Wine and Heart Health

[Surreal]

I just noticed that a better variable is simply DRINK. DRINK asks you if you do drink, EVDRINK asks you if you have ever drank alcohol.

Here are the results for “Do you drink?” according to their WORDSUM score.

0 42.7
1 49.7
2 52.6
3 62.4
4 62.4
5 70.2
6 71.7
7 78.7
8 78.7
9 83.9
10 85.7

Astounding.

[KarlGauss]

@Surreal

I agree that causality has not been established. One of the major confounds of alcohol consumption is that it tends to be a proxy for both social class and intellect.

From the General Social Survey, here is the percentage of adults who do not drink according to their WORDSUM (a 10-word basic vocabulary test) score:

1 60.7
2 33.4
3 41.5
4 21.5
5 29.4
6 21.2
7 11.1
8 16.3
9 10.2
10 10.6

As you can see, less intelligent people are far more likely to be teetotalers. Since intelligence is strongly associated better health outcomes across the board, it’s possible that the association might disappear if the studies I listed above controlled for intellect.

That’s very interesting (and instructive)!

[Galileo]

@cgg419

Not that you need another cite at this point, but I find this one particularly interesting

Heavy Drinkers Outlive Nondrinkers, Study Finds - TIME

I’m wondering about the study referenced in the Time article.

The original study is here: Late-Life Alcohol Consumption and 20-Year Mortality.

It seems that all the variables were measured just once, at the beginning of study, with the key variable being how much alcohol the subject was consuming. And, it seems that there was no follow-up measure of consumption throughout the 20-year period.

Also, a “non-drinker” is defined as someone who was not drinking alcohol at the beginning of the study, *but not someone who had never consumed alcohol. *(“Based on the aims of the parent project, lifetime abstainers were excluded.”) The non-drinkers were significantly worse in many ways compared to drinkers but no explanation is given of why:

In fact, in tests of potential covariates, we found among these older adults that at baseline, abstainers were significantly more likely to have had prior drinking problems, to be obese, and to smoke cigarettes than moderate drinkers and significantly higher than moderate drinkers on health problems, depressive symptoms, and avoidance coping. Moreover, at baseline, abstainers were significantly lower than moderate drinkers on SES, physical activity, number of close friends, and quality of friend support and significantly less likely to be married than moderate drinkers. Moreover, all of these covariates significantly predicted mortality.

At baseline, i.e. at the beginning of the 20-year period.

Seems that many of the “non-drinkers” had been heavy drinkers who had stopped before the study began, and, given that there’s no indication of what their alcohol consumption was later on, it’s possible that many of them “fell off the wagon” and went back to being heavy drinkers.

Seems to me that this study is seriously flawed. What am I missing?