Association between alcohol consumption and systolic ventricular function: A population-based study Haroon Yousaf, MD, a Richard J. Rodeheffer, MD, b Timothy E. Paterick, MD, JD, a Zain Ashary, BS, a Mirza Nubair Ahmad, MBBS, a and Khawaja Afzal Ammar, MD a Milwaukee, WI and Rochester, MN
Background Although moderate alcohol consumption is associated with decreased clinical heart failure, there are no population-based studies evaluating the relationship between alcohol consumption and left ventricular (LV) systolic function. We sought to evaluate the relationship between alcohol consumption and LV systolic function in the community. Methods In a population-based random sample of 2,042 adults, age ≥45 years, we assessed alcohol consumption by a self-administered questionnaire. Responders were categorized by alcohol consumption level: abstainer, former drinker, light drinker (b1 drink a day), moderate drinker (1-2 drinks a day), and heavy drinker (N2 drinks a day). Systolic function was assessed by echocardiography. Results We identified 38 cases of systolic dysfunction in 182 abstainers, 309 former drinkers, 1,028 light drinkers, 251 moderate drinkers, and 146 heavy drinkers. A U-shaped relationship was observed between alcohol consumption and moderate systolic dysfunction (LV ejection fraction [LVEF] ≤40%), with the lowest prevalence in light drinkers (0.9%) compared to the highest prevalence in heavy drinkers (5.5%) (odds ratio 0.14, 95% CI 0.04-0.43). This association persisted across different strata of risk factors of systolic dysfunction as well as in multivariate analysis. No significant association between alcohol consumption and systolic function was seen in subjects with LVEF N50% or ≤50%. Conclusions
There is a U-shaped relationship between alcohol consumption volume and LVEF, with the lowest risk of moderate LV dysfunction (LVEF ≤40%) observed in light drinkers (b1 drink a day). These findings are parallel to the relationship between alcohol consumption and cardiovascular disease prevalence. (Am Heart J 2014;0:1-8.)
A U-shaped relationship exists between the level of alcohol consumption and prevalence of clinical heart failure. 1 Regular heavy alcohol consumption is associated with impairment of left ventricular (LV) function 2,3 and occasionally results in overt cardiomyopathy. 4 Alternatively, regular but moderate alcohol consumption is associated with prevention of heart failure. 5,6 Although this relationship between alcohol consumption and clinical heart failure has been extensively studied, the reverse parabolic association between alcohol consumption and ventricular systolic function has not been evaluated in a population-based study. Because both
moderate drinking and heavy drinking are prevalent in the community, we hypothesized that moderate alcohol consumption would be associated not only with reduced prevalence of clinical cardiovascular disease but also with reduced prevalence of systolic dysfunction. We also postulated that heavy alcohol consumption would be associated with increased prevalence of LV systolic dysfunction. This hypothesis was tested in a randomly sampled, population-based cohort of 2,042 people ≥45 years old in Olmsted County, Minnesota.
Methods
From the aAurora Cardiovascular Services, Aurora Sinai/Aurora St. Luke’s Medical Centers, University of Wisconsin School of Medicine and Public Health, Milwaukee, WI, and bMayo Clinic and Foundation, Rochester, MN. Financial support: US Public Health Services, National Institutes of Health, HL-RO1555902. Submitted December 6, 2013; accepted February 25, 2014. Reprint requests: Khawaja Afzal Ammar, MD, Aurora Cardiovascular Services, 2801 W Kinnickinnic River Pkwy, #840, Milwaukee, WI 53215. E-mail: [email protected] http://dx.doi.org/10.1016/j.ahj.2014.02.014 0002-8703 © 2014, Mosby, Inc. All rights reserved.
The Olmsted County Heart Function Study was approved by the Mayo Foundation and Olmsted Medical Center Institutional Review boards, and written informed consent was obtained from the subjects. The present study was carried out with funding from US Public Health Services, National Institutes of Health, grant HL-RO1-555902. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the paper, and its final contents.
Study setting Of the 124,277 residents of Olmsted County, as of the 2000 census, 90% were white. Other characteristics of this
American Heart Journal Month Year
2 Yousaf et al
population and its unique resources for population-based epidemiological research have been previously described. 7,8
Population sampling and data collection A random sample of residents ≥45 years old as of January 1, 1997, was identified. A 7% sampling fraction was applied within each gender- and age- (5 years) specific stratum. Of the 4,203 subjects invited, 2,042 (48.6%) participated. Enrollment took place from January 1, 1997, to September 30, 2000. Subjects completed a self-administered questionnaire and underwent physical examination, electrocardiography, and echocardiography.
Definitions of selected clinical conditions Medical record review was carried out by 4 trained chart abstractor nurses. The median length of participant medical record archive was 36 years. Heart failure was diagnosed if Framingham criteria were fulfilled. 9,10 Myocardial infarction and hypertension were diagnosed with criteria from the World Health Organization and Sixth Report of the Joint National Committee, respectively. 11,12 Coronary artery disease, angina, diabetes, cerebrovascular accident, transient ischemic attack, and peripheral vascular disease were diagnosed if documented by a physician. Angina was assessed based on a selfadministered questionnaire by Rose et al. 13 Previous smokers were those who had smoked ≥100 cigarettes in the past but were not currently smoking. In accordance with the National Cholesterol Education Program Adult Treatment Panel III criteria, 14 metabolic syndrome was defined by the presence of ≥3 of the following criteria: (1) central obesity defined as a waist circumference ≥102 cm in men and ≥88 cm in women, (2) triglyceride level N150 mg/dL (to convert to millimoles per liter, multiply by 0.0113), (3) high-density lipoprotein cholesterol level b40 mg/dL in men and b50 mg/dL in women (to convert to millimoles per liter, multiply by 0.0259), (4) blood pressure of ≥130/85 mm Hg, and (5) fasting glucose level of ≥110 mg/dL (to convert to millimoles per liter, multiply by 0.0555). Anthropometric measurements were carried out by a trained research nurse with the subjects in the standing position. Body mass index was measured as weight (kilograms)/height (meters). 2
Echocardiographic analysis All subjects underwent echocardiography following a standardized protocol as previously described. 15–17 A single cardiologist, blinded to clinical status, interpreted all echocardiograms.
Measurement of LV systolic function Ejection fraction was measured using M-mode echocardiography, 2-dimensional echocardiography by Simpson’s biplane method, and visual estimate. As LV ejection fraction (LVEF) from these 3 methods was highly correlated and the visual estimate was available in N99% of participants, visual estimate was used in analysis.
Quantification of alcohol consumption Subjects completed a self-administered questionnaire that first asked if they had ever consumed alcohol. A negative answer resulted in classification as an abstainer. Then they were asked if alcohol had been consumed in the last 12 months. A negative answer resulted in classification as a former drinker. An affirmative answer to both questions led the subject to 3 separate sections on beer, wine, and liquor. The 12 frequency categories ranged from 3 to 4 times a day to once a year. The 16 quantity categories ranged from 1 to ≥16 drinks per day. The estimated alcohol content per drink was 12.8 g for beer, 11 g for wine, and 14 g for spirits (liquor or whiskey). Responders were categorized further by alcohol consumption level: light drinker (b1 drink a day, b11 g/d), moderate drinker (1-2 drinks a day, 11-28 g/d), and heavy drinker (N2 drinks a day, N28 g/d).
Statistical analysis Categorical data are summarized as a percentage of the group total with corresponding 95% CI based on normal approximation, and comparison between groups was based on the χ 2 test for association. Continuous variables are summarized as mean ± standard deviation, and comparisons between groups were based on analysis of variance models. Post–analysis of variance comparisons of continuous variables were based on t test, but no adjustments for multiple comparisons were made. Multivariate analysis was carried out with logistic regression for categorical variables and multiple linear regressions for continuous variables. Two-sided P values b.05 were considered significant. Analyses were performed using JMP Genomics, 5.0 (SAS Institute, Cary, NC).
Results Characteristics of study participants Letters of invitations were sent to 4,203 individuals. Of the 2,042 individuals (48.6%) who participated, 1,916 (941 men and 975 women) had valid data and were included in the present analysis. The majority of participants were light drinkers (n = 1,028, 53.7%), with smaller proportions of abstainers (n = 182, 9.5%), former drinkers (n = 309, 16.1%), moderate drinkers (n = 251, 13.1%), and heavy drinkers (n = 146, 7.6%). Table I shows the baseline characteristics of participants categorized by their alcohol consumption habits. Abstainers were more likely to be women, were less likely to be smokers, but were found to have a high prevalence of diabetes mellitus and hypertension. Light drinkers (b1 drinks a day) represented a healthy subset of the population with the lowest mean age, 2% current smokers, 6% diabetic, and 27% hypertensive. Heavy drinkers (N2 drinks a day) had the highest prevalence of current smokers (22%) and mean high-density lipoprotein levels (48.3 ± 16.1). Alcohol consumption and prevalence of clinical cardiovascular disease Table II shows the prevalence of clinical cardiovascular disease (all cardiovascular diseases combined) categorized by alcohol consumption. The prevalence
American Heart Journal Volume 0, Number 0
Yousaf et al 3
Table I. Characteristics of participants categorized by alcohol consumption volume
Age, y Women, n = 975 Body mass index, kg/m 2 Obese (body mass index ≥30) Smoking history Past Current Never Diabetes mellitus Hypertension HDL Metabolic syndrome
Abstainers (n = 182)
Former drinkers (n = 309)
b1 drink/d (n = 1028)
1-2 drinks/d (n = 251)
N2 drinks/d (n = 146)
P
68.0 ± 10.8 138 (75.8) 29.4 ± 6.3 72 (39.6)
63.0 ± 11.2 173 (56.0) 29.1 ± 5.5 121 (39.1)
60.9 ± 10.2 546 (53.1) 28.3 ± 5.4 320 (31.1)
61.6 ± 9.8 89 (35.5) 27.6 ± 4.6 60 (23.9)
62.1 ± 9.6 29 (19.9) 28.5 ± 4.7 47 (32.2)
b.0001 b.0001 b.0001 .0008
22 (12.1) 4 (2.2) 156 (85.7) 23 (12.6) 69 (37.9) 45.4 ± 14.4 84 (45.1)
122 (39.5) 27 (8.7) 160 (51.8) 39 (12.6) 109 (35.3) 43.9 ± 13.7 132 (42.4)
430 (41.8) 78 (7.6) 520 (50.6) 61 (5.9) 280 (27.2) 45.3 ± 14.3 378 (36.7)
134 (53.4) 34 (13.5) 83 (33.1) 17 (6.8) 62 (24.7) 48.2 ± 15.0 84 (33.3)
87 (59.6) 32 (21.9) 27 (18.5) 5 (3.4) 42 (28.8) 48.3 ± 16.1 55 (37.1)
1-2 drinks/d (n = 251)
N2 drinks/d (n = 146)
P
8 (3.2) 33 (13.2) 6 (2.4) 22 (8.8) 3 (1.2) 6 (2.4) 7 (2.8) 43 (17.1)
7 (4.8) 22 (15.1) 3 (2.1) 12 (8.2) 2 (1.4) 2 (1.4) 4 (2.7) 26 (17.8)
.008 .015 .078 .006 .13 .11 .09 b.0001
b.0001 b.0001 .0019 .0021 .046
Data are expressed as mean ± SD or number (percentage). HDL, High-density lipoprotein.
Table II. Prevalence of cardiovascular disease according to alcohol consumption volume Abstainers (n = 182) Myocardial infarction Coronary artery disease Heart failure Angina Cerebrovascular accidents Transient ischemic attack Peripheral vascular disease Any cardiovascular disease
17 30 8 22 4 8 9 47
(9.3) (16.5) (4.4) (12.1) (2.2) (4.4) (4.9) (25.8)
Former drinkers (n = 309) 23 48 12 36 10 8 9 70
(7.4) (15.5) (3.9) (11.7) (3.2) (2.6) (2.9) (22.7)
b1 drink/d (n = 1028) 43 102 17 63 12 15 17 138
(4.2) (9.9) (1.7) (6.1) (1.2) (1.5) (1.7) (13.4)
Data are expressed as number (percentage).
of combined cardiovascular disease indicated a U-shaped relationship (Table II, row 8). When cardiovascular diseases were evaluated individually, the U-shaped relationship persisted, maintaining statistical significance only in myocardial infarction, coronary artery disease, and angina.
Alcohol consumption and systolic function Table III shows the mean LVEF and prevalence of systolic dysfunction categorized by alcohol consumption. There was no clinically meaningful difference in mean LVEF in different alcohol consumption groups. Similarly, there was no statistical association between alcohol consumption group and prevalence of normal systolic function (LVEF N50%) (Table III; row 2) or any systolic dysfunction (LVEF ≤50%) (Table III, row 3). However, when the participants were stratified with respect to moderate systolic dysfunction (LVEF ≤40%) (Table III; row 4), a U-shaped relationship emerged. The prevalence of moderate systolic dysfunction (LVEF ≤40%) was the lowest in light drinkers, with prevalence increasing with increasing or decreasing consumption of alcohol, indi-
cating a U-shaped relationship (Figure 1). In those with LVEF ≤40% (Table III, row 4; Figure 1), the prevalence of systolic dysfunction increased 6-fold between light drinkers (0.9%) and heavy drinkers (5.5%; P b .001). The light drinkers were further stratified into habitual drinkers (those who drink at least once a week) and nonhabitual drinkers (those who drink less than once a week). The prevalence of systolic dysfunction (LVEF ≤50) was lower in habitual drinkers compared to nonhabitual drinkers (4.06% vs 5.91%). A similar trend was seen in the prevalence of moderate systolic dysfunction (LVEF ≤40), which was lower in habitual drinkers (0.24%) compared to nonhabitual drinkers (1.15%). The association between alcohol consumption and moderate systolic dysfunction is further evaluated in Table IV and Figure 2 by stratification of the participants into multiple strata of systolic dysfunction risk factors. The beneficial association was not statistically significant in the young (≤65 years), smokers, nonobese, and those without metabolic syndrome. The U-shaped association between alcohol consumption and LV systolic function persisted across different
American Heart Journal Month Year
4 Yousaf et al
Table III. Left ventricular systolic function association with alcohol consumption volume
LVEF LVEF N50% LVEF ≤50% LVEF ≤40%
Abstainers (n = 182)
Former drinkers (n = 309)
b1 drink/d (n = 1028)
1-2 drinks/d (n = 251)
N2 drinks/d (n = 146)
P
63.9 ± 7.9 172 (94.5) 10 (5.5) 6 (3.3)
62.4 ± 8.0 279 (90.3) 30 (9.7) 6 (1.9)
63.1 ± 6.4 977 (95.0) 51 (5.0) 9 (0.9)
61.8 ± 7.5 232 (92.4) 19 (7.6) 9 (3.6)
61.4 ± 7.9 137 (93.8) 9 (6.2) 8 (5.5)
b.001 .064 .11 b.001
Data are expressed as mean ± (SD) or number (percentage).
systolic function. In the Olmsted County Heart Study, the prevalence of moderate LV systolic dysfunction was the lowest in light drinkers (Table III, row 4; Figure 1), with the prevalence increasing with increasing or decreasing consumption of alcohol, indicating a U-shaped relationship. This relationship remained statistically significant for moderate systolic dysfunction even after adjustment in a multivariate model. This U-shaped relationship between alcohol consumption and LV systolic function was not seen in those with normal systolic function (LVEF N50%) or any impairment of systolic function (LVEF ≤50%). These results enhance our understanding of the beneficial and detrimental effects of alcohol consumption on systolic heart function as measured by echocardiography.
Figure 1
Alcohol intake and prevalence of moderate systolic dysfunction. The lowest prevalence of moderate systolic dysfunction (LVEF ≤40%) was seen in those who consumed b1 drink a day. The prevalence of systolic dysfunction increased 6-fold between light drinkers (0.9%) and heavy drinkers (5.5%).
Table IV. Odds of LV moderate systolic dysfunction according to alcohol consumption volume Unadjusted baseline model Abstainers Former drinkers b1 drink/d 1-2 drinks/d N2 drinks/d
OR (95% CI) 1.00 0.57 0.14 1.56 4.75
(0.12-2.11) (0.04-0.43) (0.39-5.20) (1.18-15.98)
P
.44 b.001 .49 .02
strata of risk factors of systolic dysfunction (Figure 2), including gender, age, hypertension, diabetes, smoking, obesity, and metabolic syndrome. It also persisted in a multivariate model after adjustment for all risk factors (odds ratio [OR] 0.20, 95% CI 0.06-0.64, P b .0001).
Discussion Principal findings This population-based study demonstrated a U-shaped relationship between alcohol consumption and LV
Association of alcohol with clinical cardiovascular diseases Evidence of moderate alcohol consumption reducing clinical cardiovascular events has been shown by numerous observational, 18,19 case control, 20,21 and cohort studies 22,23 both in healthy subjects 24 and in participants with risk factors such as diabetes. 25 In the current study, as previously reported, 26 the results confirm the salutatory effects of alcohol on any clinical cardiovascular disease, whether evaluated individually or collectively in the general population, including myocardial infarction, coronary artery disease, congestive heart failure, angina, cerebrovascular accident, transient ischemic attack, and peripheral artery disease (Table II). This suggests that the current study population is similar to prior studies in terms of association of alcohol consumption with clinical cardiovascular disease, which makes this study population an ideal setting to evaluate an association that has never been described (ie, alcohol and clinical systolic dysfunction). Figure 3 demonstrates the U-shaped relationship between alcohol and clinical heart failure (Table II, row 3; Figure 3). The higher prevalence of diabetes mellitus and hypertension in abstainers may have contributed to higher prevalence of clinical heart failure (Figure 3). Alcohol has been shown to reduce insulin resistance and blood pressure, which also may explain the U-shaped relationship of diabetes mellitus and hypertension with alcohol consumption as seen in our study. Abstainers had much higher odds of clinical congestive heart failure
American Heart Journal Volume 0, Number 0
Yousaf et al 5
Figure 2
Forest plot depicting bivariate stratified analysis of odds of LV moderate systolic dysfunction in light versus heavy drinkers. The association between alcohol consumption and moderate systolic dysfunction, as evaluated by stratification of the participants into multiple strata, persists for most strata in light versus heavy drinkers.
compared to light drinkers (OR 2.7, 95% CI 1.09-6.18, P = .022). When adjusted for both hypertension and diabetes mellitus, the OR dropped meaningfully to 1.9 (95% CI 0.76-4.60, P = .14). This meaningful drop (N20%) in OR indicates that, in abstainers, diabetes mellitus and hypertension were the intervening variables for clinical heart failure.
Association of alcohol with systolic dysfunction Although ejection fraction is a measure of global LV systolic function, LV performance can be further evaluated with multiple other parameters, including measurement of LV dimensions, calculation of LV mass index, regional wall motion abnormalities, as well as diastolic dysfunction. The data on these indices are not relevant to this manuscript because the focus of this study is LVEF. Prior animal (case control) 27,28 and human (case control 29,30 and cohort 3,31) studies have demonstrated heavy alcohol use to be associated with LV systolic dysfunction and alcoholic cardiomyopathy. 32 Among heavy drinkers, reduction in alcohol consumption has shown improvement in LV function. 32,33 In contrast, moderate alcohol consumption is known to protect against heart failure. 1,5,6,34 In studies of patients consuming moderate amount of alcohol, with known heart failure, no deterioration of cardiac
function was seen. 35–37 Our data build on these studies by providing information about systolic function in this population-based sample. It demonstrates light drinkers as having the lowest prevalence of systolic dysfunction with progressive worsening seen with increasing alcohol consumption. Former drinkers were noted to have a lower prevalence of moderate systolic dysfunction (LVEF ≤40%) as compared to moderate drinkers or heavy drinkers. When the entire population was evaluated (Table III, row 1), the Ushaped relationship was statistically significant but not clinically meaningful, with LVEF ranging from 61.4% to 63.9%. No association between alcohol consumption and prevalence of normal systolic function (Table III, row 2) was seen. The U-shaped relationship becomes clearer and more statistically significant in the stratum of participants with LVEF ≤40%, (Table III, row 4) where the prevalence of systolic dysfunction increases from 0.9% to 5.5% in light drinkers as compared to heavy drinkers, a 6-fold rise in prevalence. This suggests that the detrimental effect of alcohol on ventricular function becomes stronger as ventricular systolic function deteriorates. Current National Institutes of Health (National Institute on Alcohol Abuse and Alcoholism) guidelines recommend that patients with alcoholic cardiomyopathy should not drink any alcohol. 38 The findings of this study create an interesting clinical paradox, as those who
6 Yousaf et al
Figure 3
Alcohol intake and prevalence of heart failure. The lowest prevalence of heart failure was seen in those who consumed b1 drink a day.
stand to benefit the most from light alcohol consumption, that is, subjects with LVEF ≤40%, are also those who are likely to be harmed if their consumption exceeds N1 drink a day. Whether those with moderate systolic dysfunction and a proven track record of light alcohol consumption should be encouraged to continue their behavior cannot be proved from these data and requires a prospective study. Current guidelines recommend that, in the community, moderate alcohol consumption (≤1 drink a day for women and elderly and 1-2 drinks a day for men) may reasonably be continued based on existing data on clinical cardiovascular disease. 38 The current study suggests that, at least from the systolic dysfunction perspective, these guidelines may need further consideration and investigation. There is no beneficial association between alcohol consumption and LV systolic function in healthy individuals with LVEF N50%. The association is insignificant in those with LVEF ≤50% and becomes significant only in those with LVEF ≤40%. Prior large clinical studies have suggested a clinical benefit of moderate alcohol consumption on the occurrence of clinical heart failure, but they do not have data on ejection fraction. 1,6,34 Although our study confirms the U-shaped relationship in clinical heart failure (Table II, row 3; Figure 3), it also extends the findings to demonstrate that the beneficial effect is limited to those with moderate systolic dysfunction. The salutatory association between alcohol and moderate systolic dysfunction was not statistically significant in the young (age ≤65 years), smokers, nonobese, and those without metabolic syndrome. Several observational studies 39 have shown similar cardioprotective effects of alcohol in both smokers and nonsmokers, suggesting that alcohol consumption affects heart function similarly in smokers and nonsmokers. The prevalence of LV systolic
American Heart Journal Month Year
dysfunction was almost 1½ times greater in current smokers than in the general population (2.7% vs 1.9%) and more than 2 times greater in current smokers than in noncurrent smokers (2.7% vs 1.3%). Whether the lack of association is due to small sample size or current smoking nullifies the beneficial effect of alcohol can only be speculated. Other strata that failed to show a beneficial association (young, nonobese, and those without metabolic syndrome) in our study cannot be explained and might be a result of either small sample size (n = 38 with systolic dysfunction in the entire cohort) or confounding due to unmeasured variables. Further studies are needed to evaluate these associations.
Strengths and limitations The strengths of the present study include its randomly selected population-based sample, the availability of medical records with a 36-year median length of archive, uniform collection of cross-sectional data, and standardized echocardiographic measurements. Conclusions from this predominantly white cohort including only persons ≥45 years cannot be generalized to the entire US population. Participation bias was evaluated by medical record abstraction of a random selection of 500 participants and 500 invited nonparticipants, which showed no difference in cardiovascular disease. 40 The assessment for alcohol consumption was based on self-reporting and, therefore, may result in misclassification of exposure status. Research participants have been known to underreport alcohol consumption by as much as 39% for men and 44% for women. 41 However, selfadministered questionnaire data reflect the quality of information available to medical practitioners and have been shown to provide useful estimates of alcohol intake in epidemiologic studies. 42 In addition, the ability to make cause-effect inferences and assess time-dependent changes was limited because of the cross-sectional nature of the study. Although the current study has a large sample size (n = 2,042), the study suffers from a lack of power, especially in those with systolic dysfunction. For example, the numbers of patients with moderate systolic dysfunction in the 5 categories are very small (n = 6, 6, 9, 9, and 8 in abstainers, former drinkers, b1 drink a day, 1-2 drinks a day, and N2 drinks a day, respectively). However, it is in this stratum that the statistically significant results are seen. Whether statistically significant results could have been observed in other strata if the sample sizes were greater is debatable. However, the level of statistical significance with P b .001 suggests that, in those with moderate systolic dysfunction, the results are extremely unlikely to have arisen by chance. This is a commonly observed phenomenon in other studies where the maximum benefit from an intervention is secured by those with the worst disease (eg, in the CASS study, 43 the
American Heart Journal Volume 0, Number 0
maximum survival benefit from coronary artery bypass grafting was reaped by those who had moderate systolic dysfunction or severe angina). This observation generates the hypothesis that the maximum benefit from alcohol may be in those with moderate systolic dysfunction.
Conclusion In a community setting, light alcohol consumption (b1 drink a day) was associated with a reduced prevalence of moderate systolic dysfunction (LVEF ≤40%).
Acknowledgements The authors gratefully acknowledge Joe Grundle and Katie Klein for editorial preparation of the manuscript, Brian Miller and Brian Schurrer for their help in preparing figures, and Chi Cho for his help in creating the forest plot.
Disclosures Conflict of interests: The present study was carried out with funding from US Public Health Services, National Institutes of Health, grant HL-RO1-555902. The authors have no competing interests to disclose.
References 1. Bryson CL, Mukamal KJ, Mittleman MA, et al. The association of alcohol consumption and incident heart failure: the Cardiovascular Health Study. J Am Coll Cardiol 2006;48:305-11. 2. Urbano-Marquez A, Estruch R, Navarro-Lopez F, et al. The effects of alcoholism on skeletal and cardiac muscle. N Engl J Med 1989;320:409-15. 3. Kajander OA, Kupari M, Laippala P, et al. Dose dependent but non-linear effects of alcohol on the left and right ventricle. Heart 2001;86:417-23. 4. Laonigro I, Correale M, Di Biase M, et al. Alcohol abuse and heart failure. Eur J Heart Fail 2009;11:453-62. 5. Walsh CR, Larson MG, Evans JC, et al. Alcohol consumption and risk for congestive heart failure in the Framingham Heart Study. Ann Intern Med 2002;136:181-91. 6. Abramson JL, Williams SA, Krumholz HM, et al. Moderate alcohol consumption and risk of heart failure among older persons. JAMA 2001;285:1971-7. 7. Kurland LT, Molgaard CA. The patient record in epidemiology. Sci Am 1981;245:54-63. 8. Melton III LJ. History of the Rochester Epidemiology Project. Mayo Clin Proc 1996;71:266-74. 9. Kannel WB, Belanger AJ. Epidemiology of heart failure. Am Heart J 1991;121(3 Pt 1):951-7. 10. McKee PA, Castelli WP, McNamara PM, et al. The natural history of congestive heart failure: the Framingham study. N Engl J Med 1971;285:1441-6. 11. Gillum RF, Fortmann SP, Prineas RJ, et al. International diagnostic criteria for acute myocardial infarction and acute stroke. Am Heart J 1984;108:150-8. 12. Canzanello VJ, Sheps SG. The sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: what's new? What's different? Cardiol Rev 1998;6:272-7.
Yousaf et al 7
13. Rose G, McCartney P, Reid DD. Self-administration of a questionnaire on chest pain and intermittent claudication. Br J Prev Soc Med 1977;31:42-8. 14. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation 2002;106:3143-421. 15. Redfield MM, Jacobsen SJ, Burnett Jr JC, et al. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA 2003;289:194-202. 16. Munagala VK, Jacobsen SJ, Mahoney DW, et al. Association of newer diastolic function parameters with age in healthy subjects: a population-based study. J Am Soc Echocardiogr 2003;16: 1049-56. 17. Redfield MM, Jacobsen SJ, Borlaug BA, et al. Age- and gender-related ventricular-vascular stiffening: a community-based study. Circulation 2005;112:2254-62. 18. Reynolds K, Lewis B, Nolen JD, et al. Alcohol consumption and risk of stroke: a meta-analysis. JAMA 2003;289:579. [Erratum in: JAMA 2003;289:2798]. 19. Kloner RA, Rezkalla SH. To drink or not to drink? That is the question. Circulation 2007;116:1306-17. 20. Hennekens CH, Willett W, Rosner B, et al. Effects of beer, wine, and liquor in coronary deaths. JAMA 1979;242: 1973-4. 21. Wells S, Broad J, Jackson R. Alcohol consumption and its contribution to the burden of coronary heart disease in middle-aged and older New Zealanders: a population-based case-control study. N Z Med J 2004;117:U793. [Erratum in: N Z Med J 2004;117:4]. 22. Camargo Jr CA, Stampfer MJ, Glynn RJ, et al. Moderate alcohol consumption and risk for angina pectoris or myocardial infarction in U.S. male physicians. Ann Intern Med 1997;126: 372-5. 23. Mukamal KJ, Ascherio A, Mittleman MA, et al. Alcohol and risk for ischemic stroke in men: the role of drinking patterns and usual beverage. Ann Intern Med 2005;142:11-9. 24. Mukamal KJ, Chiuve SE, Rimm EB. Alcohol consumption and risk for coronary heart disease in men with healthy lifestyles. Arch Intern Med 2006;166:2145-50. 25. Ajani UA, Gaziano JM, Lotufo PA, et al. Alcohol consumption and risk of coronary heart disease by diabetes status. Circulation 2000;102:500-5. 26. Ammar KA, Samee S, Colligan R, et al. Is self-reported "moderate" drinking in the cardiovascular benefit range associated with alcoholic behavior? A population based study. J Addict Dis 2009;28:243-9. 27. Capasso JM, Li P, Guideri G, et al. Left ventricular dysfunction induced by chronic alcohol ingestion in rats. Am J Physiol 1991;261(1 Pt 2):H212-9. 28. el-Mas MM, Abdel-Rahman AA. Exacerbation of myocardial dysfunction and autonomic imbalance contributes to the estrogen-dependent chronic hypotensive effect of ethanol in female rats. Eur J Pharmacol 2012;679:95-100. 29. McKenna CJ, Codd MB, McCann HA, et al. Alcohol consumption and idiopathic dilated cardiomyopathy: a case control study. Am Heart J 1998;135(5 Pt 1):833-7. 30. Urbano-Márquez A, Estruch R, Fernández-Solá J, et al. The greater risk of alcoholic cardiomyopathy and myopathy in women compared with men. JAMA 1995;274:149-54.
8 Yousaf et al
31. Aguilar D, Skali H, Moyé LA, et al. Alcohol consumption and prognosis in patients with left ventricular systolic dysfunction after a myocardial infarction. J Am Coll Cardiol 2004;43:2015-21. 32. George A, Figueredo VM. Alcoholic cardiomyopathy: a review. J Card Fail 2011;17:844-9. 33. Gavazzi A, De Maria R, Parolini M, et al. Alcohol abuse and dilated cardiomyopathy in men. Am J Cardiol 2000;85:1114-8. 34. Djoussé L, Gaziano JM. Alcohol consumption and risk of heart failure in the Physicians' Health Study I. Circulation 2007;115:34-9. 35. Nicolás JM, Fernández-Solà J, Estruch R, et al. The effect of controlled drinking in alcoholic cardiomyopathy. Ann Intern Med 2002;136:192-200. 36. Cooper HA, Exner DV, Domanski MJ. Light-to-moderate alcohol consumption and prognosis in patients with left ventricular systolic dysfunction. J Am Coll Cardiol 2000;35:1753-9. 37. Klatsky AL, Chartier D, Udaltsova N, et al. Alcohol drinking and risk of hospitalization for heart failure with and without associated coronary artery disease. Am J Cardiol 2005;96: 346-51.
American Heart Journal Month Year
38. U.S. Department of Health Human Services, Public Health Service, National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism. The physician's guide to helping patients with alcohol problems. Washington, DC: National Institutes of Health, NIH publication. 199595-3769. 39. Corrao G, Rubbiati L, Bagnardi V, et al. Alcohol and coronary heart disease: a meta-analysis. Addiction 2000;95:1505-23. 40. Jacobsen SJ, Mahoney DW, Redfield MM, et al. Participation bias in a population-based echocardiography study. Ann Epidemiol 2004;14:579-84. 41. Harford TC, Dorman N, Feinhandler SJ. Alcohol consumption in bars: validation of self-reports against observed behavior. Drinking Drug Pract Surveyor 1976;11:13-5. 42. Giovannucci E, Colditz G, Stampfer MJ, et al. The assessment of alcohol consumption by a simple self-administered questionnaire. Am J Epidemiol 1991;133:810-7. 43. Chaitman BR, Ryan TJ, Kronmal RA, et al. Coronary Artery Surgery Study (CASS): comparability of 10 year survival in randomized and randomizable patients. J Am Coll Cardiol 1990;16:1071-8.
Background Although moderate alcohol consumption is associated with decreased clinical heart failure, there are no population-based studies evaluating the relationship between alcohol consumption and left ventricular (LV) systolic function. We sought to evaluate the relationship between alcohol consumption and LV systolic function in the community. Methods In a population-based random sample of 2,042 adults, age ≥45 years, we assessed alcohol consumption by a self-administered questionnaire. Responders were categorized by alcohol consumption level: abstainer, former drinker, light drinker (b1 drink a day), moderate drinker (1-2 drinks a day), and heavy drinker (N2 drinks a day). Systolic function was assessed by echocardiography. Results We identified 38 cases of systolic dysfunction in 182 abstainers, 309 former drinkers, 1,028 light drinkers, 251 moderate drinkers, and 146 heavy drinkers. A U-shaped relationship was observed between alcohol consumption and moderate systolic dysfunction (LV ejection fraction [LVEF] ≤40%), with the lowest prevalence in light drinkers (0.9%) compared to the highest prevalence in heavy drinkers (5.5%) (odds ratio 0.14, 95% CI 0.04-0.43). This association persisted across different strata of risk factors of systolic dysfunction as well as in multivariate analysis. No significant association between alcohol consumption and systolic function was seen in subjects with LVEF N50% or ≤50%. Conclusions
There is a U-shaped relationship between alcohol consumption volume and LVEF, with the lowest risk of moderate LV dysfunction (LVEF ≤40%) observed in light drinkers (b1 drink a day). These findings are parallel to the relationship between alcohol consumption and cardiovascular disease prevalence. (Am Heart J 2014;0:1-8.)
A U-shaped relationship exists between the level of alcohol consumption and prevalence of clinical heart failure. 1 Regular heavy alcohol consumption is associated with impairment of left ventricular (LV) function 2,3 and occasionally results in overt cardiomyopathy. 4 Alternatively, regular but moderate alcohol consumption is associated with prevention of heart failure. 5,6 Although this relationship between alcohol consumption and clinical heart failure has been extensively studied, the reverse parabolic association between alcohol consumption and ventricular systolic function has not been evaluated in a population-based study. Because both
moderate drinking and heavy drinking are prevalent in the community, we hypothesized that moderate alcohol consumption would be associated not only with reduced prevalence of clinical cardiovascular disease but also with reduced prevalence of systolic dysfunction. We also postulated that heavy alcohol consumption would be associated with increased prevalence of LV systolic dysfunction. This hypothesis was tested in a randomly sampled, population-based cohort of 2,042 people ≥45 years old in Olmsted County, Minnesota.
Methods
From the aAurora Cardiovascular Services, Aurora Sinai/Aurora St. Luke’s Medical Centers, University of Wisconsin School of Medicine and Public Health, Milwaukee, WI, and bMayo Clinic and Foundation, Rochester, MN. Financial support: US Public Health Services, National Institutes of Health, HL-RO1555902. Submitted December 6, 2013; accepted February 25, 2014. Reprint requests: Khawaja Afzal Ammar, MD, Aurora Cardiovascular Services, 2801 W Kinnickinnic River Pkwy, #840, Milwaukee, WI 53215. E-mail: [email protected] http://dx.doi.org/10.1016/j.ahj.2014.02.014 0002-8703 © 2014, Mosby, Inc. All rights reserved.
The Olmsted County Heart Function Study was approved by the Mayo Foundation and Olmsted Medical Center Institutional Review boards, and written informed consent was obtained from the subjects. The present study was carried out with funding from US Public Health Services, National Institutes of Health, grant HL-RO1-555902. The authors are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the paper, and its final contents.
Study setting Of the 124,277 residents of Olmsted County, as of the 2000 census, 90% were white. Other characteristics of this
American Heart Journal Month Year
2 Yousaf et al
population and its unique resources for population-based epidemiological research have been previously described. 7,8
Population sampling and data collection A random sample of residents ≥45 years old as of January 1, 1997, was identified. A 7% sampling fraction was applied within each gender- and age- (5 years) specific stratum. Of the 4,203 subjects invited, 2,042 (48.6%) participated. Enrollment took place from January 1, 1997, to September 30, 2000. Subjects completed a self-administered questionnaire and underwent physical examination, electrocardiography, and echocardiography.
Definitions of selected clinical conditions Medical record review was carried out by 4 trained chart abstractor nurses. The median length of participant medical record archive was 36 years. Heart failure was diagnosed if Framingham criteria were fulfilled. 9,10 Myocardial infarction and hypertension were diagnosed with criteria from the World Health Organization and Sixth Report of the Joint National Committee, respectively. 11,12 Coronary artery disease, angina, diabetes, cerebrovascular accident, transient ischemic attack, and peripheral vascular disease were diagnosed if documented by a physician. Angina was assessed based on a selfadministered questionnaire by Rose et al. 13 Previous smokers were those who had smoked ≥100 cigarettes in the past but were not currently smoking. In accordance with the National Cholesterol Education Program Adult Treatment Panel III criteria, 14 metabolic syndrome was defined by the presence of ≥3 of the following criteria: (1) central obesity defined as a waist circumference ≥102 cm in men and ≥88 cm in women, (2) triglyceride level N150 mg/dL (to convert to millimoles per liter, multiply by 0.0113), (3) high-density lipoprotein cholesterol level b40 mg/dL in men and b50 mg/dL in women (to convert to millimoles per liter, multiply by 0.0259), (4) blood pressure of ≥130/85 mm Hg, and (5) fasting glucose level of ≥110 mg/dL (to convert to millimoles per liter, multiply by 0.0555). Anthropometric measurements were carried out by a trained research nurse with the subjects in the standing position. Body mass index was measured as weight (kilograms)/height (meters). 2
Echocardiographic analysis All subjects underwent echocardiography following a standardized protocol as previously described. 15–17 A single cardiologist, blinded to clinical status, interpreted all echocardiograms.
Measurement of LV systolic function Ejection fraction was measured using M-mode echocardiography, 2-dimensional echocardiography by Simpson’s biplane method, and visual estimate. As LV ejection fraction (LVEF) from these 3 methods was highly correlated and the visual estimate was available in N99% of participants, visual estimate was used in analysis.
Quantification of alcohol consumption Subjects completed a self-administered questionnaire that first asked if they had ever consumed alcohol. A negative answer resulted in classification as an abstainer. Then they were asked if alcohol had been consumed in the last 12 months. A negative answer resulted in classification as a former drinker. An affirmative answer to both questions led the subject to 3 separate sections on beer, wine, and liquor. The 12 frequency categories ranged from 3 to 4 times a day to once a year. The 16 quantity categories ranged from 1 to ≥16 drinks per day. The estimated alcohol content per drink was 12.8 g for beer, 11 g for wine, and 14 g for spirits (liquor or whiskey). Responders were categorized further by alcohol consumption level: light drinker (b1 drink a day, b11 g/d), moderate drinker (1-2 drinks a day, 11-28 g/d), and heavy drinker (N2 drinks a day, N28 g/d).
Statistical analysis Categorical data are summarized as a percentage of the group total with corresponding 95% CI based on normal approximation, and comparison between groups was based on the χ 2 test for association. Continuous variables are summarized as mean ± standard deviation, and comparisons between groups were based on analysis of variance models. Post–analysis of variance comparisons of continuous variables were based on t test, but no adjustments for multiple comparisons were made. Multivariate analysis was carried out with logistic regression for categorical variables and multiple linear regressions for continuous variables. Two-sided P values b.05 were considered significant. Analyses were performed using JMP Genomics, 5.0 (SAS Institute, Cary, NC).
Results Characteristics of study participants Letters of invitations were sent to 4,203 individuals. Of the 2,042 individuals (48.6%) who participated, 1,916 (941 men and 975 women) had valid data and were included in the present analysis. The majority of participants were light drinkers (n = 1,028, 53.7%), with smaller proportions of abstainers (n = 182, 9.5%), former drinkers (n = 309, 16.1%), moderate drinkers (n = 251, 13.1%), and heavy drinkers (n = 146, 7.6%). Table I shows the baseline characteristics of participants categorized by their alcohol consumption habits. Abstainers were more likely to be women, were less likely to be smokers, but were found to have a high prevalence of diabetes mellitus and hypertension. Light drinkers (b1 drinks a day) represented a healthy subset of the population with the lowest mean age, 2% current smokers, 6% diabetic, and 27% hypertensive. Heavy drinkers (N2 drinks a day) had the highest prevalence of current smokers (22%) and mean high-density lipoprotein levels (48.3 ± 16.1). Alcohol consumption and prevalence of clinical cardiovascular disease Table II shows the prevalence of clinical cardiovascular disease (all cardiovascular diseases combined) categorized by alcohol consumption. The prevalence
American Heart Journal Volume 0, Number 0
Yousaf et al 3
Table I. Characteristics of participants categorized by alcohol consumption volume
Age, y Women, n = 975 Body mass index, kg/m 2 Obese (body mass index ≥30) Smoking history Past Current Never Diabetes mellitus Hypertension HDL Metabolic syndrome
Abstainers (n = 182)
Former drinkers (n = 309)
b1 drink/d (n = 1028)
1-2 drinks/d (n = 251)
N2 drinks/d (n = 146)
P
68.0 ± 10.8 138 (75.8) 29.4 ± 6.3 72 (39.6)
63.0 ± 11.2 173 (56.0) 29.1 ± 5.5 121 (39.1)
60.9 ± 10.2 546 (53.1) 28.3 ± 5.4 320 (31.1)
61.6 ± 9.8 89 (35.5) 27.6 ± 4.6 60 (23.9)
62.1 ± 9.6 29 (19.9) 28.5 ± 4.7 47 (32.2)
b.0001 b.0001 b.0001 .0008
22 (12.1) 4 (2.2) 156 (85.7) 23 (12.6) 69 (37.9) 45.4 ± 14.4 84 (45.1)
122 (39.5) 27 (8.7) 160 (51.8) 39 (12.6) 109 (35.3) 43.9 ± 13.7 132 (42.4)
430 (41.8) 78 (7.6) 520 (50.6) 61 (5.9) 280 (27.2) 45.3 ± 14.3 378 (36.7)
134 (53.4) 34 (13.5) 83 (33.1) 17 (6.8) 62 (24.7) 48.2 ± 15.0 84 (33.3)
87 (59.6) 32 (21.9) 27 (18.5) 5 (3.4) 42 (28.8) 48.3 ± 16.1 55 (37.1)
1-2 drinks/d (n = 251)
N2 drinks/d (n = 146)
P
8 (3.2) 33 (13.2) 6 (2.4) 22 (8.8) 3 (1.2) 6 (2.4) 7 (2.8) 43 (17.1)
7 (4.8) 22 (15.1) 3 (2.1) 12 (8.2) 2 (1.4) 2 (1.4) 4 (2.7) 26 (17.8)
.008 .015 .078 .006 .13 .11 .09 b.0001
b.0001 b.0001 .0019 .0021 .046
Data are expressed as mean ± SD or number (percentage). HDL, High-density lipoprotein.
Table II. Prevalence of cardiovascular disease according to alcohol consumption volume Abstainers (n = 182) Myocardial infarction Coronary artery disease Heart failure Angina Cerebrovascular accidents Transient ischemic attack Peripheral vascular disease Any cardiovascular disease
17 30 8 22 4 8 9 47
(9.3) (16.5) (4.4) (12.1) (2.2) (4.4) (4.9) (25.8)
Former drinkers (n = 309) 23 48 12 36 10 8 9 70
(7.4) (15.5) (3.9) (11.7) (3.2) (2.6) (2.9) (22.7)
b1 drink/d (n = 1028) 43 102 17 63 12 15 17 138
(4.2) (9.9) (1.7) (6.1) (1.2) (1.5) (1.7) (13.4)
Data are expressed as number (percentage).
of combined cardiovascular disease indicated a U-shaped relationship (Table II, row 8). When cardiovascular diseases were evaluated individually, the U-shaped relationship persisted, maintaining statistical significance only in myocardial infarction, coronary artery disease, and angina.
Alcohol consumption and systolic function Table III shows the mean LVEF and prevalence of systolic dysfunction categorized by alcohol consumption. There was no clinically meaningful difference in mean LVEF in different alcohol consumption groups. Similarly, there was no statistical association between alcohol consumption group and prevalence of normal systolic function (LVEF N50%) (Table III; row 2) or any systolic dysfunction (LVEF ≤50%) (Table III, row 3). However, when the participants were stratified with respect to moderate systolic dysfunction (LVEF ≤40%) (Table III; row 4), a U-shaped relationship emerged. The prevalence of moderate systolic dysfunction (LVEF ≤40%) was the lowest in light drinkers, with prevalence increasing with increasing or decreasing consumption of alcohol, indi-
cating a U-shaped relationship (Figure 1). In those with LVEF ≤40% (Table III, row 4; Figure 1), the prevalence of systolic dysfunction increased 6-fold between light drinkers (0.9%) and heavy drinkers (5.5%; P b .001). The light drinkers were further stratified into habitual drinkers (those who drink at least once a week) and nonhabitual drinkers (those who drink less than once a week). The prevalence of systolic dysfunction (LVEF ≤50) was lower in habitual drinkers compared to nonhabitual drinkers (4.06% vs 5.91%). A similar trend was seen in the prevalence of moderate systolic dysfunction (LVEF ≤40), which was lower in habitual drinkers (0.24%) compared to nonhabitual drinkers (1.15%). The association between alcohol consumption and moderate systolic dysfunction is further evaluated in Table IV and Figure 2 by stratification of the participants into multiple strata of systolic dysfunction risk factors. The beneficial association was not statistically significant in the young (≤65 years), smokers, nonobese, and those without metabolic syndrome. The U-shaped association between alcohol consumption and LV systolic function persisted across different
American Heart Journal Month Year
4 Yousaf et al
Table III. Left ventricular systolic function association with alcohol consumption volume
LVEF LVEF N50% LVEF ≤50% LVEF ≤40%
Abstainers (n = 182)
Former drinkers (n = 309)
b1 drink/d (n = 1028)
1-2 drinks/d (n = 251)
N2 drinks/d (n = 146)
P
63.9 ± 7.9 172 (94.5) 10 (5.5) 6 (3.3)
62.4 ± 8.0 279 (90.3) 30 (9.7) 6 (1.9)
63.1 ± 6.4 977 (95.0) 51 (5.0) 9 (0.9)
61.8 ± 7.5 232 (92.4) 19 (7.6) 9 (3.6)
61.4 ± 7.9 137 (93.8) 9 (6.2) 8 (5.5)
b.001 .064 .11 b.001
Data are expressed as mean ± (SD) or number (percentage).
systolic function. In the Olmsted County Heart Study, the prevalence of moderate LV systolic dysfunction was the lowest in light drinkers (Table III, row 4; Figure 1), with the prevalence increasing with increasing or decreasing consumption of alcohol, indicating a U-shaped relationship. This relationship remained statistically significant for moderate systolic dysfunction even after adjustment in a multivariate model. This U-shaped relationship between alcohol consumption and LV systolic function was not seen in those with normal systolic function (LVEF N50%) or any impairment of systolic function (LVEF ≤50%). These results enhance our understanding of the beneficial and detrimental effects of alcohol consumption on systolic heart function as measured by echocardiography.
Figure 1
Alcohol intake and prevalence of moderate systolic dysfunction. The lowest prevalence of moderate systolic dysfunction (LVEF ≤40%) was seen in those who consumed b1 drink a day. The prevalence of systolic dysfunction increased 6-fold between light drinkers (0.9%) and heavy drinkers (5.5%).
Table IV. Odds of LV moderate systolic dysfunction according to alcohol consumption volume Unadjusted baseline model Abstainers Former drinkers b1 drink/d 1-2 drinks/d N2 drinks/d
OR (95% CI) 1.00 0.57 0.14 1.56 4.75
(0.12-2.11) (0.04-0.43) (0.39-5.20) (1.18-15.98)
P
.44 b.001 .49 .02
strata of risk factors of systolic dysfunction (Figure 2), including gender, age, hypertension, diabetes, smoking, obesity, and metabolic syndrome. It also persisted in a multivariate model after adjustment for all risk factors (odds ratio [OR] 0.20, 95% CI 0.06-0.64, P b .0001).
Discussion Principal findings This population-based study demonstrated a U-shaped relationship between alcohol consumption and LV
Association of alcohol with clinical cardiovascular diseases Evidence of moderate alcohol consumption reducing clinical cardiovascular events has been shown by numerous observational, 18,19 case control, 20,21 and cohort studies 22,23 both in healthy subjects 24 and in participants with risk factors such as diabetes. 25 In the current study, as previously reported, 26 the results confirm the salutatory effects of alcohol on any clinical cardiovascular disease, whether evaluated individually or collectively in the general population, including myocardial infarction, coronary artery disease, congestive heart failure, angina, cerebrovascular accident, transient ischemic attack, and peripheral artery disease (Table II). This suggests that the current study population is similar to prior studies in terms of association of alcohol consumption with clinical cardiovascular disease, which makes this study population an ideal setting to evaluate an association that has never been described (ie, alcohol and clinical systolic dysfunction). Figure 3 demonstrates the U-shaped relationship between alcohol and clinical heart failure (Table II, row 3; Figure 3). The higher prevalence of diabetes mellitus and hypertension in abstainers may have contributed to higher prevalence of clinical heart failure (Figure 3). Alcohol has been shown to reduce insulin resistance and blood pressure, which also may explain the U-shaped relationship of diabetes mellitus and hypertension with alcohol consumption as seen in our study. Abstainers had much higher odds of clinical congestive heart failure
American Heart Journal Volume 0, Number 0
Yousaf et al 5
Figure 2
Forest plot depicting bivariate stratified analysis of odds of LV moderate systolic dysfunction in light versus heavy drinkers. The association between alcohol consumption and moderate systolic dysfunction, as evaluated by stratification of the participants into multiple strata, persists for most strata in light versus heavy drinkers.
compared to light drinkers (OR 2.7, 95% CI 1.09-6.18, P = .022). When adjusted for both hypertension and diabetes mellitus, the OR dropped meaningfully to 1.9 (95% CI 0.76-4.60, P = .14). This meaningful drop (N20%) in OR indicates that, in abstainers, diabetes mellitus and hypertension were the intervening variables for clinical heart failure.
Association of alcohol with systolic dysfunction Although ejection fraction is a measure of global LV systolic function, LV performance can be further evaluated with multiple other parameters, including measurement of LV dimensions, calculation of LV mass index, regional wall motion abnormalities, as well as diastolic dysfunction. The data on these indices are not relevant to this manuscript because the focus of this study is LVEF. Prior animal (case control) 27,28 and human (case control 29,30 and cohort 3,31) studies have demonstrated heavy alcohol use to be associated with LV systolic dysfunction and alcoholic cardiomyopathy. 32 Among heavy drinkers, reduction in alcohol consumption has shown improvement in LV function. 32,33 In contrast, moderate alcohol consumption is known to protect against heart failure. 1,5,6,34 In studies of patients consuming moderate amount of alcohol, with known heart failure, no deterioration of cardiac
function was seen. 35–37 Our data build on these studies by providing information about systolic function in this population-based sample. It demonstrates light drinkers as having the lowest prevalence of systolic dysfunction with progressive worsening seen with increasing alcohol consumption. Former drinkers were noted to have a lower prevalence of moderate systolic dysfunction (LVEF ≤40%) as compared to moderate drinkers or heavy drinkers. When the entire population was evaluated (Table III, row 1), the Ushaped relationship was statistically significant but not clinically meaningful, with LVEF ranging from 61.4% to 63.9%. No association between alcohol consumption and prevalence of normal systolic function (Table III, row 2) was seen. The U-shaped relationship becomes clearer and more statistically significant in the stratum of participants with LVEF ≤40%, (Table III, row 4) where the prevalence of systolic dysfunction increases from 0.9% to 5.5% in light drinkers as compared to heavy drinkers, a 6-fold rise in prevalence. This suggests that the detrimental effect of alcohol on ventricular function becomes stronger as ventricular systolic function deteriorates. Current National Institutes of Health (National Institute on Alcohol Abuse and Alcoholism) guidelines recommend that patients with alcoholic cardiomyopathy should not drink any alcohol. 38 The findings of this study create an interesting clinical paradox, as those who
6 Yousaf et al
Figure 3
Alcohol intake and prevalence of heart failure. The lowest prevalence of heart failure was seen in those who consumed b1 drink a day.
stand to benefit the most from light alcohol consumption, that is, subjects with LVEF ≤40%, are also those who are likely to be harmed if their consumption exceeds N1 drink a day. Whether those with moderate systolic dysfunction and a proven track record of light alcohol consumption should be encouraged to continue their behavior cannot be proved from these data and requires a prospective study. Current guidelines recommend that, in the community, moderate alcohol consumption (≤1 drink a day for women and elderly and 1-2 drinks a day for men) may reasonably be continued based on existing data on clinical cardiovascular disease. 38 The current study suggests that, at least from the systolic dysfunction perspective, these guidelines may need further consideration and investigation. There is no beneficial association between alcohol consumption and LV systolic function in healthy individuals with LVEF N50%. The association is insignificant in those with LVEF ≤50% and becomes significant only in those with LVEF ≤40%. Prior large clinical studies have suggested a clinical benefit of moderate alcohol consumption on the occurrence of clinical heart failure, but they do not have data on ejection fraction. 1,6,34 Although our study confirms the U-shaped relationship in clinical heart failure (Table II, row 3; Figure 3), it also extends the findings to demonstrate that the beneficial effect is limited to those with moderate systolic dysfunction. The salutatory association between alcohol and moderate systolic dysfunction was not statistically significant in the young (age ≤65 years), smokers, nonobese, and those without metabolic syndrome. Several observational studies 39 have shown similar cardioprotective effects of alcohol in both smokers and nonsmokers, suggesting that alcohol consumption affects heart function similarly in smokers and nonsmokers. The prevalence of LV systolic
American Heart Journal Month Year
dysfunction was almost 1½ times greater in current smokers than in the general population (2.7% vs 1.9%) and more than 2 times greater in current smokers than in noncurrent smokers (2.7% vs 1.3%). Whether the lack of association is due to small sample size or current smoking nullifies the beneficial effect of alcohol can only be speculated. Other strata that failed to show a beneficial association (young, nonobese, and those without metabolic syndrome) in our study cannot be explained and might be a result of either small sample size (n = 38 with systolic dysfunction in the entire cohort) or confounding due to unmeasured variables. Further studies are needed to evaluate these associations.
Strengths and limitations The strengths of the present study include its randomly selected population-based sample, the availability of medical records with a 36-year median length of archive, uniform collection of cross-sectional data, and standardized echocardiographic measurements. Conclusions from this predominantly white cohort including only persons ≥45 years cannot be generalized to the entire US population. Participation bias was evaluated by medical record abstraction of a random selection of 500 participants and 500 invited nonparticipants, which showed no difference in cardiovascular disease. 40 The assessment for alcohol consumption was based on self-reporting and, therefore, may result in misclassification of exposure status. Research participants have been known to underreport alcohol consumption by as much as 39% for men and 44% for women. 41 However, selfadministered questionnaire data reflect the quality of information available to medical practitioners and have been shown to provide useful estimates of alcohol intake in epidemiologic studies. 42 In addition, the ability to make cause-effect inferences and assess time-dependent changes was limited because of the cross-sectional nature of the study. Although the current study has a large sample size (n = 2,042), the study suffers from a lack of power, especially in those with systolic dysfunction. For example, the numbers of patients with moderate systolic dysfunction in the 5 categories are very small (n = 6, 6, 9, 9, and 8 in abstainers, former drinkers, b1 drink a day, 1-2 drinks a day, and N2 drinks a day, respectively). However, it is in this stratum that the statistically significant results are seen. Whether statistically significant results could have been observed in other strata if the sample sizes were greater is debatable. However, the level of statistical significance with P b .001 suggests that, in those with moderate systolic dysfunction, the results are extremely unlikely to have arisen by chance. This is a commonly observed phenomenon in other studies where the maximum benefit from an intervention is secured by those with the worst disease (eg, in the CASS study, 43 the
American Heart Journal Volume 0, Number 0
maximum survival benefit from coronary artery bypass grafting was reaped by those who had moderate systolic dysfunction or severe angina). This observation generates the hypothesis that the maximum benefit from alcohol may be in those with moderate systolic dysfunction.
Conclusion In a community setting, light alcohol consumption (b1 drink a day) was associated with a reduced prevalence of moderate systolic dysfunction (LVEF ≤40%).
Acknowledgements The authors gratefully acknowledge Joe Grundle and Katie Klein for editorial preparation of the manuscript, Brian Miller and Brian Schurrer for their help in preparing figures, and Chi Cho for his help in creating the forest plot.
Disclosures Conflict of interests: The present study was carried out with funding from US Public Health Services, National Institutes of Health, grant HL-RO1-555902. The authors have no competing interests to disclose.
References 1. Bryson CL, Mukamal KJ, Mittleman MA, et al. The association of alcohol consumption and incident heart failure: the Cardiovascular Health Study. J Am Coll Cardiol 2006;48:305-11. 2. Urbano-Marquez A, Estruch R, Navarro-Lopez F, et al. The effects of alcoholism on skeletal and cardiac muscle. N Engl J Med 1989;320:409-15. 3. Kajander OA, Kupari M, Laippala P, et al. Dose dependent but non-linear effects of alcohol on the left and right ventricle. Heart 2001;86:417-23. 4. Laonigro I, Correale M, Di Biase M, et al. Alcohol abuse and heart failure. Eur J Heart Fail 2009;11:453-62. 5. Walsh CR, Larson MG, Evans JC, et al. Alcohol consumption and risk for congestive heart failure in the Framingham Heart Study. Ann Intern Med 2002;136:181-91. 6. Abramson JL, Williams SA, Krumholz HM, et al. Moderate alcohol consumption and risk of heart failure among older persons. JAMA 2001;285:1971-7. 7. Kurland LT, Molgaard CA. The patient record in epidemiology. Sci Am 1981;245:54-63. 8. Melton III LJ. History of the Rochester Epidemiology Project. Mayo Clin Proc 1996;71:266-74. 9. Kannel WB, Belanger AJ. Epidemiology of heart failure. Am Heart J 1991;121(3 Pt 1):951-7. 10. McKee PA, Castelli WP, McNamara PM, et al. The natural history of congestive heart failure: the Framingham study. N Engl J Med 1971;285:1441-6. 11. Gillum RF, Fortmann SP, Prineas RJ, et al. International diagnostic criteria for acute myocardial infarction and acute stroke. Am Heart J 1984;108:150-8. 12. Canzanello VJ, Sheps SG. The sixth report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure: what's new? What's different? Cardiol Rev 1998;6:272-7.
Yousaf et al 7
13. Rose G, McCartney P, Reid DD. Self-administration of a questionnaire on chest pain and intermittent claudication. Br J Prev Soc Med 1977;31:42-8. 14. National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). Third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation 2002;106:3143-421. 15. Redfield MM, Jacobsen SJ, Burnett Jr JC, et al. Burden of systolic and diastolic ventricular dysfunction in the community: appreciating the scope of the heart failure epidemic. JAMA 2003;289:194-202. 16. Munagala VK, Jacobsen SJ, Mahoney DW, et al. Association of newer diastolic function parameters with age in healthy subjects: a population-based study. J Am Soc Echocardiogr 2003;16: 1049-56. 17. Redfield MM, Jacobsen SJ, Borlaug BA, et al. Age- and gender-related ventricular-vascular stiffening: a community-based study. Circulation 2005;112:2254-62. 18. Reynolds K, Lewis B, Nolen JD, et al. Alcohol consumption and risk of stroke: a meta-analysis. JAMA 2003;289:579. [Erratum in: JAMA 2003;289:2798]. 19. Kloner RA, Rezkalla SH. To drink or not to drink? That is the question. Circulation 2007;116:1306-17. 20. Hennekens CH, Willett W, Rosner B, et al. Effects of beer, wine, and liquor in coronary deaths. JAMA 1979;242: 1973-4. 21. Wells S, Broad J, Jackson R. Alcohol consumption and its contribution to the burden of coronary heart disease in middle-aged and older New Zealanders: a population-based case-control study. N Z Med J 2004;117:U793. [Erratum in: N Z Med J 2004;117:4]. 22. Camargo Jr CA, Stampfer MJ, Glynn RJ, et al. Moderate alcohol consumption and risk for angina pectoris or myocardial infarction in U.S. male physicians. Ann Intern Med 1997;126: 372-5. 23. Mukamal KJ, Ascherio A, Mittleman MA, et al. Alcohol and risk for ischemic stroke in men: the role of drinking patterns and usual beverage. Ann Intern Med 2005;142:11-9. 24. Mukamal KJ, Chiuve SE, Rimm EB. Alcohol consumption and risk for coronary heart disease in men with healthy lifestyles. Arch Intern Med 2006;166:2145-50. 25. Ajani UA, Gaziano JM, Lotufo PA, et al. Alcohol consumption and risk of coronary heart disease by diabetes status. Circulation 2000;102:500-5. 26. Ammar KA, Samee S, Colligan R, et al. Is self-reported "moderate" drinking in the cardiovascular benefit range associated with alcoholic behavior? A population based study. J Addict Dis 2009;28:243-9. 27. Capasso JM, Li P, Guideri G, et al. Left ventricular dysfunction induced by chronic alcohol ingestion in rats. Am J Physiol 1991;261(1 Pt 2):H212-9. 28. el-Mas MM, Abdel-Rahman AA. Exacerbation of myocardial dysfunction and autonomic imbalance contributes to the estrogen-dependent chronic hypotensive effect of ethanol in female rats. Eur J Pharmacol 2012;679:95-100. 29. McKenna CJ, Codd MB, McCann HA, et al. Alcohol consumption and idiopathic dilated cardiomyopathy: a case control study. Am Heart J 1998;135(5 Pt 1):833-7. 30. Urbano-Márquez A, Estruch R, Fernández-Solá J, et al. The greater risk of alcoholic cardiomyopathy and myopathy in women compared with men. JAMA 1995;274:149-54.
8 Yousaf et al
31. Aguilar D, Skali H, Moyé LA, et al. Alcohol consumption and prognosis in patients with left ventricular systolic dysfunction after a myocardial infarction. J Am Coll Cardiol 2004;43:2015-21. 32. George A, Figueredo VM. Alcoholic cardiomyopathy: a review. J Card Fail 2011;17:844-9. 33. Gavazzi A, De Maria R, Parolini M, et al. Alcohol abuse and dilated cardiomyopathy in men. Am J Cardiol 2000;85:1114-8. 34. Djoussé L, Gaziano JM. Alcohol consumption and risk of heart failure in the Physicians' Health Study I. Circulation 2007;115:34-9. 35. Nicolás JM, Fernández-Solà J, Estruch R, et al. The effect of controlled drinking in alcoholic cardiomyopathy. Ann Intern Med 2002;136:192-200. 36. Cooper HA, Exner DV, Domanski MJ. Light-to-moderate alcohol consumption and prognosis in patients with left ventricular systolic dysfunction. J Am Coll Cardiol 2000;35:1753-9. 37. Klatsky AL, Chartier D, Udaltsova N, et al. Alcohol drinking and risk of hospitalization for heart failure with and without associated coronary artery disease. Am J Cardiol 2005;96: 346-51.
American Heart Journal Month Year
38. U.S. Department of Health Human Services, Public Health Service, National Institutes of Health, National Institute on Alcohol Abuse and Alcoholism. The physician's guide to helping patients with alcohol problems. Washington, DC: National Institutes of Health, NIH publication. 199595-3769. 39. Corrao G, Rubbiati L, Bagnardi V, et al. Alcohol and coronary heart disease: a meta-analysis. Addiction 2000;95:1505-23. 40. Jacobsen SJ, Mahoney DW, Redfield MM, et al. Participation bias in a population-based echocardiography study. Ann Epidemiol 2004;14:579-84. 41. Harford TC, Dorman N, Feinhandler SJ. Alcohol consumption in bars: validation of self-reports against observed behavior. Drinking Drug Pract Surveyor 1976;11:13-5. 42. Giovannucci E, Colditz G, Stampfer MJ, et al. The assessment of alcohol consumption by a simple self-administered questionnaire. Am J Epidemiol 1991;133:810-7. 43. Chaitman BR, Ryan TJ, Kronmal RA, et al. Coronary Artery Surgery Study (CASS): comparability of 10 year survival in randomized and randomizable patients. J Am Coll Cardiol 1990;16:1071-8.
