Should Patients With Diabetes Drink to Their Health?

In this issue of THE JOURNAL, Valmadrid and other investigators from the Wisconsin Epidemiologic Study of Diabetic Retinopathy (WESDR)¹ report a progressive reduction in risk of coronary heart disease (CHD) death and overall mortality with increasing levels of alcohol consumption, within the mild to moderate range, in patients with older-onset diabetes. This finding complements and extends findings from numerous previous studies that have reported reduced CHD and all-cause mortality in light to moderate drinkers.² Case-control, population, and twin studies have observed a relation of moderate alcohol consumption and reduced insulin resistance (with lower fasting and postprandial glucose and insulin levels).^{3 - 5} Other studies, meanwhile, have found that higher amounts of acute or long-term alcohol ingestion increase insulin resistance,^{6 - 7} as well as triglyceride levels, blood pressure, and all-cause mortality. The present study by Valmadrid et al was unable to evaluate risk with high levels of alcohol consumption, since only 1.8% of subjects consumed more than 3 drinks per day, but the study showed, in the highest consumption category, a discordance of CHD and total mortality benefit.

The importance of these observations for patients with diabetes rests squarely on the apparent benefit on total mortality; indeed, the reported reductions in risk for CHD and total mortality exceed those in most prior studies. A greater absolute benefit actually would be expected in older patients with diabetes compared with unselected populations, because they are at higher baseline risk of disease, comparable with subjects with known cardiac disease.⁸ By contrast, no benefit from alcohol consumption at any level has been identified in men younger than 40 years or premenopausal women.^{9 - 10} Of note, the relative risk reduction observed in the study by Valmadrid et al also exceeded that in prior studies; this finding could occur if alcohol increased insulin sensitivity, directly targeting the mechanism of heightened risk (increased insulin resistance), a viable speculation. Consistent with this possibility is a report by Hein et al¹¹ showing that alcohol-related benefits were most dramatic in the 10% of the population with a blood phenotype linked to the insulin resistance syndrome. However, this mechanism is not supported clearly by the study by Valmadrid et al because inclusion of the available diabetes-associated variables (glycosylated hemoglobin and C-peptide levels) in multivariate models produced no attenuation in the relative risk estimates, thus failing to confirm a mediating role for reduced insulin resistance.

A role for high-density lipoprotein cholesterol (HDL-C) in mediating the effect of moderate alcohol consumption on CHD risk has been consistently demonstrated and appears to explain approximately 50% of the benefit of light to moderate alcohol consumption.^{12 - 13} However, the present study also failed to demonstrate an HDL pathway, despite a "statistically significant" correlation between alcohol and HDL-C. Oddly, neither HDL-C nor any other lipid parameter showed a relationship to CHD mortality in this study, an anomalous finding since lipid levels do predict outcome in type 2 diabetes¹⁴ and lipid lowering with 3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors provides dramatic benefit.¹⁵ It seems possible that the use of insulin and oral hypoglycemic agents, as well as antihypertensive drugs and lipid-lowering medications, obscured the ability to demonstrate an insulin sensitivity or HDL pathway in this cohort. Thrombotic factors also have been touted as possible mediating variables,¹⁶ but apparently were not available in this study.

Could the observationally noted "benefit" from light alcohol consumption be the spurious consequence of high CHD risk in ex-drinkers, who may have quit because they were sick? The authors looked for and failed to find evidence of such a "sick quitter" bias. Of relevance, the authors suggest that abstinence rates in the diabetic cohort are higher than in nondiabetic individuals, hinting that some diabetic patients abstain because they presume drinking may jeopardize their health. This seems reasonable, but abstinence rates in the WESDR study were similar to those in older nondiabetic populations.¹⁷

What does all this mean for the clinician caring for diabetic patients? Previous reports have warned of the dangers of recommending alcohol as a pharmaceutical for cardioprotection.² While evidence is mounting that light to moderate alcohol consumption may be associated with reduced insulin resistance and lower CHD risk, serious problems preclude extrapolating these findings to treatment recommendations. First, abstainers often restrict drinking for a reason, such as family or personal history of alcohol abuse, intolerance, medical contraindications, or other problems (in themselves or family members) potentiated by alcohol; it would seem folly to recommend alcohol to such individuals. Clearly, those who avoid alcohol consumption include an overrepresentation of persons destined for a less favorable risk-benefit ratio were they to drink.

Second, in patients with diabetes, alcohol may both induce and mask potentially severe hypoglycemia by exaggerating hypoglycemic effects caused by other factors (eg, exercise or insulin, sulfonylureas, β-blockers, or other drugs).^{18 - 19} Even at relatively low doses, alcohol may induce hypoglycemia despite low serum insulin levels and high serum glucagon levels.¹⁹ Moreover, heavy alcohol intake may worsen diabetic neuropathy²⁰ and produce or exacerbate insulin resistance.^{6 - 7}

Third, even overall mortality does not adequately address the spectrum of morbidity from alcohol; concerns include, among many others, increased risk of cancer, liver disease, drug interactions, depression, unintentional injuries, and social discord. Public health repercussions not only influence drinkers, but have potent implications for family, associates, and those who share the highway with them—particularly since 50% of drinkers will experience some alcohol-related problem in their lifetime.²¹

Once again, the discussion and the recommendation regarding alcohol intake fall on the shoulders of treating physicians, who, with the increasing demands of managed care, have less time than ever to discuss lifestyle issues (eg, smoking, exercise, alcohol consumption, diet) with their patients. But what is true for most patients with diabetes is true for other patients at high risk for CHD—light to moderate alcohol consumption likely provides benefit, but is contraindicated in anyone who, for whatever reason, cannot restrict his or her drinking to light to moderate levels. For some patients, 1 drink is plenty, 2 is too many, and, unfortunately, 3 is not half enough. As always, judicious recommendations can be made in individual cases when the patient is well known to the clinician, but without a randomized trial, alcohol consumption should not be a general recommendation, whether or not the patient has diabetes.