards the reduction of incident AF with increasing caffeine intake was detected, with the lowest values corresponding to the highest intake. Figure 1. Open in new tabDownload slide Cumulative incidence of atrial fibrillation in the three tertiles of caffeine intake. Table 2 shows the trend of incidence of AF in AA, CC, A carrier and C carrier subjects. A significantly lower incidence was always detected in the third than in the first or second tertile. The reduction was present in all the genotypes, although mildly steeper in the CC homozygous or in the C carriers than in the others. When the difference between the third and the first tertile was considered, the difference in this difference between CC and AA was 6% in favour of the former (χ2 = 3.84, P = 0.05), and that between C carriers and A carriers was 4.7% in favour of the former (χ2 = 11.36, P = 0.001). On the contrary, no significant difference was detected either between CC and C carriers or between AA and A carriers. Table 2. Crude incidence of atrial fibrillation according to –163C > A polymorphism of CYP1A2 gene, after dividing each genotype into tertiles of caffeine intake. Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Open in new tab Table 2. Crude incidence of atrial fibrillation according to –163C > A polymorphism of CYP1A2 gene, after dividing each genotype into tertiles of caffeine intake. Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Tertiles of daily caffeine intake . −163C > A polymorphism of CYP1A2 gene . AA (n = 650) . AC (n = 639) . CC (n = 186) . A-carriers (n = 1289) . C-carriers (n = 825) . 12-Year incidence of atrial fibrillation  1st 23/217 (10.6%) 18/213 (8.4%) 9/62 (14.5%) 35/430 (8.1%) 29/275 (10.5%)  2nd 12/217 (5.5%) 13/213 (6.1%) 3/62 (4.8%) 26/430 (6.0%) 15/275 (5.4%)  3rd 8/216 (3.7%) 2/213 (0.9%) 1/62 (1.6%) 15/429 (3.5%) 3/275 (1.1%)  χ2 for trend 8.95 12.84 8.52 8.80 22.92  P for trend 0.011 0.0016 0.014 0.012 <0.001 Open in new tab The incidence of ischaemic stroke decreased with increasing caffeine intake (8.5% in the first tertile, 6.5% in the second tertile, 3.0% in the third tertile; P for trend < 0.001). Multivariate analysis In the multivariate Cox model (Table 3), being in the highest tertile of caffeine intake had a protective effect against incident AF in comparison to the reference represented by the first tertile, despite significant correction for age, systolic blood pressure, history of heart failure and of coronary artery disease, and chronic pulmonary disease; HRs are shown in Figure 2. In different analyses, the –163C > A polymorphism and the caffeine × CYP1A2 interaction term were also rejected, both considering the three genotypes separately and using the categorisation AA = 1 and C carrier = 0 (estimate −0.014, P = 0.9). Table 3. Cox model for incident atrial fibrillation among 1475 unselected subjects from the general population according to caffeine intake (left panel) and in subjects without any cardiovascular (CV) disease at initial screening. Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 † Interaction term is that between caffeine intake and CYP1A2 –163C > A genotype. ‡Statistically significant. The model is adjusted for the independent variables listed in the table, logarithmised when proper (*). The first tertile of caffeine intake was considered as the reference. Open in new tab Table 3. Cox model for incident atrial fibrillation among 1475 unselected subjects from the general population according to caffeine intake (left panel) and in subjects without any cardiovascular (CV) disease at initial screening. Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 Independent variables . Whole cohort . Subjects with no CV disease at screening . Estimate (standard error) . 95% confidence intervals . P value . Estimate (standard error) . 95% confidence intervals . P value . Main predictors  2nd tertile of caffeine intake −0.540 (0.352) −1.230–0.151 0.126 −0.562 (0.356) −1.261–0.136 0.115  3rd tertile of caffeine intake −1.647 (0.631) −2.882– –0.411 0.009‡ −1.578 (0.630) −2.812– –0.344 0.012‡ Significant covariables  in the whole cohort  Age* 3.922 (1.210) 1.551–6.293 0.001‡ 4.102 (1.234) 1.682–6.521 0.001‡  Chronic pulmonary disease 1.882 (0.432) 1.036–2.728 0.0001‡ 1.928 (0.436) 1.073–2.783 0.0001‡  History of heart failure 1.586 (0.495) 0.616–2.555 0.001‡ 1.715 (0.495) 0.744–2.685 0.001‡  Systolic blood pressure* 2.560 (1.237) 0.136–4.985 0.038‡ 2.106 (1.246) −0.337–4.549 0.091 Non significant covariables  Diastolic blood pressure* −2.704 (1.426) −5.498–0.090 0.058 −2.313 (1.445) −5.145–0.520 0.110  Heart rate* 0.514 (1.019) −1.484–2.512 0.614 0.545 (1.034) −1.481–2.571 0.598  Diabetes −0.343 (0.324) −0.978–0.291 0.289 −0.317 (0.324) −0.952–0.319 0.329  −163C > A of CYP1A2 gene −0.204 (0.294) −0.780–0.373 0.488 −0.174 (0.295) −0.753–0.405 0.556  Interaction term† 0.001 (0.001) −0.001–0.004 0.220 0.001 (0.001) −0.001–0.003 0.301  Sex 0.180 (0.345) −0.496–0.856 0.602 0.106 (0.350) −0.580–0.792 0.762  Smoking (cigarettes/day) −0.031 (0.040) −0.109–0.047 0.434 −0.028 (0.040) −0.106–0.050 0.483  Ethanol intake (mmol/day) 0.001 (0.005) −0.008–0.010 0.834 0.002 (0.005) −0.007–0.011 0.700  Body mass index (kg/m2)* 1.461 (1.071) −0.638–3.560 0.172 1.532 (1.077) −0.580–3.644 0.155  History of coronary disease 0.733 (0.398) −0.047–1.513 0.065 0.606 (0.414) −0.206–1.417 0.143  Width of ECG P wave 0.006 (0.009) −0.012–0.024 0.498 0.008 (0.009) −0.010–0.026 0.390  NYHA class −0.420 (0.293) −0.994–0.154 0.152 −0.421 (0.294) −0.997–0.155 0.152 † Interaction term is that between caffeine intake and CYP1A2 –163C > A genotype. ‡Statistically significant. The model is adjusted for the independent variables listed in the table, logarithmised when proper (*). The first tertile of caffeine intake was considered as the reference. Open in new tab Figure 2. Open in new tabDownload slide Hazard ratio (with 95% confidence intervals) of incident atrial fibrillation in the three tertiles of caffeine intake (first tertile as reference). In two sensitivity analyses, physical activity (sedentary, moderately active, very active and trained) was rejected from the risk equation (estimate 0.098, standard error 0.171, 95% CI –0.238 to 0.434, P = 0.6). Selecting subjects free from cardiovascular disease at the initial screening did not alter the multivariate model (Table 3, right panels). Discussion Background Exploring the effects of caffeine in a series of papers,11,14–16 we observed an increasing interest about the possible arrhythmogenic action of caffeine. Unfortunately, the studies available in the literature on this topic are limited in number and contradictory in results. The papers showing a direct caffeine → AF association are anecdotal.3,17 Other studies emphasise the absence of any effect of caffeine on incident AF in non-population-based8 and population-based epidemiological setting18 and in short-term experiments on little numbers of subjects.19 A neutral role of caffeine emerges from some data pooling6 or meta-analysis.5 This study The data shown here demonstrate that, in a population-based epidemiological setting, the higher the intake of caffeine in normal life, the lower the incidence of AF. This is clarified both in univariate and multivariate statistics. In univariate analysis, the 12-year incidence of AF was significantly lower in the third tertile of caffeine intake (>165 mmol/day, corresponding to four cups of espresso coffee)14 than in the first and the second tertiles (–78% and –62%, respectively). In Cox analysis adjusted for confounders, being in the third tertile of caffeine intake significantly reduced the risk of incident AF with a HR of 0.25 independent of the adverse effect of age, systolic blood pressure, chronic pulmonary disease and history of heart failure, coronary artery disease and cerebrovascular disease. To our knowledge, no population-based epidemiological study has demonstrated up to now this inverse association, although other prospective or pooled data studies found similar results.6,8 Possible mechanisms Due to the epidemiologic nature of our study and to the lack of data in the literature, the reasons for this inverse association can only be the object of speculation. The possibility that caffeine directly favours the AF → sinus rhythm conversion has been suggested;20 in this model, like in our study, the beneficial effect of caffeine was partly contrasted by systolic hypertension. Furthermore, an electrophysiological study demonstrated that the cumulative window of vulnerability is significantly lower in subjects taking caffeine, leading to a reduced propensity to AF.21 Finally, stimulation of A2B receptors plays a role in the pathogenesis of cardiac fibrosis, inhibits cardiac fibroblast production of collagen in vitro and diminishes myocardial remodelling.22–24 These caffeine-induced changes in cardiac homeostasis, when translated to the general population, could play a role in shifting subjects from the category prone to AF to that resistant to AF, reducing the final incidence of AF. Unfortunately, these data are insufficient and can only serve as a guide for future experiments. Role of genetics Studies on caffeine must take into account the –163C > A polymorphism of the CYP1A2 gene,12 as this polymorphism controls the in vivo effects of caffeine and of its active metabolites.1 Actually, the effect of caffeine on cardiovascular, metabolic and cognitive patterns strictly depend on genotype.11,12,16 Univariate analysis (Table 2) seemed to suggest a protective role of the C allele, but a significant trend towards AF reduction was detected in all genotypes. In Cox models, the genotype or the caffeine × CYP1A2 interaction term were not accepted in the equation of risk nor changed the Cox model. The multivariate analysis adjusted for confounders therefore belies the simplistic and misleading univariate approach and indicates that – contrary to other effects of caffeine – the action of caffeine on AF, whatever it is, is independent of genetic control. Strengths of the study The strengths of our study are that it was population-based, that it took into consideration non-selected men and women deriving from a homogenous group of individuals living in comparable conditions in a limited geographical area, that it was performed by a sole research staff and that it took into account the –163C > A polymorphism of the CYP1A2 gene. The ecogenetic context was taken into consideration,25 not excluding a priori any role for the genetic pattern. Another strength is that a double check of caffeine intake was performed, using both a questionnaire and a dietary diary. Finally, a further confirmation of our results comes from the reduction of ischaemic cerebral events, a well-known consequence of AF, with increasing caffeine intake. Limitations of the study A limitation is represented by the fact that caffeine intake was assessed only at the initial screening; eventual changes in the dietary pattern compared to the baseline were not considered. Another limitation is that only one ECG rescreening was planned in the protocol during the follow-up, so that asymptomatic paroxysmal AF not leading to hospitalisation or to the attention of general practitioners might not have been detected. On the other hand, we followed in these aspects the protocols of the Danish Diet, Cancer, and Health Study8 and of the Framingham Study.18 In a population-based study, asymptomatic paroxysmal AF is, through force of circumstances, underestimated. Furthermore, being older and having more comorbidities, subjects in the third tertile of caffeine intake could have been under more medical supervision. The best way we found to avoid any bias was to consider tertiles rather than arbitrary classes. Unfortunately, the three tertiles were not homogeneous but we are confident that multivariate analysis was sufficient to avoid any interference. Finally, caffeine intake in this cohort was quite homogenous, a typical trait of the Italian general population. This finding could have reduced the apparent effect of different caffeine consumption, but – as demonstrated – it was not sufficient to nullify the beneficial effect of caffeine on incident AF. Conclusions A higher caffeine intake is associated with a lower incidence of AF in a 12-year epidemiological prospective setting based on the general population also after adjusting for possible confounders, and is not influenced by the slow or fast caffeine metaboliser genotype. Chronic cardiac and pulmonary disease and high systolic blood pressure26 contrast but do not nullify the beneficial effect of caffeine. Author contribution EC, VT, FA and PP contributed to the conception and design. EC, FA, FG, AM, MM, ED, MB and PS contributed to the acquisition, analysis, or interpretation of data for the work. EC, FA and PS drafted the manuscript. EC, VT, FA, FG, AM, MM, ED, MB and PP critically revised it. All authors gave final approval and agree to be accountable for all aspects of the work ensuring integrity and accuracy. Declaration of conflicting interests The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. Funding The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: this work was supported by Veneto region programmes RSF 811-98 and 178-04. References 1 Cappelletti S , Piacentino D, Sani Get al. Caffeine: cognitive and physical performance enhancer or psychoactive drug? Curr Neuropharmacol 2015 ; 13 : 71 – 88 . Google Scholar Crossref Search ADS PubMed WorldCat 2 Joung B , Lin SF, Chen Zet al. Mechanisms of sinoatrial node dysfunction in a canine model of pacing-induced atrial fibrillation . Heart Rhythm 2010 ; 7 : 88 – 95 . Google Scholar Crossref Search ADS PubMed WorldCat 3 Di Rocco JR , During A, Morelli PJet al. Atrial fibrillation in healthy adolescents after highly caffeinated beverage consumption: two case reports . J Med Case Rep 2011 ; 5 : 18 – 18 . Google Scholar Crossref Search ADS PubMed WorldCat 4 Voigt N , Heijman J, Wang Qet al. Cellular and molecular mechanisms of atrial arrhythmogenesis in patients with paroxysmal atrial fibrillation . Circulation 2014 ; 129 : 145 – 156 . Google Scholar Crossref Search ADS PubMed WorldCat 5 Caldeira D , Martins C, Alves LBet al. Caffeine does not increase the risk of atrial fibrillation: a systematic review and meta-analysis of observational studies . Heart 2013 ; 99 : 1383 – 1389 . Google Scholar Crossref Search ADS PubMed WorldCat 6 Cheng M , Hu Z, Lu Xet al. Caffeine intake and atrial fibrillation incidence: dose response meta-analysis of prospective cohort studies . Can J Cardiol 2014 ; 30 : 448 – 454 . Google Scholar Crossref Search ADS PubMed WorldCat 7 EHRA Scientific Committee Task Force . European Heart Rhythm Association (EHRA)/European Association of Cardiovascular Prevention and Rehabilitation (EACPR) position paper on how to prevent atrial fibrillation endorsed by the Heart Rhythm Society (HRS) and Asia Pacific Heart Rhythm Society (APHRS) . Eur J Prev Cardiol 2017 ; 24 : 4 – 40 . Google Scholar OpenURL Placeholder Text WorldCat 8 Frost L , Vestergaard P. Caffeine and risk of atrial fibrillation or flutter: the Danish Diet, Cancer, and Health Study . Am J Clin Nutr 2005 ; 81 : 578 – 582 . Google Scholar Crossref Search ADS PubMed WorldCat 9 Hansson A , Madsen-Härdig B, Bertil Olsson S. Arrhythmia-provoking factors and symptoms at the onset of paroxysmal atrial fibrillation: a study based on interviews with 100 patients seeking hospital assistance . BMC Cardiovasc Dis 2004 ; 4 : 13 – 13 . Google Scholar Crossref Search ADS WorldCat 10 Ruddox V , Sandven I, Munkhaugen Jet al. Atrial fibrillation and the risk for myocardial infarction, all-cause mortality and heart failure: a systematic review and meta-analysis . Eur J Prev Cardiol 2017 ; 24 : 1555 – 1566 . Google Scholar Crossref Search ADS PubMed WorldCat 11 Casiglia E , Tikhonoff V, Albertini Fet al. Effects of caffeine and coffee on incident heart failure in general population. Role of the CYP1A2 –163C > A polymorphism . WJRR 2017 ; 5 : 42 – 48 . Google Scholar Crossref Search ADS WorldCat 12 Cornelis MC , El-Sohemy A, Kabagambe EKet al. Coffee, CYP1A2 genotype, and risk of myocardial infarction . JAMA 2006 ; 295 : 1135 – 1141 . Google Scholar Crossref Search ADS PubMed WorldCat 13 Tikhonoff V , Casiglia E, Guidotti Fet al. Body fat and the cognitive pattern: a population-based study . Obesity 2015 ; 23 : 1502 – 1510 . Google Scholar Crossref Search ADS PubMed WorldCat 14 Pizziol A , Tikhonoff V, Paleari CDet al. Effects of caffeine on glucose tolerance: a placebo-controlled study . Eur J Clin Nutr 1998 ; 52 : 846 – 849 . Google Scholar Crossref Search ADS PubMed WorldCat 15 Casiglia E , Paleari CD, Petucco Set al. Haemodynamic effects of coffee and purified caffeine in normal volunteers: a placebo-controlled clinical study . J Hum Hypertens 1992 ; 6 : 95 – 99 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 16 Palatini P . Coffee consumption and risk of type 2 diabetes . Diabetologia 2015 ; 58 : 199 – 200 . Google Scholar Crossref Search ADS PubMed WorldCat 17 Thyagarajan B , Alagusundaramoorthy SS, Agrawal A. Atrial fibrillation due to over the counter stimulant drugs in a young adult . J Clin Diagn Res 2015 ; 9 : OD05 – 07 . Google Scholar PubMed OpenURL Placeholder Text WorldCat 18 Shen J , Johnson VM, Sullivan LMet al. Dietary factors and incident atrial fibrillation: the Framingham Heart Study . Am J Clin Nutr 2011 ; 93 : 261 – 266 . Google Scholar Crossref Search ADS PubMed WorldCat 19 Lemery R , Pecarskie A, Bernick Jet al. A prospective placebo controlled randomized study of caffeine in patients with supraventricular tachycardia undergoing electrophysiologic testing . J Cardiovasc Electrophysiol 2015 ; 26 : 1 – 6 . Google Scholar Crossref Search ADS PubMed WorldCat 20 Mattioli AV , Farinetti A, Miloro Cet al. Influence of coffee and caffeine consumption on atrial fibrillation in hypertensive patients . Nutr Metab Cardiovasc Dis 2011 ; 21 : 412 – 417 . Google Scholar Crossref Search ADS PubMed WorldCat 21 Rashid A , Hines M, Scherlag BJet al. The effects of caffeine on the inducibility of atrial fibrillation . J Electrocardiol 2006 ; 39 : 421 – 425 . Google Scholar Crossref Search ADS PubMed WorldCat 22 Anty R , Marjoux S, Iannelli Aet al. Regular coffee but not espresso drinking is protective against fibrosis in a cohort mainly composed of morbidly obese European women with NAFLD undergoing bariatric surgery . J Hepatol 2012 ; 57 : 1090 – 1096 . Google Scholar Crossref Search ADS PubMed WorldCat 23 Dubey RK , Gillespie DG, Jackson EK. Adenosine inhibits collagen and protein synthesis in cardiac fibroblasts: role of A2B receptors . Hypertension 1998 ; 31 : 943 – 948 . Google Scholar Crossref Search ADS PubMed WorldCat 24 Wakeno M , Minamino T, Seguchi Oet al. Long-term stimulation of adenosine A2b receptors begun after myocardial infarction prevents cardiac remodeling in rats . Circulation 2006 ; 114 : 1923 – 1932 . Google Scholar Crossref Search ADS PubMed WorldCat 25 Stolarz K , Staessen JA, Kawecka-Jaszcz Ket al. Genetic variation in CYP11B2 and AT1R influences heart rate variability conditional on sodium excretion . Hypertension 2004 ; 44 : 156 – 162 . Google Scholar Crossref Search ADS PubMed WorldCat 26 Mourtzinis G , Schiöler L, Kahan Tet al. Antihypertensive control and new-onset atrial fibrillation: results from the Swedish Primary Care Cardiovascular Database (SPCCD) . Eur J Prev Cardiol 2017 ; 24 : 1206 – 1211 . Google Scholar Crossref Search ADS PubMed WorldCat © The European Society of Cardiology 2018 This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) © The European Society of Cardiology 2018

Journal

European Journal of Preventive CardiologyOxford University Press

Published: Jul 1, 2018

Keywords: atrial fibrillation; caffeine; cytochrome p-450 cyp1a2; cardiac arrhythmia

There are no references for this article.