Author + information
- Received March 27, 2001
- Revision received August 22, 2001
- Accepted August 31, 2001
- Published online December 1, 2001.
- Mihaela Tanasescu, MD*,* (, )
- Frank B Hu, MD*,
- Walter C Willett, MD*,†,‡,
- Meir J Stampfer, MD*,†,‡ and
- Eric B Rimm, ScD*,†
- ↵*Reprint requests and correspondence: Dr. Mihaela Tanasescu, Department of Nutrition, Harvard School of Public Health, 665 Huntington Avenue, Boston, Massachusetts 02115 USA
The goal of this study was to examine the relationship between alcohol intake and risk of coronary heart disease (CHD) among men with type 2 diabetes.
Type 2 diabetes is associated with an increased risk of CHD. Emerging evidence suggests that moderate alcohol intake is associated with an important reduction in risk of CHD in individuals with type 2 diabetes.
We studied 2,419 men who reported a diagnosis of diabetes at age 30 or older in the Health Professionals’ Follow-up study (HPFS). During 11,411 person-years of follow-up after diagnosis, we documented 150 new cases of CHD (81 nonfatal myocardial infarction [MI] and 69 fatal CHD). Relative risks (RR) were estimated from pooled logistic regression adjusting for potential confounders.
Alcohol use was inversely associated with risk of CHD in men with type 2 diabetes. The age-adjusted RRs corresponding to intakes of ≤0.5 drinks/day, 0.5 to 2 drinks/day and >2 drinks/day were 0.76 (95% confidence interval: [CI]: 0.52 to 1.12), 0.64 (95% CI: 0.40 to 1.02) and 0.59 (95% CI: 0.32 to 1.09), respectively, as compared with nondrinkers (p for trend = 0.06). When we controlled for body mass index, smoking, family history of MI, hypertension, hypercholesterolemia, duration of diabetes, physical activity level, vitamin E supplements and intake of trans fat, polyunsaturated fat, fiber and folate, RRs were 0.78 (95% CI: 0.52 to 1.15), 0.62 (95% CI: 0.40 to 1.00) and 0.48 (95% CI: 0.25 to 0.94) (p for trend = 0.03). The benefits of moderate consumption did not statistically differ by beverage type.
Moderate alcohol consumption is associated with lower risk of CHD in men with type 2 diabetes.
Atherosclerosis is the most frequent complication of type 2 diabetes and the underlying cause of death in approximately 80% of patients with diabetes (1). Insulin resistance, hypertension, dyslipidemia, glucose intolerance, hyperinsulinemia, dysfibrinolysis and endothelial dysfunction may contribute to the macrovascular complications of diabetes.
The association between moderate alcohol drinking and coronary heart disease (CHD) in the general population is well established (2). Four recent studies point to reductions in CHD risk associated with moderate alcohol intake in individuals with type 2 diabetes (3–6). However, the importance of beverage type and the influence of possible confounders like diet, obesity or concurrent health conditions have not been adequately addressed. This study investigated the association between alcohol consumption and CHD among men with type 2 diabetes in the Health Professionals’ Follow-up study (HPFS).
The HPFS study is a prospective cohort study of 51,529 men aged 40 to 75 in 1986. The current study included 2,419 men who reported a physician’s diagnosis of diabetes mellitus at the age of 30 or older on any questionnaire from 1986 through 1994. Men who reported a diagnosis of diabetes were mailed a supplementary questionnaire that included ascertained prediagnostic symptoms, clinical testing that led to diagnosis and current treatment regimen. Cases were considered confirmed if any of the following were met: one or more classic symptoms plus a raised fasting (≥7.8 mmol/l) or random (≥11.1 mmol/l) plasma glucose concentration; at least two raised plasma glucose concentrations on separate occasions (fasting, ≥7.8 mmol/l or random, ≥11.1 mmol/l, or 11.1 mmol/l after 2 h or more on glucose tolerance testing) in the absence of symptoms; treatment with a hypoglycemic drug: insulin or oral hypoglycemic agent (7). Because the cases of diabetes for this analysis were diagnosed before 1994, we used the diagnostic criteria in place for that time period and not the more recent classifications (8). Validity of self-reported diabetes was verified in a subsample of 71 men from the HPFS cohort. Of these, 12 had incomplete records while the diagnosis of type 2 diabetes was confirmed in 57 (97%) of the remaining 59.
Average daily consumption of beer, wine and spirits was included in the semiquantitative food frequency questionnaire in 1986, 1990 and 1994. For each beverage, a commonly used portion size was specified, and the participants were asked how often, on average, over the past year they consumed that amount. Standard portion sizes were specified as a can/bottle or glass for beer, 4-oz glass for wine and one drink or shot for liquor. A previous study found this questionnaire to measure alcohol intake with a high degree of validity (9). The intake derived from the questionnaire and that from multiple week diet records were highly correlated (Spearman r = 0.86), while mean intakes were very similar. Also, serum high-density lipoprotein cholesterol (HDL-C) levels were significantly correlated with alcohol intakes estimated from the questionnaire among men in the HPFS.
CHD end points
Fatal CHD and nonfatal myocardial infarction (MI) occurring between the return of the 1986 questionnaire and 1996 were the end points of the analysis. Self-reported MIs were confirmed by reviewing medical records if they met World Health Organization criteria: characteristic symptoms with either typical electrocardiographic changes or elevations of cardiac enzymes. Probable cases of MI (no available records but confirmation of hospitalization and information from telephone interview/letter representing 0.9% of cases) were also included in the analysis after ensuring that results were not appreciably different from the definite cases.
Deaths were reported by next of kin, work associates and postal authorities. In case of persistent nonresponse, the National Death Index was used to identify deceased cohort members. Fatal CHD was confirmed by reviewing medical records or autopsy reports with the permission of the next of kin. The cause listed on the death certificate was not sufficient alone to confirm a coronary death. Sudden deaths (i.e., death within 1 h of symptom onset in a man without known disease that could explain death) were included in the fatal MI category. For subjects with multiple end points, follow-up ended with onset of the first event.
Person-months of follow-up were accumulated for patients with diabetes starting with the date of return of the 1986 questionnaire for men diagnosed with diabetes at enrollment or with the return of the questionnaire on which type 2 diabetes was first reported until the occurrence of a CHD end point, death or January 31, 1996. We excluded subjects with a history of CHD (MI, angina or coronary revascularization), stroke or cancer reported at baseline. Patients with a prior MI and cancer were subsequently excluded from analyses.
Men were divided into four categories of alcohol intake: nondrinkers, ≤0.5 drinks/day, 0.5 to 2 drinks/day and >2 drinks/day. Relative risks (RR) were calculated as the CHD incidence rate for a given category of alcohol intake compared with the nondrinkers category. Tests for trend were calculated by assigning the median value to increasing categories of alcohol use. P values of less than 0.05 were considered statistically significant. Relative risks were initially calculated adjusting for age. Multivariate logistic regression was then used to estimate RRs of CHD over each two-year follow-up interval using the alcohol intake reported on the most recent questionnaire, adjusting for other potential confounders: physical activity (quintiles of metabolic equivalents spent weekly), body mass index (BMI) (≤23 kg/m2, 23 to 24.9, 25 to 29.9, ≥30), smoking (never smoker, <10, 10 to 24, 25 to 44, 45 to 64, 65+ pack years), history of hypertension, high cholesterol, family history of MI, use of vitamin E supplements and dietary intake of trans fat, polyunsaturated fat (percent of total energy intake), fiber, folate and total calories. We additionally controlled for insulin and oral hypoglycemic drug use in analyses on men with definite and symptomatic type 2 diabetes. Person-years with missing information for alcohol intake due to an unreturned questionnaire were not considered in the analysis. Alcohol’s interaction with CHD risk factors (smoking, BMI, hypertension, family history of MI and hypercholesterolemia) were assessed by the likelihood ratio statistic (difference in deviance or in −2 Log Likelihood) between the model containing the interaction of alcohol with each risk factor and the main effects model following a chi square distribution with 2 degrees of freedom.
Subjects with type 2 diabetes in the cohort had predominantly light-to-moderate intake of alcohol, with 39% of person-time accounted for by men with no alcohol intake, 31% with ≤0.5 drink/day, 20% between 0.5 and 2 drinks/day and only 10% drinking >2 alcoholic drinks/day. The median intake in this highest category was 2.8 drinks/day.
Baseline characteristics of prevalent cases of diabetes are shown in Table 1. Compared with the reference group (nondrinkers), men who drank alcohol tended to have a lower intake of fiber, a lower proportion of people with family history of MI and a higher proportion of current smokers, people with hypercholesterolemia and hypertensives.
During 11,411 person-years of follow-up, we documented 150 new cases of CHD (81 nonfatal MI and 69 fatal CHD). The age-adjusted rate RRs across categories of alcohol consumption (≤0.5 drinks/day, 0.5 to 2 drinks/day and >2 drinks/day) were 0.76 (95% confidence interval [CI]: 0.52 to 1.12), 0.64 (95% CI: 0.40 to 1.02) and 0.59 (95% CI: 0.32 to 1.09), respectively, as compared with nondrinkers (p for trend = 0.06). This inverse association became slightly stronger after adjusting for potential covariates: 0.78 (95% CI: 0.52 to 1.15), 0.62 (95% CI: 0.40 to 1.00), 0.48 (95% CI: 0.25 to 0.94) (p for trend = 0.03) (Table 2). To assess the effect of alcohol intake in different subgroups of type 2 diabetics, data were stratified by smoking status, BMI, presence of hypertension, family history of MI and presence of hypercholesterolemia. The inverse association was observed in all subgroups (Table 3). Formal testing for interaction showed that family history of MI marginally modified the effect of alcohol intake on CHD risk (p = 0.09). In further analyses, when nondrinkers without family history of MI were considered in the reference group, men with a family history of MI had a risk ratio for CHD of 3.13 (95% CI: 1.80 to 5.43) if they drank no alcohol, of 1.71 (95% CI: 0.85 to 3.45) if they drank ≤0.5 drinks/day and 0.65 (95% CI: 0.23 to 1.84) if they had an intake of more than 0.5 drinks/day. This interaction awaits confirmation in other populations.
To determine if “sick quitters” were contributing to higher rates of CHD among the abstainers, we conducted analyses excluding 637 nondrinkers who reported a reduction in alcohol intake on the baseline questionnaire or who had a lifetime exposure of more than 50 alcoholic drinks. After exclusions, the multivariate RRs were 0.64 (95% CI: 0.36 to 1.15), 0.53 (95% CI: 0.28 to 1.00) and 0.42 (95% CI: 0.18 to 0.95) for intakes of ≤0.5 drinks/day, 0.5 to 2 drinks/day and >2 drinks/day, respectively.
Intakes of beer, wine or liquor were each inversely associated with CHD (Table 4). The multivariate RRs corresponding to an intake of liquor of ≤0.5 drinks/day and greater than 0.5 drinks/day were 0.78 (95% CI: 0.52 to 1.16) and 0.48 (95% CI: 0.27 to 0.86) (p for trend = 0.02). The corresponding RRs for beer were 0.71 (95% CI: 0.48 to 1.04) and 0.48 (95% CI: 0.20 to 1.14) (p for trend = 0.11) and the RRs for wine were 0.70 (95% CI: 0.48 to 1.02) and 0.75 (95% CI: 0.37 to 1.52) (p for trend 0.34). The observed effects did not statistically differ by beverage type.
When we examined rates of CHD within each category of total alcohol consumption among nondiabetic men, the age-adjusted incidence of MI was approximately 2.5 to 4.5 times less than that observed in type 2 diabetics. Alcohol was associated with a greater absolute decrease in risk subjects with type 2 diabetes than it was in nondiabetics. Using the nondrinkers nondiabetics as a reference group, we cross-classified men by their alcohol and diabetes status (Fig. 1). In multivariate analyses, nondiabetics who drank >2 drinks/day had the lowest RR for CHD (RR = 0.56, 95% CI: 0.44 to 0.70). Increasing consumption of alcohol among subjects with diabetes decreased risk of CHD. At an alcohol intake of 0.5 to 2 drinks/day and >2 drinks/day, subjects with diabetes did not have a significantly elevated risk of CHD in comparison with nondiabetic nondrinkers (RR = 1.17, 95% CI: 0.79 to 1.74 and RR = 1.14, 95% CI: 0.68 to 1.91, respectively).
Summary of findings, strengths and limitations
Moderate alcohol intake was associated with a lower risk of CHD in men with type 2 diabetes, beer, wine and liquor all contributing to this association. The inverse association was not confounded by other independent risk factors for CHD: duration of diabetes, physical activity, BMI, current smoking, presence of hypertension, high cholesterol, family history of MI, vitamin E supplements and intakes of trans fat, polyunsaturated fat, fiber and folate.
The prospective design reduces the potential for bias from differential reporting of alcohol between cases and noncases. Furthermore, the availability of updated measures of alcohol intake and covariates over the follow-up period enabled us to incorporate changes in lifestyle characteristics into the analysis. One limitation of this study is the ascertainment of diabetes status through self-report and the fact that some cases may have been diagnosed during routine screening. To determine if screening bias altered our results, we conducted subanalyses of the main association among men who had definite diabetes (n = 1,513) as identified by responses on the supplementary questionnaire and among men who had diabetic symptoms at the time of diagnosis (n = 942). The multivariate RRs of CHD across categories of alcohol consumption (≤0.5 drinks/day, 0.5 to 2 drinks/day and >2 drinks/day) were 0.72 (95% CI: 0.43 to 1.20), 0.43 (95% CI: 0.20 to 0.89), 0.38 (95% CI: 0.13 to 1.07) among the subset of definite subjects with diabetes and 0.84 (95% CI: 0.45 to 1.56), 0.59 (95% CI: 0.24 to 1.45), 0.46 (95% CI: 0.11 to 2.01) among subjects with diabetes with symptoms at diagnosis.
Previously, we found that alcohol intake was reported with a high degree of validity as shown on a subsample of this cohort (9). However, underreporting by some respondents at the higher end of consumption may still have occurred, in which case the range of benefit may be somewhat underestimated.
Review of other epidemiologic studies
Our results are consistent with those of the four recent studies that focused on the effect of alcohol on occurrence of MI (4), cardiovascular death (3,5)and both fatal and nonfatal CHD (6)among patients with diabetes.
In a population of 983 men and women with type 2 diabetes aged 68.6 ± 11.0 years in the Wisconsin Epidemiologic Study of Diabetic Retinopathy, Valmadrid et al. (3)found important reductions of mortality from CHD among individuals consuming <2, 2 to 13 and 14 g or more alcohol/day, compared with never drinkers: 0.54 (95% CI: 0.33 to 0.90), 0.44 (95% CI: 0.23 to 0.84) and 0.21 (95% CI: 0.09 to 0.48), respectively. Former drinkers’ risk of death from CHD tended to be lower than that of never drinkers, but this difference was not statistically significant. The authors controlled for severity of diabetes using glycosylated hemoglobin level and plasma C-peptide. After controlling for HDL-C in a subgroup of 451 patients, alcohol intake of more than 2 g/day was still significantly associated with a lower risk of CHD, which suggests that additional cardioprotective mechanisms may be operating (e.g., decreases in fibrinogen and increases in insulin sensitivity).
In a study by Solomon et al. (4), 5,103 women in the Nurses’ Health study who reported type 2 diabetes at age 30 or older were followed from 1980 to 1994. Risk ratios for nonfatal or fatal CHD were 0.72 (95% CI: 0.54 to 0.96) and 0.45 (95% CI: 0.29 to 0.68) for those who reported 0.1 to 4.9 g alcohol/day and ≥5 g/day, respectively, compared with nondrinkers.
Similarly, among 2,790 men in the Physicians’ study with diabetes at baseline Ajani et al. (6)found risk reductions for CHD death of 1.11 (95% CI: 0.66 to 1.89), 0.67 (95% CI: 0.42 to 1.07), 0.42 (95% CI: 0.23 to 0.77) (p for trend = 0.002) corresponding to alcohol levels of monthly, weekly and daily as compared with rarely/never drinkers. In a subsample of 510 diabetics followed for CHD incidence, RR were 0.84 (95% CI: 0.46 to 1.54), 0.75 (95% CI: 0.45 to 1.26), 0.66 (95% CI: 0.38 to 1.16) for the same categories of intake.
Among 133 male patients with diabetes in the World Health Organization Multinational Study of Vascular Disease in Diabetes, Diem et al. (5)found a significant inverse association (odds ratio = 0.90, 95% CI: 0.81 to 1.0) for an intake of 16 g to 30 g alcohol/day, compared with nondrinkers. Lower and higher intakes were not associated with reduced risk.
Potential mechanisms underlying the association
The inverse association between alcohol and CHD, documented in over 40 prospective studies of healthy populations, is probably in large part achieved through the direct effects of ethanol on increasing HDL-C (2,10). This mechanism may be even more important among diabetics, as low levels of HDL-C are a common feature of type 2 diabetes.
Diabetic hyperglycemia leads to oxidative stress in the vessel wall through formation of advanced glycation end products (AGE) and enhanced binding to the receptors for AGE (11,12). Alcohol may lower the production of AGE through its metabolite acetaldehyde (13).
Also, type 2 diabetes is associated with decreased fibrinolysis attributable to elevated concentrations of plasminogen activator inhibitor 1 (PAI-1) (14). In vitro studies show that alcohol downregulates the expression of the PAI-1 gene (15). In free living populations, alcohol may be positively associated with plasma levels of the thrombolytic tissue type plasminogen activator (t-PA), independent of HDL-C levels (16). Additionally, platelet aggregation is inhibited by ethanol (17,18), which may be particularly important in patients with type 2 diabetes who are prone to increased platelet aggregability. Finally, fibrinogen is increased in type 2 diabetes (19), yet moderate alcohol consumption is associated with lower levels of fibrinogen (2,20).
The effects of moderate alcohol consumption on insulin sensitivity may also partly explain the benefit we found. Although increased insulin resistance is induced by heavy drinking, light and moderate alcohol intake (1 to 3 drinks/day) is associated with increased insulin sensitivity (21,22). During prolonged fasting, alcohol may induce hypoglycemia and hyperketonaemia (23), whereas, during the fed state, alcohol may induce hyperglycemia (24). Despite these possibilities, multiple studies show no alteration of glucose homeostasis when moderate alcohol is consumed with meals (25,26). It has also been shown that moderate alcohol consumption is associated with decreased risk of diabetes (27).
Risk and benefit of alcohol consumption among diabetics
Although heavy drinking is associated with diabetic neuropathy (28)and retinopathy (29), moderate alcohol intake poses little, if any, additional risk for these complications (30,31). Other alcohol-related risks particularly relevant for this population are hypertension and stroke. Available data do not support an increased risk of hypertension with moderate drinking in the general population (32), and recent evidence suggests moderate drinking is inversely associated with total stroke, mostly through reductions in the more common ischemic stroke (33,34).
The results of this study are consistent with previous reports in other populations showing CHD risk reduction due to light and moderate alcohol intake among patients with type 2 diabetes. The risk reduction between drinkers and nondrinkers was stronger among subjects with diabetes than it was among nondiabetics in our population, albeit the confidence intervals were larger. Nonetheless, recommending alcohol to abstainers is still controversial. The benefit of reduced CHD needs to be weighed against the potential for risk among populations consuming more than 1 drink/day. In healthy populations, light-to-moderate alcohol intake is associated with reduced total mortality (35,36). In our sample of men with diabetes, the RRs of death from all causes for intakes of ≤0.5 drinks/day, 0.5 to 2 drinks/day and >2 drinks/day were 0.72 (95% CI: 0.54 to 0.97), 0.67 (95% CI: 0.47 to 0.95) and 0.62 (95% CI: 0.40 to 0.97) as compared with nondrinkers. Also, the RRs for any cardiovascular death were 0.73 (95% CI: 0.48 to 1.11), 0.69 (95% CI: 0.42 to 1.14) and 0.53 (95% CI: 0.27 to 1.05) for the same categories of intake. It has been repeatedly shown that people in these age groups are less likely than younger individuals to abuse alcohol, binge drink or drive under its influence (37). Light-to-moderate drinking with meals may be an attractive and beneficial lifestyle component for the patient with diabetes. Those with personal or family history of abuse, intolerance, psychiatric problems, contraindications or other alcohol-related problems should certainly be advised not to drink. It remains for the clinician to decide whether to include alcohol recommendations as part of the treatment plan for patients with type 2 diabetes. For most, the benefits would likely outweigh the risks. Studies assessing the effect of moderate alcohol intake on biological markers associated with CHD in patients with diabetes would help in generating recommendations to be included in diabetes treatment guidelines.
☆ Supported by research grants CA 55075, HL 35464 and AA 11181 from the National Institutes of Health. Dr. Hu’s work is supported, in part, by an American Diabetes Association Research Award.
- advanced glycation end products
- body mass index
- coronary heart disease
- confidence interval
- high-density lipoprotein cholesterol
- Health Professionals’ Follow-up study
- myocardial infarction
- plasminogen activator inhibitor 1
- relative risk
- tissue type plasminogen activator
- Received March 27, 2001.
- Revision received August 22, 2001.
- Accepted August 31, 2001.
- American College of Cardiology
- Rimm E.B,
- Williams P,
- Fosher K,
- Criqui M,
- Stampfer M.J
- Solomon C.G,
- Hu F.B,
- Stampfer M.J,
- et al.
- Diem P,
- Fajfr R,
- Mueller B,
- et al.
- Ajani U.A,
- Gaziano J.M,
- Lotufo P.A,
- et al.
- National Diabetes Data Group
- Expert Committee on the Diagnosis and Classification of Diabetes Mellitus
- Giovannucci E,
- Colditz G,
- Stampfer M.J,
- et al.
- Al-Abed Y,
- Mitsuhashi T,
- Li H,
- et al.
- Sobel B.E,
- Woodcock-Mitchell J,
- Schneider D.J,
- Holt R.E,
- Marutsuka K,
- Gold H
- McKenzie C.R,
- Eisenberg P.R
- Facchini F,
- Chen Y.D,
- Reaven G.M
- Kiechl S,
- Willeit J,
- Poewe W,
- et al.
- Burge M.R,
- Zeise T.M,
- Sobhy T.A,
- Rassam A.G,
- Schade D.S
- Koivisto V.A,
- Tulokas S,
- Toivonen M,
- Haapa E,
- Pelkonen R
- Rimm E.B,
- Chan J,
- Stampfer M.J,
- Colditz G.A,
- Willett W.C
- Bell D.S
- Young R.J,
- McCulloch D.K,
- Prescott R.J,
- Clarke B.F
- Mitchell B.D,
- Vinik A
- Camargo C.A,
- Rimm E.B
- Hillborn M,
- Juvela S