Alcohol Consumption and Heart Failure in Hypertensive US Male Physicians
Article Outline
Although alcohol drinking increases blood pressure and heavy drinking has been associated with alcoholic cardiomyopathy, little is known about the association between light to moderate drinking and risk of heart failure (HF) in hypertensive subjects. Thus, the association between light to moderate drinking and incident HF in 5,153 hypertensive male physicians who were free of stroke, myocardial infarction, or major cancers at baseline was prospectively examined. Alcohol consumption was self-reported and classified as <1, 1 to 4, 5 to 7, and ≥8 drinks/week. HF was ascertained using follow-up questionnaires and validated using Framingham criteria. Average age was 58 years, and about 70% of subjects consumed 1 to 7 drinks/week. A total of 478 incident HF cases occurred in this cohort during follow-up. Compared with subjects consuming <1 drink/week, hazard ratios for HF were 0.89 (95% confidence interval [CI] 0.70 to 1.12), 0.72 (95% CI 0.57 to 0.91), and 0.38 (95% CI 0.20 to 0.72) for alcohol consumption of 1 to 4, 5 to 7, and ≥8 drinks/week after adjustment for age, body mass index, smoking, randomization group, use of multivitamins, vegetable consumption, breakfast cereal, exercise, and history of atrial fibrillation, respectively (p for trend <0.001). Similar results were obtained for subjects with HF with and without antecedent myocardial infarction and those without diabetes mellitus. In conclusion, our data suggested that light to moderate alcohol consumption was associated with a lower risk of HF in hypertensive male physicians.
Moderate drinking may lower the risk of heart failure (HF) caused by antecedent coronary artery disease. Recent alcohol consumption was not associated with the subsequent risk of HF in patients who had a myocardial infarction.1, 2 In addition, in patients after myocardial infarction with left ventricular dysfunction, moderate drinking was not associated with incident HF.1, 2 Conversely, other investigators have reported an inverse relation between light to moderate drinking and HF risk in the general population.3, 4, 5, 6 No previous study examined the association between moderate drinking and incident HF in subjects with hypertension. Because moderate alcohol intake could lower the risk of HF (through its beneficial effects on coronary artery disease and neurohormonal changes that might prevent the clinical onset of HF7, 8, 9) and increase the risk of HF through hypertensive cardiomyopathy, it is important to understand the net benefit, if any, of light to moderate drinking (≤2 drinks/day for men) in people with hypertension. Thus, we prospectively assessed whether alcohol consumption was associated with a lower risk of incident HF in US male physicians with prevalent hypertension at baseline. In addition, we examined whether alcohol consumption influenced the risk of HF with and without antecedent myocardial infarction in this population.
Methods
We used data from the Physicians' Health Study (PHS) I, a randomized trial of aspirin and β carotene for the primary prevention of cardiovascular disease and cancer in 22,071 US male physicians. A detailed description of the PHS I has been published previously.10 At baseline, subjects were asked (1) to report their present blood pressure (item 13.a of the questionnaire) and (2) whether they had ever received drug treatment for hypertension (possible answers were never, past only, or present; item 13.b of the questionnaire). Prevalent hypertension was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or positive response to present/past treatment for hypertension. Of 5,176 subjects with prevalent hypertension at baseline, we excluded 23 subjects with missing data for covariates, leaving a final sample of 5,153 subjects for the present analyses. Each subject gave written informed consent, and the Institutional Review Board at Brigham and Women's Hospital, Boston, Massachusetts, approved the study protocol.
Information about usual alcohol consumption was self-reported on a standard questionnaire. Subjects were asked, “How often do you usually consume alcoholic beverages?” Possible response categories included rarely/never, 1 to 3 times/month, 1 time/week, 2 to 4 times/week, 5 to 6 times/week, daily, and ≥2 times/day. The response was interpreted as number of alcoholic drinks consumed during the specified period. In addition, similar alcohol questions were repeated on the 84-month follow-up questionnaire.
A questionnaire was mailed to each subject every 6 months during the first year and has been mailed annually thereafter to obtain information about compliance with the intervention and the occurrence of new medical diagnoses, including HF. In a pilot study, 66 of 73 self-reported HF cases (90%) were confirmed using Framingham criteria (details about HF validation in the PHS were reported previously6). Self-reported information about age, height, weight, body mass index, cigarette smoking, physical activity, prevalent atrial fibrillation, valvular disease, and diabetes mellitus was collected at baseline. In addition, data about such selected foods as fruits and vegetables, breakfast cereals, and history of hypercholesterolemia (defined as reported total cholesterol ≥240 mg/dl or present/past treatments for increased cholesterol) were obtained at baseline.
We computed person-time of follow-up from baseline until the first occurrence of (1) HF, (2) death, or (3) censoring date, the date of receipt of the last follow-up questionnaire. Within each category of alcohol consumption, we calculated the incidence rate of HF by dividing the number of cases by the corresponding person-time. We used Cox proportional hazard models to compute multivariable adjusted hazard ratios (HRs) with corresponding 95% confidence intervals (CIs) using subjects in the lowest category of alcohol consumption as the reference group. We assessed confounding by using 10% change in HR. Assumptions for the proportional hazard models were tested (by including main effects and product terms of covariates and time factor) and were met (all p >0.05). The parsimonious model controlled for age (<45, 45 to 49, 50 to 54, 55 to 59, 60 to 64, and ≥65 years), body mass index (<25, 25 to 29.9, and ≥30 kg/m2), and smoking (never, past, and present smokers). The final multivariable model also controlled for randomization arm (4 groups), use of multivitamins (never, past, and present), breakfast cereal consumption (0, <1, 2 to 6, and ≥7 servings/week), exercise (<1, 1, 2 to 4, and ≥5 times/week), vegetable consumption (<1, 1, and ≥2 servings/day), and history of atrial fibrillation (yes/no). Because alcohol consumption improves insulin sensitivity, we also examined whether diabetes mellitus was an intermediate factor by adding prevalent diabetes in the multivariable model (we expected attenuation of the HRs if the relation between alcohol consumption and HF was partially mediated by diabetes mellitus). To explore that sick quitters may inflate HRs, we repeated these analyses after exclusion of nondrinkers and used present drinkers of 1 to 4 drinks/week as reference. The p value for linear trend was obtained by fitting an ordinal alcohol variable (assigned a value of 1, 2, 3, and 4 from the lowest to the highest category of alcohol, respectively).
In secondary analyses, we examined the association between alcohol consumption and HF with and without antecedent myocardial infarction. Furthermore, we repeated the main analysis while excluding (1) subjects with prevalent diabetes, and (2) subjects with follow-up <2 years. All analyses were completed using SAS, version 9.1 (SAS Institute, Cary, North Carolina). Significance level was set at 0.05.
Results
Of 5,153 PHS subjects who reported prevalent hypertension at baseline, mean age at randomization was 58 ± 10 years (range 40 to 86). About 70% of subjects consumed 1 to 7 drinks/week, and only 4% reported consumption of ≥8 drinks/week. Table 1 lists baseline characteristics of study subjects. During an average follow-up of 18 years, 478 new cases of HF occurred in this cohort. In a multivariable Cox regression model, alcohol consumption was associated with a lower risk of HF in a dose–response fashion (p for linear trend <0.001; Table 2). Additional adjustment for diabetes mellitus (possible intermediate factor) led to a modest attenuation of relative risk estimates (HRs 1.0 [reference], 0.96 [95% CI 0.76 to 1.21], 0.78 [95% CI 0.62 to 0.99], and 0.40 [95% CI 0.21 to 0.76] from the lowest to the highest alcohol category using the fully adjusted model, p for trend 0.002). Excluding subjects with <2 years of follow-up or those with prevalent diabetes mellitus did not alter results (p for trend <0.001 and 0.002, respectively). When alcohol was updated at the 84-month questionnaire, corresponding HRs were 1.0, 1.05 (95% CI 0.84 to 1.31), 0.77 (95% CI 0.61 to 0.98), and 0.58 (95% CI 0.35 to 0.96), respectively (p for trend = 0.008). After exclusion of present nondrinkers, we observed a similar relation among present drinkers (Table 3). Furthermore, in the multivariable adjusted Cox model, alcohol consumption was associated with a lower risk of HF with and without antecedent myocardial infarction (Table 4, Table 5, respectively).
Table 1. Baseline characteristics of 5,153 hypertensive men according to alcohol consumption
| Characteristics | Alcohol Consumption (drinks/wk) | |||
|---|---|---|---|---|
| <1 | 1–4 | 5–7 | ≥8 | |
| (n = 1,312) | (n = 1,661) | (n = 1,958) | (n = 222) | |
| Age (yrs) | 58 ±10 | 56±10 | 59±10 | 60±10 |
| Body mass index (kg/m2) | 26±3 | 26±3 | 25±3 | 26±4 |
| Vegetable intake (servings/d) | 1.1±0.7 | 1.2±0.7 | 1.2±0.7 | 1.2±0.8 |
| Smoker (%) | ||||
| Never | 58% | 48% | 35% | 23% |
| Past | 32% | 43% | 52% | 51% |
| Present | 10% | 10% | 14% | 26% |
| Atrial fibrillation (%) | 1% | 1% | 2% | 3% |
| Prevalent diabetes mellitus (%) | 10% | 5% | 4% | 6% |
| Breakfast cereal consumption | ||||
| None (%) | 31% | 33% | 39% | 53% |
| ≥7 servings/wk (%) | 23% | 17% | 17% | 14% |
| Exercise (times/wk) | ||||
| <1 (%) | 37% | 29% | 28% | 39% |
| ≥5 (%) | 15% | 13% | 16% | 16% |
| Present use of multivitamin (%) | 23% | 20% | 26% | 28% |
| Hypercholesterolemia (%) | 16% | 16% | 18% | 18% |
| Aspirin arm (%) | 50% | 48% | 51% | 50% |
Table 2. Adjusted hazard ratios (HRs) for heart failure (HF) according to alcohol consumption
| Alcohol Intake (drinks/wk) | No. of Cases | HR for HF | ||
|---|---|---|---|---|
| Model 1⁎ | Model 2† | Model 3‡ | ||
| <1 | 135 | 1.0 | 1.0 | 1.0 |
| 1–4 | 156 | 0.91(0.73–1.15) | 0.89(0.71–1.13) | 0.89(0.70–1.12) |
| 5–7 | 177 | 0.78(0.62–0.98) | 0.75(0.59–0.94) | 0.72(0.57–0.91) |
| ≥8 | 10 | 0.46(0.24–0.87) | 0.40(0.21–0.76) | 0.38(0.20–0.72) |
| p Value for linear trend | <0.01 | <0.001 | <0.001 | |
⁎Adjusted for age (<45, 45 to 49, 50 to 54, 55 to 59, 60 to 64, and ≥65 years). |
†Adjusted for age (6 categories as above), body mass index (<25, 25 to 29.9, and ≥30 kg/m2), and smoking (never, past, and present smokers). |
‡Adjusted for variables in model 2 plus randomization arm (4 groups), use of multivitamin (never, past, and present), breakfast cereal consumption (0, <1, 2 to 6, and ≥7 servings/week), exercise (<1, 1, 2 to 4, and ≥5 times/week), vegetable consumption (<1, 1, and ≥2 servings/day), and history of atrial fibrillation. |
Table 3. Adjusted hazard ratios (HRs) for heart failure (HF) according to alcohol consumption using moderate drinkers as reference
| Alcohol Intake (drinks/wk) | No. of Cases | HR for HF | ||
|---|---|---|---|---|
| Model 1⁎ | Model 2† | Model 3‡ | ||
| 1–4 | 156 | 1.0 | 1.0 | 1.0 |
| 5–7 | 177 | 0.85(0.68–1.05) | 0.84(0.67–1.04) | 0.82(0.65–1.02) |
| ≥8 | 10 | 0.50(0.26–0.95) | 0.45(0.24–0.86) | 0.44(0.23–0.84) |
| p Value for linear trend | 0.02 | 0.01 | <0.01 | |
⁎Adjusted for age (<45, 45 to 49, 50 to 54, 55 to 59, 60 to 64, and ≥65 years). |
†Adjusted for age (6 categories as above), body mass index (<25, 25 to 29.9, and ≥30 kg/m2), and smoking (never, past, and present smokers). |
‡Adjusted for variables in model 2 plus randomization arm (4 groups), use of multivitamin (never, past, and present), breakfast cereal consumption (0, <1, 2 to 6, and ≥7 servings/week), exercise (<1, 1, 2 to 4, and ≥5 times/week), vegetable consumption (<1, 1, and ≥2 servings/day), and history of atrial fibrillation. |
Table 4. Adjusted hazard ratios (HRs) for heart failure (HF) with antecedent myocardial infarction by alcohol consumption
| Alcohol Intake (drinks/wk) | No. of Cases | HR for HF | ||
|---|---|---|---|---|
| Model 1⁎ | Model 2† | Model 3‡ | ||
| <1 | 26 | 1.0 | 1.0 | 1.0 |
| 1–4 | 34 | 1.01(0.60–1.68) | 0.98(0.58–1.64) | 0.97(0.58–1.63) |
| 5–7 | 22 | 0.51(0.29–0.90) | 0.50(0.28–0.89) | 0.47(0.26–0.84) |
| ≥8 | 3 | § | — | — |
| p Value for linear trend | 0.02 | 0.02 | 0.01 | |
⁎Adjusted for age (<45, 45 to 49, 50 to 54, 55 to 59, 60 to 64, and ≥65 years). |
†Adjusted for age (6 categories as above), body mass index (<25, 25 to 29.9, and ≥30 kg/m2), and smoking (never, past, and present smokers). |
‡Adjusted for variables in model 2 plus randomization arm (4 groups), use of multivitamin (never, past, and present), breakfast cereal consumption (0, <1, 2 to 6, and ≥7 servings/week), exercise (<1, 1, 2 to 4, and ≥5 times/week), vegetable consumption (<1, 1, and ≥2 servings/day), and history of atrial fibrillation. |
§Insufficient number of subjects for stable estimates. |
Table 5. Adjusted hazard ratios (HRs) for heart failure (HF) without antecedent myocardial infarction by alcohol consumption
| Alcohol Intake (drinks/wk) | No. of Cases | HR for HF | ||
|---|---|---|---|---|
| Model 1⁎ | Model 2† | Model 3‡ | ||
| <1 | 109 | 1.0 | 1.0 | 1.0 |
| 1–4 | 122 | 0.89(0.69–1.15) | 0.87(0.67–1.13) | 0.86(0.66–1.13) |
| 5–7 | 155 | 0.84(0.66–1.08) | 0.80(0.62–1.03) | 0.78(0.60–1.00) |
| ≥8 | 7 | 0.39(0.18–0.84) | 0.34(0.16–0.72) | 0.32(0.15–0.70) |
| p Value for linear trend | 0.03 | <0.01 | <0.01 | |
⁎Adjusted for age (<45, 45 to 49, 50 to 54, 55 to 59, 60 to 64, and ≥65 years). |
†Adjusted for age (6 categories as above), body mass index (<25, 25 to 29.9, and ≥30 kg/m2), and smoking (never, past, and present smokers). |
‡Adjusted for variables in model 2 plus randomization arm (4 groups), use of multivitamin (never, past, and present), breakfast cereal consumption (0, <1, 2 to 6, and ≥7 servings/week), exercise (<1, 1, 2 to 4, and ≥5 times/week), vegetable consumption (<1, 1, and ≥2 servings/day), and history of atrial fibrillation. |
Discussion
In this cohort of hypertensive US male physicians, we found that light to moderate alcohol consumption was associated with a lower risk of HF. Furthermore, the inverse relation between light to moderate alcohol consumption and HF was observed for patients with HF with and without antecedent myocardial infarction. This was the first study to report an association between light to moderate drinking and risk of HF in hypertensive subjects.
Previous studies showed that excessive drinking was associated with left ventricular dysfunction,11, 12 dilated cardiomyopathy,13, 14 and hypertension.15, 16 However, current reports provided evidence supporting the beneficial effects of moderate drinking on myocardial infarction and other cardiovascular diseases17, 18, 19 and mortality.20 It is thus possible that some of the HF cases might have been averted through prevention of myocardial infarction by light to moderate drinking. However, the fact that we observed an inverse association between alcohol consumption and HF without antecedent myocardial infarction suggested that other biologic mechanisms could have a role in the association between alcohol intake and HF.
Earlier reports showed beneficial effects of moderate alcohol consumption on other risk factors for HF. For example, alcohol consumption was shown to increase high-density lipoprotein cholesterol,21 improve insulin sensitivity and lower the risk of diabetes mellitus,22, 23 decrease plasma inflammatory markers,24 and increase plasma adiponectin.25, 26 In experimental studies, adiponectin was shown to attenuate cardiac hypertrophy in response to pressure overload in adiponectin knockout mice.27 These antihypertrophic actions of adiponectin are possibly mediated through activation of adenosine monophosphate–activated protein kinase.27 Furthermore, alcohol consumption has diuretic effects and was associated with plasma atrial natriuretic peptide.9 Such a diuretic effect could prevent volume overload and delay the onset of signs and symptoms of HF. One or more of these physiologic mechanisms could partially or completely explain the observed association between light to moderate alcohol consumption and HF without antecedent myocardial infarction in our study.
Our study had some limitations. First, our sample consisted of male physicians who may have different lifestyle factors than the general population, thus limiting the generalizability of our findings. In addition, the lack of data for women limited our ability to make inference on the relation between moderate drinking and HF in hypertensive women or explore a gender–alcohol interaction on HF. Furthermore, identification of subjects with prevalent hypertension was based on self-reported information. However, systolic and diastolic blood pressure measured on a standardized examination showed a reasonable correlation with self-reported values in physicians (correlation coefficients 0.72 and 0.60 for systolic and diastolic blood pressure, respectively; all p <0.0001).28 Second, because most subjects were light to moderate drinkers (70% reported consumption of 1 to 7 drinks/week, whereas only 4% reported alcohol consumption ≥8 drinks/week), we could not examine the association between heavy drinking and risk of HF in hypertensive subjects. Third, underreporting of alcohol drinking could have obscured the true association between light to moderate drinking and risk of HF. It is possible that sick quitters may have inflated the observed association because we did not have data to separate lifelong abstainers from former drinkers. However, this was less likely because we showed similar inverse relations between alcohol intake and HF after exclusion of present abstainers. Fourth, we did not have information for drinking patterns, beverage types, or whether alcohol was consumed with meals to examine their influences on the studied association. Fifth, given the observational nature of our data, we cannot exclude residual confounding or confounding by unmeasured factors as possible explanation of our findings. Nevertheless, the large sample size; long-term follow-up; robustness of the data in sensitivity analyses, including use of updated alcohol at 84 months from baseline measurements; and standardized and complete follow-up in the PHS were major strengths of this study.
Acknowledgments
We thank the PHS subjects for outstanding commitment and cooperation and the entire PHS staff for expert and unfailing assistance.
References
- . Lack of effect of recent alcohol consumption on the course of acute myocardial infarction. Am Heart J. 1999;138:926–933
- Alcohol consumption and prognosis in patients with left ventricular systolic dysfunction after a myocardial infarction. J Am Coll Cardiol. 2004;43:2015–2021
- . Alcohol consumption and risk for congestive heart failure in the Framingham Heart Study. Ann Intern Med. 2002;136:181–191
- . Moderate alcohol consumption and risk of heart failure among older persons. JAMA. 2001;285:1971–1977
- . The association of alcohol consumption and incident heart failure: the Cardiovascular Health Study. J Am Coll Cardiol. 2006;48:305–311
- . Alcohol consumption and risk of heart failure in the Physicians' Health Study I. Circulation. 2007;115:34–39
- . Increased plasma atrial natriuretic peptide after ingestion of low doses of ethanol in humans. Alcohol Clin Exp Res. 1997;21:162–170
- . A possible role of atrial natriuretic peptide in ethanol-induced acute diuresis. Life Sci. 1991;48:635–642
- . Alcohol consumption and plasma atrial natriuretic peptide (from the HyperGEN Study). Am J Cardiol. 2006;98:628–632
- . Final report on the aspirin component of the ongoing Physicians' Health Study. N Engl J Med. 1989;321:129–135
- . Unrecognized left ventricular dysfunction in an apparently healthy alcohol abuse population. Drug Alcohol Depend. 1991;28:113–119
- . Left ventricular size, mass and function in relation to the duration and quantity of heavy drinking in alcoholics. Am J Cardiol. 1991;67:274–279
- . Alcohol consumption and idiopathic dilated cardiomyopathy: a case control study. Am Heart J. 1998;135:833–837
- . Alcohol and heart failure. J Card Fail. 2002;8:239–246
- . Alcohol consumption and risk for hypertension in middle-aged Japanese men. J Hypertens. 2001;19:851–855
- . Potassium, calcium, alcohol intake and blood pressure: the Zutphen Study. Am J Clin Nutr. 1985;41:1299–1304
- . Meta-analysis of the relationship between alcohol consumption and coronary heart disease and mortality in type 2 diabetic patients. Diabetologia. 2006;49:648–652
- . Alcohol and coronary heart disease. Int J Epidemiol. 2001;30:724–729
- . Meta-analysis of wine and beer consumption in relation to vascular risk. Circulation. 2002;105:2836–2844
- . Alcohol dosing and total mortality in men and women: an updated meta-analysis of 34 prospective studies. Arch Intern Med. 2006;166:2437–2445
- . Moderate alcohol intake, increased levels of high-density lipoprotein and its subfractions, and decreased risk of myocardial infarction. N Engl J Med. 1993;329:1829–1834
- . A prospective study of drinking patterns in relation to risk of type 2 diabetes among men. Diabetes. 2001;50:2390–2395
- . Alcohol consumption and type 2 diabetes among older adults: the Cardiovascular Health Study. Obesity. 2007;15:1758–1765
- . Overall alcohol intake, beer, wine, and systemic markers of inflammation in western Europe: results from three MONICA samples (Augsburg, Glasgow, Lille). Eur Heart J. 2004;25:2092–2100
- . Effect of moderate alcohol consumption on adipokines and insulin sensitivity in lean and overweight men: a diet intervention study. Eur J Clin Nutr. 2007;Epub 2007 June 6
- . The effect of moderate alcohol consumption on adiponectin oligomers and muscle oxidative capacity: a human intervention study. Diabetologia. 2007;50:1388–1392
- Adiponectin-mediated modulation of hypertrophic signals in the heart. Nat Med. 2004;10:1384–1389
- . Validity of physicians' self-reports of cardiovascular disease risk factors. Ann Epidemiol. 1993;3:442–447
The Physicians' Health Study was supported by Grants No. CA-34944, CA-40360, and CA-097193 from the National Cancer Institute, Bethesda, Maryland, and Grants No. HL-26490 and HL-34595 from the National Heart, Lung, and Blood Institute, Bethesda, Maryland. Dr. Djoussé is Principal Investigator on a K01 HL-70444 from the National Heart, Lung, and Blood Institute, Bethesda, Maryland.
PII: S0002-9149(08)00761-3
doi:10.1016/j.amjcard.2008.04.031
© 2008 Elsevier Inc. All rights reserved.
