Critique 182: A major meta-analysis on the association of alcohol consumption with the risk of diabetes mellitus — 8 March 2016

Li X-H, Yu F-F, Zhou Y-H, He J.  Association between alcohol consumption and the risk of incident type 2 diabetes: a systematic review and dose-response meta-analysis.  Pre-publication: Am J Clin Nutr 2016; doi: 10.3945/ajcn.115.114389.

Authors’ Abstract

Background: Previous cohort studies have shown that moderate alcohol consumption was associated with a lower risk of type 2 diabetes (T2D).  However, whether these associations differ according to the characteristics of patients with T2D remains controversial.

Objective: The purpose of this study was to explore and summarize the evidence on the strength of the association between alcohol consumption and the subsequent risk of T2D by using a dose-response meta-analytic approach.

Design: We identified potential studies by searching the PubMed, Embase, and Cochrane Library databases up to 24 March 2015.  Prospective observational studies that evaluated the relation between alcohol consumption and the risk of T2D and reported its effect estimates with 95% CIs were included.

Results: Analyses were based on 706,716 individuals (275,711 men and 431,005 women) from 26 studies with 31,621 T2D cases.  We detected a nonlinear relation between alcohol consumption and the risk of T2D, which was identified in all cohorts (P-trend, 0.001, P-nonlinearity, 0.001), in men (P-trend, 0.001, P-nonlinearity, 0.001), and in women (P-trend, 0.001, P-nonlinearity, 0.001).  Compared with the minimal category of alcohol consumption, light (RR: 0.83; 95% CI: 0.73, 0.95; P, 0.005) and moderate (RR: 0.74; 95% CI: 0.67, 0.82; P, 0.001) alcohol consumption was associated with a lower risk of T2D. However, heavy alcohol consumption had little or no effect on subsequent T2D risk.  Furthermore, the summary RR ratio (RRR; male to female) of the comparison between moderate alcohol consumption and the minimal alcohol categories for T2D was significantly higher, and the pooled RRR (current smoker to never smoker) of light alcohol consumption was significantly reduced.

Conclusions: Light and moderate alcohol consumption was associated with a lower risk of T2D, whereas heavy alcohol consumption was not related to the risk of T2D.

Forum Comments

Most prospective cohort studies (e.g., Stampfer et al, de Vegt et al, Wannamethee et al) have consistently shown that light-to-moderate drinkers are at a lower risk of developing type 2 diabetes mellitus (DM) than are non-drinkers or heavy drinkers.  There have also been meta-analyses and reviews that support beneficial effects on the risk of DM of moderate alcohol consumption (Koppes et al, Baliunas et al, Carlsson et al).  The present meta-analysis is based on a very large sample-size (more than 700,000 subjects with over 31,000 cases of DM) which has allowed the investigators to not only evaluate the overall main effect but to carry out a number of sub-group analyses for estimating the effects of alcohol on DM risk.

In the present study, after an extensive review of the literature and exclusion of studies with incomplete or inappropriate information, data from a total of 26 observational studies were used.  Eleven studies included in the meta-analysis were from the USA, 9 from Europe, 4 from Asia, and 2 from Australia.  As with any meta-analysis, there were problems with heterogeneity among papers.  The authors used appropriate statistical methods for attempting to put each study’s assessments of alcohol into consumption categories of “light” (≤ 12 g/day), “moderate” (>12- <24 g/day), and “heavy” (≥ 24 g/day).  The referent group was made up of the lowest alcohol intake category in each study, but it appears that in all but 2 studies (in which the comparison group was “none”), the investigators used “never drinkers” for the referent group, in other words removing ex-drinkers from such a group.

Overall, the study reports that light and moderate drinkers had a significantly reduced risk of developing DM.  For light drinkers, in comparison with non-drinkers, the overall risk ratio for DM was 0.83, with 95% CIs of 0.73, 0.95 (P = 0.005).  For moderate drinkers the RR was 0.74, with 95% CIs of 0.67, 0.82 (P = 0.001).  Thus, data from this meta-analysis indicate a 17% and 26% reduction in the risk of DM for these two drinking categories, respectively.  For subjects classified as heavy users of alcohol, the RR was 0.98, with 95% CIs of 0.83, 1.09, P = 0.480), interpreted as no effect.  In a figure in the paper showing the separate results for each individual study included, with only one or two exceptions the point estimates for the risk of DM were 1.0 or less than 1.0 for light and moderate drinkers.  Thus, data from almost all of the studies support the overall finding of a decrease in risk of DM for light or moderate drinkers.

The authors also carried out analyses to evaluate potential confounding or modification by sex, age, BMI, smoking, physical activity, and family history of DM.  In each sub-category, the point estimate of the risk ratio associated with light or moderate alcohol consumption was less than 1.0, adding further to the robustness of the overall conclusions of a reduction in risk of DM from alcohol intake.  Similarly, for all subjects as well as for men and women separately, there was a clear U-shaped curve for the association between alcohol and DM.  The nadir of the effect (the strongest protective effect) was just over 20 grams of alcohol per day (about 2 typical drinks), and even the risk for heavy drinkers did not reach that of non-drinkers.

Specific comments on the study by Forum members:  Forum member Ellison noted: “The data presented indicate that almost all of the separate studies included in the meta-analysis had adjusted their results for BMI or waist circumference.  Many previous studies have shown that light-to-moderate drinkers have less obesity than non-drinkers, which raises the possibility that the authors have adjusted for one of the mechanisms by which alcohol reduces the risk of DM.  If this is the case, the adjusted estimates of effect of alcohol on DM found in this paper could be under-estimates of the true effect.”  Reviewer Skovenborg agreed that this was a well-done meta-analysis that confirmed the expected outcomes.  He also noted: “It is somewhat surprising to me that the relation between alcohol consumption and risk of type 2 diabetes differs according to smoking status.”

Reviewer Finkel stated: “Although we’ve seen these results before, this redux is welcome in the face of the massive and still increasing numbers of diabetics among us.  DM is a mean disease, which viciously challenges the cardiovascular system and creates other havocs.  As for patients who have already developed diabetes, endocrinologists of my acquaintance and the conservative American Diabetes Association agree that moderate alcohol consumption is beneficial to such patients.  I have long held that diabetics, more than most, need that extra edge, as long as they remember to eat (so as to avoid hypoglycemia).  It is shocking to continue to encounter physicians so ill-informed or incorrectly convinced that diabetes is a contraindication to even modest quantities of alcohol.”

Forum member de Gaetano agreed that some physicians are not aware that the protective effects of moderate alcohol are especially important for patients who already have diabetes.  A number of randomized clinical trials have confirmed beneficial effects of moderate drinking among diabetics, most recently studies by Shai et al and Gepner et al.

Forum member Ursini stated: “I agree with the comments from other Forum members that there is a strong a positive effect of alcohol (better wine with foods) on diabetes; this is supported by what we learned from basic science.  There is a problem, however, with getting this message to the medical establishment and the public.  Many of the diabetologists I know refuse to accept this evidence and keep telling their patients that alcohol is a poison, particularly for diabetics.  My disappointing feeling is that this view is based on something other than scientific evidence.”

Reviewer Stockley and others pointed out the marked increase in the occurrence of diabetes worldwide, and added: “Almost 300 people in Australia are newly diagnosed with diabetes every day, and the incidence is projected to approximately double in the next 15 years.  This adds to the importance of providing simple messages to the public as to what constitutes a healthy diet and lifestyle that may reduce the risk of this disease.”  Forum members agreed that, for most adults, moderate alcohol consumption should be considered as a component of such a healthy lifestyle.

Are there differences in effect according to the type of alcoholic beverage consumed?   The authors did not provide results separately according to the type of alcoholic beverage consumed.  This is unfortunate, as people generally do not drink alcohol per se, but various beverages that contain alcohol.  Forum member Van Velden thought that this was a well done meta-analysis, stating that their own studies on this topic supported similar conclusions.  “We could not explain the reason for the lower incidence associated with alcohol intake, but in our studies it was red wine intervention that stimulated insulin secretion that resulted in lower blood glucose levels.”

Reviewer Estruch had similar concerns: “This is a well-performed meta-analysis that confirms the protective role against developing diabetes of the light-to-moderate intake of alcoholic beverages; this is probably due to an increase in insulin sensitivity.  However, I miss in the discussion whether this protective effect should be attributed to alcohol (ethanol) or other components of alcoholic beverages (for instance, polyphenols).  In our groups we have performed short-term randomized clinical trials (e.g., Chive-Blanch et al) exploring the different effects of alcohol and non-alcoholic compounds of wine on insulin sensitivity.  We have found that fasting glucose remained constant throughout the studies, while mean adjusted plasma insulin and HOMA-IR decreased after both red wine and dealcoholized red wine. These results support a beneficial effect of the non-alcoholic fraction of red wine (mainly polyphenols) on insulin resistance, conferring greater protective effects on cardiovascular disease from red wine than from other alcoholic beverages.  With beer, the effect was not so clear, perhaps because beer contains a lesser content of polyphenols.”

Forum member Mattivi noted: “I am reluctant to accept the odd concept that alcohol is one of the most widely consumed beverages; ethanol should rather be considered as an important ingredient of beverages having a rather different composition.  Can we consider ethanol, from a mechanistic point of view, as the sole active agent?  Very likely ethanol is a main player here, but at least in some beverages there are several other compounds with reported anti-diabetic activity.  I do hope in the future to see meta-analyses considering the separate effects of the different alcoholic beverages.”  Forum reviewer Lanzmann-Petithory agreed strongly that wine has shown greater protection against DM, and many other chronic diseases, than have other types of beverages containing alcohol.

References from Forum Review

Baliunas DO, Taylor BJ, Irving H, Roerecke M, Patra J, Mohapatra S, Rehm J.  Alcohol as a risk factor for type 2 diabetes: a systematic review and meta-analysis. Diabetes Care 2009;32:2123–2132.

Carlsson S, Hammar N, Grill V.  Alcohol consumption and type 2 diabetes: meta-analysis of epidemiological studies indicates a U-shaped relationship.  Diabetologia 2005;48:1051–1054.

Chiva-Blanch G, Urpi-Sarda M, Ros E, Valderas-Martinez P, Casas R, Arranz S, Guillén M,  Lamuela-Raventós RM, Llorach R, Andres-Lacueva C, Estruch R.  Effects of red wine polyphenols and alcohol on glucose metabolism and the lipid profile: A randomized clinical trial.  Clin Nutr 2013;32:200-206 DOI: http://dx.doi.org/10.1016/j.clnu.2012.08.022.

de Vegt F, Dekker JM, Groeneveld WJ, Nijpels G, Stehouwer CD, Bouter LM, Heine RJ. Moderate alcohol consumption is associated with lower risk for incident diabetes and mortality: the Hoorn Study. Diabetes Res Clin Pract 2002;57:53–60.

Gepner Y, Golan R, Harman-Boehm I,  . . .  Stampfer MJ, Shai I, et al.  Effects of Initiating Moderate Alcohol Intake on Cardiometabolic Risk in Adults with Type 2 Diabetes.  A 2-Year Randomized, Controlled Trial.  Ann Intern Med 2015;163:569-579. doi:10.7326/M14-1650

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.

Shai I, Wainstein J, Harman-Boehm I, Raz I, Fraser D, Rudich A, et al.  Glycemic effects of moderate alcohol intake among patients with type 2 diabetes: a multicenter, randomized, clinical intervention trial. Diabetes Care 2007;30:3011-3016. [PMID: 17848609]

Stampfer MJ, Colditz GA, Willett WC, Manson JE, Arky RA, Hennekens CH, Speizer FE.  A prospective study of moderate alcohol drinking and risk of diabetes in women.  Am J Epidemiol 1988;128:549–558.

Wannamethee SG, Camargo CA Jr., Manson JE, Willett WC, Rimm EB.  Alcohol drinking patterns and risk of type 2 diabetes mellitus among younger women. Arch Intern Med 2003;163:1329–1336.

Forum Summary

Most previous studies have shown that consumers of light-to-moderate amounts of alcoholic beverages tend to have a significant reduction in their subsequent risk of developing Type II diabetes mellitus (DM).  The purpose of the present study was to explore and summarize the evidence on the strength of the association between alcohol consumption and the subsequent risk of DM by using a dose-response meta-analytic approach.  The authors identified 26 prospective cohort studies providing data appropriate for a meta-analysis; their analyses were based on 706,716 individuals (275,711 men and 431,005 women) with 31,621 cases of DM.

This meta-analysis reports that light and moderate drinkers have a significantly reduced risk of developing DM.  For “light” drinkers (defined as an average of ≤ 12 g/day of alcohol), in comparison with non-drinkers, the overall risk ratio for DM was 0.83, with 95% CIs of 0.73, 0.95 (P = 0.005).  For “moderate” drinkers (>12- <24 g/day), the RR was 0.74, with 95% CIs of 0.67, 0.82 (P = 0.001).  Thus, data from this meta-analysis indicate a 17% and 26% reduction in the risk of DM, respectively, for these two drinking categories.

For subjects classified as “heavy” users of alcohol (reported intake averaging ≥ 24 g/day), the RR was 0.98, with 95% CIs of 0.83, 1.09, P = 0.480), interpreted as no effect.  In a figure in the paper showing the separate results for each individual study included, the point estimates for the risk of DM were 1.0 or less than 1.0 for light and moderate drinkers in essentially all studies.  Thus, data from the individual studies support the overall finding of a decrease in risk of DM for light or moderate drinkers.

Sub-group analyses showed that when subjects were stratified by sex, age, BMI, smoking, physical activity, and family history of DM, the point estimates of the risk ratios associated with light or moderate alcohol consumption were less than 1.0 in all groups, adding further to the robustness of the overall conclusions of a reduction in risk of DM from alcohol intake.  Similarly, for all subjects, as well as for men and women separately, there was a clear U-shaped curve for the association.  The nadir of the effect was just over 20 grams of alcohol per day (about 2 typical drinks), and even the risk of heavy drinkers did not reach the estimated risk of non-drinkers.

Forum members considered this to be a well-done analysis that confirms most previous results from prospective studies indicating a reduction in the risk of developing DM associated with moderate drinking.  Further, an increasing number of randomized clinical trials are supporting such beneficial effects on the development and clinical treatment of DM.   The Forum thought it unfortunate that beverage-specific results were not available in this study, as increasingly it is being shown that, beyond alcohol effects, there are polyphenols and other substances in wine and beer that provide additional protection against diabetes.  Further, the Forum felt it important to also emphasize the protective effects against cardiovascular disease among subjects who already have DM, who are especially vulnerable to coronary heart disease and other effects of atherosclerosis.

Overall, this meta-analysis based on a large number of subjects indicates that the risk of DM is considerable lower among light and moderate drinkers than among abstainers.  This finding supports the contention that, for most middle-aged and older adults (with the exception of individuals with specific prohibitions against alcohol such as former drug or alcohol abuse, certain types of neurological or severe hepatic disease, etc.), moderate alcohol consumption should be considered as a component of a “healthy lifestyle” that reduces the risk of diabetes and most of the chronic diseases of ageing.

                         *                    *                  *                      *                      *                      *     

Comments on this critique by the International Scientific Forum on Alcohol Research were provided by the following members:

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

R. Curtis Ellison, MD, Professor of Medicine & Public Health, Boston University School of Medicine, Boston, MA, USA

Ramon Estruch, MD, PhD.  Associate Professor of Medicine, University of Barcelona, Spain

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

Dominique Lanzmann-Petithory,MD, PhD, Nutrition/Cardiology, Praticien Hospitalier Hôpital Emile Roux, Paris, France

Fulvio Mattivi, MSc, Head of the Department of Food Quality and Nutrition, Research and Innovation Centre, Fondazione Edmund Mach, in San Michele all’Adige, Italy

Erik Skovenborg, MD, specialized in family medicine, member of the Scandinavian Medical Alcohol Board, Aarhus, Denmark

Diewertje Sluik, DrPH, Division of Human Nutrition, Wageningen University, NL.

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

Arne Svilaas, MD, PhD, general practice and lipidology, Oslo University Hospital, Oslo, Norway

Fulvio Ursini, MD, Dept. of Biological Chemistry, Universityof Padova, Padova, Italy

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

Andrew L. Waterhouse, PhD, Department of Viticulture and Enology, University of California, Davis, USA