Drinking your way to....health? oblivion? cancer? More on 'evidence based' medicine

Perhaps the main concern of our blog is to understand biological causation. Our interests are general, but the issue comes up disproportionately in understanding the findings of medical research, because that is naturally what a large fraction of funding supports.

Well, last week the BBC reported a study in Spain that says that men's health is substantially improved by moderate daily drinking (sorry, women, maybe your turn will come with the next discovery). That is, 3 or so drinks a day reduces heart disease risk by 35 to even 50 percent-- a very substantial difference indeed!

But why did this story make the news? How many times does alcohol consumption have to be studied in regard to health risks? It is included in many, perhaps the vast majority of epidemiological studies, and it has been so for many decades. How could such an effect have been missed? Indeed, how could it possibly be that we don't have solid, irrefutable knowledge by this time? Why would even a Euro cent have to be spent to study its effects any further?

This is highly relevant to the notion of 'evidence based' medicine, because the recommendation about alcohol use bounces around like silly putty. The only thing that is uniformly agreed on is that too much is, well, too much (but the greatest heart disease risk reduction includes those Spanish guys downing 11+ drinks a day).

Here is a case in which culture is part of the nature of 'evidence'. In prudish America, alcohol is considered something so pleasurable as to be necessarily a sin and is studied intensely. It has been controversial whether hospitals should offer patients a dinner glass of wine. Officials dread to recommend drinking at all. So the 'evidence' required to make a recommendation depends on subjective value judgments. But even if one were to be a hard-nosed empiricist, we again ask how we could possibly not know the answers with indisputable rigor.

If this is the nature of evidence, then what evidence do we accept? Is it always the latest study? Why do we think that is any better than the next latest study to come down the pike tomorrow? And if so, why don't we ignore today's study? Why do we think former studies were wrong (some may be identifiably so, but most aren't obviously flawed). Is some aggregate set of studies to be believed? Is it the study that let's business as usual be carried on, for whatever reason?

Our answer is that in addition to the cultural side-issues, there are so many complex factors at play, both causal in regard to what alcohol does in the body and what the body does to it, and in regard to confounding factors, that there is no simple 'truth' and hence it is unclear what counts as 'evidence.' Confounders are factors that may not be known or measured but that are highly correlated with the measured variable of interest (daily alcohol consumption) so that cause itself is hard to identify.Confounders may be causal on their own, or may causally interact with the factor under study.

For example, if the more you drink the more you smoke or the less sleep you get, or the more sex you enjoy (which, since pleasurable, is a sin and must be harmful) these other factors, rather than the alcohol, could be what affect your heart disease risk directly.

Confounders are confoundedly difficult to identify or tease out. Their exposure patterns and even their identity can change with lifestyle changes. And then there are many potentially directly relevant variables, too. When do you drink? What do you drink? With or without olives or a twist of lemon (or salt on the rim)? How uniform is your daily consumption whose average is measured on a survey? Even if these things were known, the future exposure patterns cannot be known, so that today's evidence is really about yesterday's exposures and so the accuracy of future risks based on this evidence is inherently unknowable.

Finally, if drinking is encouraged and heart disease is reduced, will this be good for public health? Or will it increase the number of, say, fatal accidents or violent crimes? Or is it -- well, it is -- a kind of Get-Cancer program? Why? Because if you don't get heart disease you'll live longer and that by itself increases your cancer risk. Not to mention the risk of Alzheimer's, hearing and vision problems, and a host of other older-age problems.

Perhaps the oldest advice in western medical history is from Hippocrates, about 400 BCE. That was "moderation in all things." In today's world, with our romantic notions about the powers of science, such advice is so non-specific and non-technical, that it is considered a cop-out that is not 'evidence-based'. Maybe so, but it's still the best advice. That's because it implicitly includes the unmeasured and unpredictable risk-factor regimes to which people are exposed--and that is evidence.

"Just the facts" sounds like raw empiricism, the kind of rational empiricism our society values. But the 'facts' weave a tangled web.