An interesting controversy that has simmered in the medical literature for years is the effect on health and survival of moderate alcohol consumption. Epidemiological studies consistently show that moderate drinkers live longer than abstainers, while heavy drinkers fair the worst. Before we discuss possible mechanisms for this effect, we have to consider whether or not the effect is real – and here controversy remains.

One of the reasons for the controversy is that the data supporting a protective effect of moderate alcohol is all observational, rather than experimental. It is simply not possible to do a study in which subjects are randomly assigned to either no, moderate, or heavy alcohol consumption (especially in a blinded fashion). Therefore researchers must be content with mere correlation. This is useful but very problematic evidence.

The weakness of observational evidence like this is confounding factors. Since people self-select their own alcohol consumption there may be many other lifestyle choices that go along with that and affect health and survival. There may even be genetic types that lead people to drink and also confer greater longevity. Without randomization the possible confounding factors are enormous.One factor that initially skewed the data was including former drinkers in the non-drinker group. Former drinkers include heavy drinkers that have already damaged their health. More recent studies control for this, and the apparent beneficial effect of light or moderate drinking remains (although is much smaller) when ex-drinkers are excluded from the non-drinker group.

The evidence is a bit mixed, and depends on how the data is examined. Various reviews show that there is a mild protective effect for coronary heart disease (CHD) for mild to moderate regular alcohol consumption, and no protective effect for occasional alcohol consumption. There is no apparent benefit for stroke. However, all-cause mortality is higher in some studies in the light to moderate alcohol group, even when CHD is slightly lower. Heavy drinking is consistently associated with higher mortality.

It does appear that alcohol may increase HDL (good cholesterol) but it is not clear if this is enough to provide any reduction in CHD.

Also, there does not appear to be any benefit to taking up light to moderate drinking in middle age, or for men who have CHD.

Type of drink probably does not matter. While some studies have shown that wine is associated with lower mortality than beer, review of shopping receipts show that wine purchasers also purchase more healthful food items, while beer buyers also tend to buy fatty and salty foods.

While there is room for a range of interpretations, my take on the totality of research is that there may be a small beneficial effect to CHD (through raising HDL or some other mechanism) to light to moderate alcohol consumption, but this benefit has not been clearly established. Further, any such small effect is likely counterbalanced by many negative health effects associated with regular alcohol consumption. And in any case there is no evidence that for any individual taking up regular alcohol consumption is of any health benefit. Further, I think the small effects being seen in these studies are overshadowed by possible confounding factors and the weaknesses of observational studies.

Other lifestyle changes, like quitting smoking and regular exercise, have a well-established and much larger effect than any possible benefit from alcohol.

Into this mix now comes another study which is getting a lot of press exposure. This is a study of late life (55-65 yo) drinking and 20 year mortality. The researchers found:

Controlling only for age and gender, compared to moderate drinkers, abstainers had a more than 2 times increased mortality risk, heavy drinkers had 70% increased risk, and light drinkers had 23% increased risk. A model controlling for former problem drinking status, existing health problems, and key sociodemographic and social-behavioral factors, as well as for age and gender, substantially reduced the mortality effect for abstainers compared to moderate drinkers. However, even after adjusting for all covariates, abstainers and heavy drinkers continued to show increased mortality risks of 51 and 45%, respectively, compared to moderate drinkers.

The strengths of this study are that the authors tried to account for as many confounding factors as they could, and also did a long follow up. They acknowledge that confounding factors have a huge effect of the outcomes, but there remains a significant effect even when confounders are removed. However, in my opinion this study demonstrates how large the effect of confounders can be and all it would take is one or two missed confounders, or improperly accounting for known confounders, to produce the results seen.

Taken at face value, these results indicate that moderate regular alcohol consumption had the lowest mortality, followed by light drinking, then heavy drinking, and then non-drinkers. This last bit is the surprising result, as prior studies have always shown that heavy drinkers have the highest mortality. Of course, that is also the bit that press reports latched onto – heavy drinkers outlive non-drinkers. The effect is tiny once confounders are removed (51% increase in mortality over moderate drinkers for abstainers and 45% for heavy drinkers), but the effect remains.

Given the totality of existing research this is the part of this study that to me says the results are not reliable. It is far easier to believe that there are missing confounding factors in this study or some systematic bias in data analysis rather than that heavy drinkers outlive non-drinkers, despite a large amount of prior research that says the opposite.

In the conclusion the authors focus on the apparent advantage of moderate drinking and ignore the difference between non-drinkers and heavy drinkers, probably because they realize this is a small and likely artifactual result.

Even after this study I am left unconvinced that there is any advantage to consuming moderate amounts of alcohol. I think we are still likely dealing with an artifact of observational studies. But even if there is a small health benefit (which would be from decreased CHD), there does not appear to be any benefit to taking up drinking, and so this has no implications to lifestyle choices. Meanwhile the potential downside to increased alcohol consumption (despite this one study) is well established. Public health measures would be better off focusing on better established and larger lifestyle factors – quitting smoking, diet, and exercise.