Q. Alcohol, even starting in middle age, is good for your heart per this month's American Journal of Medicine. True?

A. Remember how we all thought that estrogens reduced heart disease? Estrogens improve cholesterol, and a 1985 study of 32,000 women found that those on hormones had 50 percent less heart disease.

But a later study showed estrogens increase heart disease in older women. Why do we believe this one? Because it was a much more accurate randomized, experimental study, while the 1985 study was an observational one that followed two groups of women — those who decided to take estrogen, and those who didn't.

But the women who took estrogen exercised more, ate better, and got better healthcare. These confounding factors were the real reasons they had less heart attacks.

With observational studies, you can never be sure you've discovered all the confounders. In randomized studies, folks are randomly assigned to two groups: One gets the real McCoy and the other gets a fake placebo. After weeks to years, you compare the two groups to uncover any differences. Only randomized studies can eliminate confounders and bias and show the true effects of our potions.

There are no randomized studies of the effect of alcohol on heart disease. Too bad. Like the women who didn't take estrogen, nondrinkers may be sicker than you think.

Dr. George Vaillant of Harvard found that 33 percent of lifelong abstainers had poor childhood health versus 14 percent of moderate drinkers (moderate is an average of two or less drinks a day for men, and one or less a day for women).

Nondrinkers are less likely to have had warm childhood homes. And ill folks may stop drinking. So if nondrinkers have a disproportionate share of sickies with poor memory or weak hearts, then observational studies would find that drinking improves memory or reduces heart disease, when in fact the opposite may be true.

Indeed, Tim Naimi from the Centers for Disease Control and others looked at 30 things that increase heart disease, like diabetes, no health insurance, divorce and physical inactivity, and found 27 to be more common in nondrinkers.

He concluded that the apparent protective effect of moderate alcohol intake on heart disease found in most studies may therefore be false and due to confounding.

Observational studies like the American Journal of Medicine try to correct for confounders, but there'd always be doubt — the 1985 study mistakenly found 50 percent less heart disease even after correcting for obesity, smoking, blood pressure, cholesterol and diabetes.

In short, alcohol and red wine may not be as heart healthy as everyone says.

Dr. Landis Lum is a family-practice physician for Kaiser Permanente and an associate clinical professor at the University of Hawai'i's John A. Burns School of Medicine. Send your questions to: Prescriptions, Island Life, The Advertiser, P.O. Box 3110, Honolulu, HI 96802; islandlife@honoluluadvertiser.com; or fax 535-8170. This column is not intended to provide medical advice.