Critique 153: Moderate alcohol intake may lower the risk of developing diabetes among obese subjects — 11 December 2014

Wakabayashi U.  Light-to-Moderate Alcohol Drinking Reduces the Impact of Obesity on the Risk of Diabetes Mellitus.  J Stud Alcohol Drugs 2014;75:1032–1038.

Author’s Abstract

Objective:  Light-to-moderate alcohol drinking has been shown to reduce the risk of type 2 diabetes, for which obesity is a primary risk factor.  The aim of this study was to determine whether drinking alcohol influences the relationship between obesity and hyperglycemia.

Method:  The relationships of adiposity indices with hyperglycemia were compared among middle-aged Japanese men (N = 12,627) who were non-, light-to-moderate (<22 g ethanol/day), heavy (≥22 and <44 g ethanol/day), and very heavy (≥44 g ethanol/day) drinkers.

Results:  There were significant positive correlations of hemoglobin A1c with body mass index (BMI) and waist-to-height ratio (WHtR), which were significantly weaker in light-to-moderate and heavy drinkers than in nondrinkers but were not significantly different in very heavy drinkers compared with nondrinkers. Odds ratios (ORs) for hyperglycemia in subjects with versus those without high BMI or WHtR were significantly higher than reference level of 1.00 in all the drinker groups and significantly lower in light-to-moderate and heavy drinkers compared with nondrinkers; however they were not significantly different in very heavy drinkers compared with nondrinkers.  ORs of the interaction term consisting of alcohol drinking and high adiposity index were significantly lower than the reference level in the light-to-moderate and heavy drinkers (OR with 95% confidence interval: high BMI, 0.61 [0.41, 0.91] in light-to-moderate drinkers and 0.64 [0.48, 0.85] in heavy drinkers; high WHtR, 0.57 [0.38, 0.85] in light-to-moderate drinkers and 0.66 [0.50, 0.88] in heavy drinkers) but were not significantly different from the reference level in very heavy drinkers (high BMI, 0.90 [0.65, 1.25]; high WHtR, 1.04 [0.74, 1.46]).

Conclusions: The associations between obesity and hyperglycemia were weaker in light-to-moderate drinkers than in nondrinkers.  Thus, light-to-moderate drinking may reduce the impact of obesity on the risk for diabetes.

Forum Comments

An inverse association between moderate alcohol and the risk of developing diabetes mellitus has been known for many decades.  In 1988, Stampfer et al reported lower risk of diabetes among moderate drinkers; lower rates of obesity among moderate drinkers explained some, but not all, of the beneficial effects of alcohol.  Howard et al provided a good summary of the research on alcohol and diabetes in 2004, giving an estimate of 33% to 56% lower incidence of diabetes for consumers of 1-3 drinks/day than was seen for non-drinkers.  A meta-analysis by Koppes et al indicated that for a wide range of alcohol intake (from about ½ to more than 3 drinks/day), the relative risk of diabetes for drinkers was about 30% lower than it was for abstainers.  A more recent study by Rasouli et al (2013) suggested a lower risk of both type 1 and type 2 diabetes for moderate drinkers.

The author of the present paper states that “The purpose of this study was to determine whether the relationship between adiposity and glycemic status differs among nondrinkers and different categories of drinkers.  Hemoglobin A1c and the obesity-related indices (body mass index [BMI] and waist-to-height ratio [WHtR]) were used to evaluate glycemic status and adiposity, respectively.”  Only subjects who stated that they consumed alcohol “every day” were considered as “drinkers;” subjects reporting up to 22 g/day of alcohol were considered to be “light-to-moderate,” ≥ 22 – <44 g/day to be “”heavy,” and ≥ 44 g/day to be “very heavy” consumers of alcohol.

Key findings of the this article include lower associations between measures of obesity (a high BMI and waist-to-height ratio) and hemoglobin A1C  (an index of long-term hyperglycemia) for subjects in the “light-to-moderate” and the “heavy” alcohol consumption groups than were seen for non-drinkers.  For “very heavy” drinkers, these associations were not significantly different from those of non-drinkers.  In other words, drinkers of less than the largest amounts of alcohol did not have values of A1c as high as did non-drinkers of the same body size.  Reviewer Thelle drew attention to the physical activity variable evaluated as a confounder in this study.  “The subjects are split into two categories, one with regular exercise every day for 30 minutes or more, and the rest.  Thus, the physical activity variable used was very crude and only distinguished between those with very high activity and everyone else.  There may be residual confounding as those with moderate alcohol may be exercising more than the abstainers, and thereby protected against diabetes and hyperglycemia.” 

Analytic methodology and interpretation:  The author presented data in two formats: (1) the risk of “hyperglycemia” in categories of alcohol consumption according to whether the subjects were obese or not; and (2) the specific levels of A1c according to quartiles of BMI and weight/height ratio in relation to alcohol intake.  Forum members Djoussé and Zhang had some concerns about the conclusions of the author from the first type of analysis, when discussing the relative risk of hyperglycemia according to alcohol intake.  Djoussé commented: “The conclusion of the author of this paper stated, ‘The associations between obesity and hyperglycemia were weaker in light-to-moderate drinkers than in nondrinkers.  Thus, light-to-moderate drinking may reduce the impact of obesity on the risk for diabetes.’ 

Djoussé continued: “The data presented from the initial analyses do not support the conclusion of a ‘weaker’ effect of obesity on hyperglycemia to be associated with alcohol consumption.  The author based this on results of stratified analysis by levels of alcohol intake with 4 different reference groups (one for each alcohol level).  However, in each stratum, the relative risk of hyperglycemia among subjects with normal BMI may not be the same; thus, the obtained ORs are not directly comparable as the ratio is very sensitive to the denominator, which varied in each group.”  A similar issue can also be applied when assessing effect of high WHtR on risk of hyperglycemia according to status of alcohol consumption.

Forum member Zhang continued: “Unlike the comparisons of ‘hyperglycemia’ in relation to alcohol, the findings in the second type of analysis on the effects of alcohol on the relation of BMI and WHtR to specific levels of hemoglobin A1C are more interesting.  The data clearly show that values of hemoglobin A1c are much lower among drinkers than non-drinkers, regardless of the BMI (or WHtR) quartile; this is likely to make the absolute difference in hemoglobin A1c levels among BMI (or WHtR) categories within the same alcohol category less impressive than that among non-drinkers.  (Such relationships could have affected the findings in the first type of analysis.)  However, differences in hemoglobin A1c among different BMI or WHtR quartile groups among alcohol drinkers are more easily interpreted and probably have greater clinical implications.”

Reviewer Ellison contends that the second analytic approach presented in this paper can support two conclusions: (1) at all levels of obesity, A1c levels are lower among drinkers than non-drinkers; and (2) at all levels of drinking, A1c levels are markedly lower for leaner subjects than more obese subjects.  The message that might be derived from these associations is that to prevent diabetes, you should be as lean as possible and consume some alcohol.”  Forum member Finkel added: “Diabetes is a very mean disease. Any measure that might assuage its ravages is worth considering.  The other possibly helpful lifestyle alterations (such as losing weight) are very difficult to maintain for most of us.”

Reviewer Skovenborg commented: “The results of the present study are not surprising but nevertheless interesting because this is the first study showing a difference in the relationship between adiposity and glycemic status in drinkers and nondrinkers.  Logistic regression analyses show that the associations between adiposity indices and glycemic status were significantly weaker in light-to-moderate and heavy drinkers than in nondrinkers.  However, the cross-sectional design of the study does not allow conclusions regarding a causal relationship.” 

Definition of “moderate” drinking:  Skovenborg also commented on the best definition of “moderate” drinking.  “The cut-off point in this study for ‘heavy’ alcohol consumption (≥ 22 g ethanol/day) may be somewhat low.  My guess is that a consumption between ≥ 22 g ethanol/day and < 44 g ethanol/day (up to about one half bottle of wine per day) would be considered as moderate by most men from the Mediterranean countries.”   

De Gaetano commented: “In a meta-analysis by Costanzo et al (2011) from 12 studies reporting separate data on wine or beer consumption in relation to cardiovascular disease, two closely overlapping dose-response curves were obtained (maximal protection of 33% at 25 g/day of alcohol).  Another meta-analysis on cardiovascular mortality in patients with cardiovascular disease showed a J-shaped pooled curve with a significant maximal protection (average 22%) by alcohol at approximately 26 g/day.  In that study, light to moderate alcohol consumption (from 5 to 25 g/day) was significantly associated with a lower incidence of cardiovascular and all-cause mortality (Costanzo et al, 2010).  Low levels of alcohol intake should be considered separately for men and women and in different countries.  In a very large meta-analysis on alcohol dosing and total mortality (Di Castelnuovo et al), 1-2 drinks per day for women and 2-4 drinks per day for men were inversely associated with total mortality.

Potential mechanisms of effect of alcohol on risk of diabetes:  Forum member Van Velden stated: “We found in our research on the influence of moderate alcohol consumption on the symptoms and signs of the Metabolic  Syndrome (MS) that alcohol increases insulin secretion in the same way as the suphonylureas.  This resulted in a decrease in the serum glucose concentration.  Moderate alcohol consumption has a beneficial effect on type II DM.  It is not clear how this affects body weight, because hyperinsulinaemia may actually increase the transformation of glucose into fat and store it by preventing it from being used for energy.  Many people with MS have insulin resistance that standard medications cannot reverse.  Insulin is thus an inflammatory hormone and a fat storage hormone, which we wish to keep as low as possible, having only the bare minimum amount necessary to remove glucose from the blood stream and turn it into energy.”  

Other research:  Reviewer Stockley points out that a previous study of the effects of alcohol among obese subjects had similar results as those reported in the present paper.  “Dixon et al showed in a cross-sectional study of 486 very obese subjects that alcohol consumers (N = 276) showed a marked reduction in the adjusted odds ratio of type 2 diabetes (odds ratio = 0.29; 95% confidence interval, 0.16 to 0.55) compared with rare or non-consumers (N = 210).  There was a U-shaped relationship between the amount and frequency of alcohol consumption and fasting triglyceride, fasting glucose, hemoglobin A1c, and an index of insulin resistance measurements.  Consumers of < 100 g/week of alcohol had more favorable measures.”  The authors of that paper (Dixon et al) concluded that “Light-to-moderate alcohol consumption is associated with a lower prevalence of type 2 diabetes, reduced insulin resistance, and more favorable vascular risk profile in the severely obese.  We would propose that light to moderate alcohol consumption should not be discouraged in the severely obese.”

References from Forum review

Costanzo S, Di Castelnuovo A, Donati MB, Iacoviello L, de Gaetano G.   Alcohol Consumption and Mortality in Patients with Cardiovascular Disease: A Meta-Analysis.  J Am Coll Cardiol 2010;55:1339-1347.

Costanzo S, Di Castelnuovo A, Donati MB, Iacoviello L, de Gaetano G.  Wine, beer or spirit drinking in relation to fatal and non-fatal cardiovascular events: a meta-analysis.  Eur J Epidemiol 2011;26:833–850.

Di Castelnuovo A, Costanzo S, Bagnardi V, Donati MD, Iacoviello L, de Gaetano G.  Alcohol Dosing and Total Mortality in Men and Women. Arch Intern Med 2006;166:2437-2445.

Dixon JB, Dixon ME, O’Brien PE.  Alcohol consumption in the severely obese: relationship with the metabolic syndrome. Obes Res 2002;10:245-252.

Howard AA, Arnsten JH, Gourevitch MN. Effect of alcohol consumption on diabetes mellitus: a systematic review. Ann Intern Med 2004;140:211-219.

Koppes LL, Dekker JM, Hendriks HF, Bouter LM, Heine RJ. Moderate alcohol consumption lowers the risk of type 2 diabetes: a meta-analysis of prospective observational studies. Diabetes Care 2005;28:719-725.

Rasouli B1, Ahlbom A, Andersson T, Grill V, Midthjell K, Olsson L, Carlsson S.  Alcohol consumption is associated with reduced risk of Type 2 diabetes and autoimmune diabetes in adults: results from the Nord-Trøndelag health study.  Diabet Med 2013;30:56-64. doi: 10.1111/j.1464-5491.2012.03713.x.

Stampfer MJ, Colditz GA, Willett WC, et al: A prospective study of moderate alcohol drinking and risk of diabetes in women.  Am J Epidemiol 1988;128:549-558.

Forum Conclusions

This case-control study compared the risk of having a high level of hemoglobin A1c, a measure of hyperglycemia, and of the waist/height ratio, a measure of obesity, according to reported alcohol consumption.  Overall, subjects reporting an average alcohol intake between 22 and 44 g/day (the equivalent of between about 2 and 4 typical drinks) had lower levels of hemoglobin A1c and lower weight/height values.  The author concludes: “The associations between obesity and hyperglycemia were weaker in light-to-moderate drinkers than in nondrinkers.  Thus, light-to-moderate drinking may reduce the impact of obesity on the risk for diabetes.”

While Forum reviewers had some concerns about the first analytic approach given in the paper, they agreed that the second approach, in which specific levels of A1c were reported according to alcohol intake and measures of obesity, had merit.  These latter data clearly indicate that hyperglycemia is affected both by obesity (which increases the risk) and alcohol intake (which decreases the risk); hyperglycemia is the leading factor associated with the development of diabetes mellitus. 

A therapeutic implication of this study may be that obese subjects who are moderate drinkers should not be advised to stop their alcohol consumption.  The results, along with a large amount of other research currently available, also suggest that obese subjects who do not drink and have no contraindications to alcohol consumption should be informed that consuming moderate amounts of alcohol may be associated with a reduction in their risk of developing diabetes. 

Comments on this paper have been provided by the following members of the International Scientific Forum on Alcohol Research:

Giovanni de Gaetano, MD, PhD, Department of Epidemiology and Prevention, IRCCS Istituto Neurologico Mediterraneo NEUROMED, Pozzilli, Italy

Luc Djoussé, MD, DSc, Dept. of Medicine, Division of Aging, Brigham & Women’s Hospital and Harvard Medical School, Boston, MA, USA

R. Curtis Ellison, MD, Section of Preventive Medicine & Epidemiology, Boston University School of Medicine, Boston, MA, USA

Harvey Finkel, MD, Hematology/Oncology, Boston University Medical Center, Boston, MA, USA

Erik Skovenborg, MD, Scandinavian Medical Alcohol Board, Practitioner, Aarhus, Denmark

Creina Stockley, PhD, MBA, Clinical Pharmacology, Health and Regulatory Information Manager, Australian Wine Research Institute, Glen Osmond, South Australia, Australia. 

Dag S. Thelle, MD, PhD, Senior Professor of Cardiovascular Epidemiology and Prevention, University of Gothenburg, Sweden; Senior Professor of Quantitative Medicine at the University of Oslo, Norway

David Van Velden, MD, Dept. of Pathology, Stellenbosch University, Stellenbosch, South Africa

Andrew L. Waterhouse, PhD, Marvin Sands Professor, Department of Viticulture and Enology, University of California, Davis; Davis, CA, USA

Yuqing Zhang, MD, DSc, Epidemiology, Boston University School of Medicine, Boston, MA, USA