The Relation of Leisure Activity and Alcohol Consumption to Longevity

A potentially very important article appeared in the European Heart Journal last week. A summary of the findings was published in Physician's First Watch, an internet service that scans the literature in various medical fields. In order to avoid copyright infringement, I shall further summarize and paraphrase their summary. The title, however, pretty much provides a good picture of the findings - unfortunately. That is, it is unfortunate that a massive and costly study could be nearly totally captured in so few words. The title of the Physician's Watch Summary is:

"Physical Activity, Moderate Alcohol Intake Associated with Improved Survival" *

It requires only a slight expansion of these words to convey nearly all of the findings of the study. These were that both "moderate," leisure time physical activity, and "moderate" alcohol intake lowered the risk of ischemic heart disease (heart attacks) and deaths by all causes (as found in a rather large group of Danes). The best results (fewest heart attacks and fewest deaths occurred in the group that both drank moderately and were moderately active. Moderate drinking was defined as 1 to 14 drinks per week, and moderate leisure time activity comprised more than four hours per week of light exercise (such as walking) or 2 to 4 hours per week of more vigorous exercise.

This study took 20 years and involved nearly 20,000 people to begin with, nearly 12,000 of whom completed the entire 20 years, and it managed to minimize the information it could have had. Moreover, although they were aware of this, they didn't seem to know how to prevent it. Let me explain.

The "moderate drinking" group comprised all people who said (yes, it was a questionnaire with little validity check) that they had between one and fourteen drinks per week (inclusive). Anyone who claimed about one drink a week was put into the same group as anyone who claimed fourteen drinks per week. In the latter case, the distribution was not asked; that is the Saturday night binger was lumped with the two glasses of wine per day drinker. Folks, this is an incredible design for a study that took twenty years to complete and involved tens of thousands of people.

For one thing, it has been long established that activity level is associated with heart health. It has been firmly established (for over forty years) that moderate alcohol intake has beneficial effects on the heart, and is inversely related to mortality. These findings were not startling. The fact that they are additive, that they can combine in the same people to give the greatest benefit to those who were both more active and moderate drinkers, while not firmly established, was a reasonable inference. It has also been shown that there is a difference between one and fourteen drinks. Perhaps surprisingly to some, the favor goes to the fourteen drinks - if they are consumed at a steady rate over the week, that is, two per night. Neither the number (from one to fourteen) nor the distribution (all in one night or spread out somehow) was reported, or, presumably, analyzed.

In the discussion section of the study, the authors admitted that data were lost because they combined the groups in order to see the combined effect of both exercise and alcohol. WHAT?! Although spouting statistical lingo as though they understood it, they failed to explain why they didn't retain the information (that which they bothered to gather) and use multivariate statistics, or simple multi-way F distributions to test the interactions, not to mention the simple effects. Possibly they failed in this explanation because there aren't any reasons that actually matter. (For the mathematically inclined among the readers, the multivariate assumptions regarding the underlying distribution of the variables are not met. However, this makes almost no difference in the conclusions.)

We learned something from this study. The size of the advantages they found were interestingly great. They confirmed some of what we already knew and some of what we only thought we knew (that they are additive in effect). But I weep for what we could have learned and didn't. And I wonder why. If there is something I don't understand about this analysis, I beg any reader to explain it to me. I rather hope I am overlooking some explanation. After all, my error after spending less than an hour reading and trying to understand the paper would be trivial compared to what appears to be a colossal oversight. The waste is monumental. Someone please prove me wrong. If the original raw data are intact, I would love to see them and try to glean the information that must be there, somewhere. All in all, for the sake of what we did learn from this study, and in despair over what we did not, I am determined to start drinking again.

* The full citation of the original; article:

European Heart Journal
doi:10.1093/eurheartj/ehm574

The combined influence of leisure-time physical
activity and weekly alcohol intake on fatal
ischaemic heart disease and all-cause mortality

Jane Østergaard Pedersen1,2*, Berit Lilienthal Heitmann2, Peter Schnohr3,
and Morten Grønbæk1