Is Good Health Merely a Guessing Game? Writer Gary Taubes argues that much of what we've been told about what is or isn't good for us is based on guesswork. His article "Do We Know What Makes Us Healthy?" appears in this Sunday's edition of the New York Times Magazine. Taubes talks with Madeleine Brand about his article.
NPR logo

Is Good Health Merely a Guessing Game?

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript
Is Good Health Merely a Guessing Game?

Is Good Health Merely a Guessing Game?

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript


Back now with DAY TO DAY. I'm Robert Smith.


And I'm Madeleine Brand.

We in the media love to report science news, especially if it's health-related; you know, the story that starts off a new study by the such and such journal says doing X will make you live longer. But there always seems to be another study a little later on that says the opposite, that X won't make you live longer. Garlic, red wine, aspirin, all once touted as beneficial, now maybe not. Well, why does that happen?

Here with an answer is writer Gary Taubes, whose cover story in this coming Sunday's New York Times Magazine is called "Unhealthy Science." And Gary Taubes is here now.

Welcome to DAY TO DAY.

Mr. GARY TAUBES (Author, "Good Calories, Bad Calories: Challenging the Conventional Wisdom on Diet, Weight Control and Disease"): Thank you. It's nice to be here.

BRAND: Well, the unhealthy science that you're referring to is epidemiology and the big epidemiological study that's in a lot of this health news comes from the Nurses Health Study. Now, describe what that is.

Mr. TAUBES: This was a study founded at Harvard in the 1970s and it follows some 120,000 nurses around the country. And they send them questionnaires every two years, dozens of pages about what do you eat, what drugs are you taking. And they record their illnesses and when they've had heart attacks and what cancers they've had. And then they try to draw these associations between some aspect of diet and lifestyle or prescription drug use and the diseases they have. And the fatal flaw is that they can tell you that, for instance, that eating monounsaturated fat associates with a lower risk of heart disease, but they can't tell you that eating monounsaturated fat causes you to have less heart disease.

BRAND: So there could be all sides of other factors.

Mr. TAUBES: There is an infinite number of other factors. And the question becomes, can you ever go from this association between two events to the implication that it's a causal relationship?

BRAND: One of the big conclusions to come out of the Nurses Health Study was the conclusion that hormone replacement therapy reduced heart disease in older women. And that was found to be not only fatally flawed, but in fact just the opposite.

Mr. TAUBES: Yeah, it's a complicated story. Back in the '80s when the Nurses Health Study first reported on hormone replacement therapy, estrogen therapy for postmenopausal women, it appeared to dramatically reduce the risk of heart disease. And what that means is women who take estrogen have much less heart disease than women who don't. By the 1990s, it went from being a treatment of hot flashes and other menopausal symptoms to being a preventive treatment for a chronic disease. And then lo and behold, they do these very expensive long-term clinical trials, you know, where you randomized tens of thousands of women to one treatment or a placebo, and you find out at least in these trials that it does not prevent heart disease. It also causes a variety of cancers and strokes, or at least it increases the risk. And it seems that the initial 20 years of observational study is telling us that this was a way to prevent heart disease appears to have been dead wrong.

BRAND: So the difference being that the Nurses Health Study and other epidemiological studies rely on what's known as observational science, right?

Mr. TAUBES: Yeah, they just...

BRAND: Observational studies, not clinical.

Mr. TAUBES: They just observe these women and they say this is what we see and we see that they take these drugs or they eat this diet or they do this exercise and they have these diseases. And this is as opposed to a randomized controlled trial where you literally say let's start with say 10, 20,000 people and pick subjects at random and say, you take the drug, you don't, you take the drug, you don't, you take the drug, you don't, and then I'm going to follow you for five or ten years.

And in that way, by randomizing these subjects into two different groups, the idea is you're controlling for all these other factors that might differ between people who choose to take the drug and people who don't.

BRAND: However, the Nurses Study, since it involved so many nurses, tens of thousands of them, wouldn't it randomize it anyway?

Mr. TAUBES: No, that's the point. No matter how many people you have in the study, by picking people who choose to do a certain thing, you're not looking at them at random. You're going to say we're going to look at every woman who takes hormone therapy or we're going to look at every man who runs five miles a day. And as soon as you do that, you select out a certain type of person who has a lot of things in common and a lot of things different to those people who don't run five miles a day or don't take hormone therapy.

So it turns out, for instance, that women who take hormone replacement therapy are very different from women who don't, and they're different in a lot of different ways. They tend to be wealthier. They tend to be more health conscious. They tend to be better educated. They tend to have better, you know, doctors who are more interested in prevention.

And so then you start - you have all these factors that are different between these two groups of women and you want to say, well, the difference in disease that we see is due to one of them, the actual drug, and now you're just guessing. Until you do this trial, where you get rid of all those, you know, various factors by random, you know, and taking subjects at random and saying you take the drug and you don't, until you do that, you don't know. You're just guessing.

BRAND: Well, what about the givens that we now have, the givens that come out of these studies, givens that we should be eating a Mediterranean diet, we shouldn't be smoking, we should be getting plenty of exercise, that all those things will keep us healthy?

Mr. TAUBES: You shouldn't be smoking. That's probably the one thing that has come out of these studies that, you know, the data was just so unambiguous that you can't explain what we're seeing any other way.

The problem with all these other things, even exercise and the Mediterranean diet, is there is other ways to explain it. There's this phenomenon called the healthy user effect. Somebody who chooses to do something that they think is healthy tends to do a lot of things that they think are healthy. So people who choose to exercise also don't smoke, you know, they get plenty of sleep. They live life in moderation. There's a whole infinite number or cluster of things that they might do, and you can't really believe it's the exercise. I mean these people who did the studies believe it, but from the scientific point of view, they're guessing.

BRAND: So next week when we hear about the latest study that's telling us to pop this pill or, you know, ride this exercise bike or eat this olive oil to keep us healthy, when that comes out next week, what are we supposed to think?

Mr. TAUBES: Be skeptical. You know, if it's telling you that by doing it you'll be healthier, unless it's a randomized control trial, you don't know what they're missing.

One of the reasons I wrote the article is I'm trying to get my colleagues in the press to stop reporting these stories. And it would be great if I could the epidemiologists themselves to stop making that step from association to maybe it implies cause and effect. You know, you read their articles and it's fascinating because in the bulk of the article they'll talk about all the - you know, they'll use all the correct terminology, this is an association, it's a risk factor, it's blah, blah, blah, blah, blah. And then in the abstract, the conclusion, they'll say this implies that A may cause B or C may cause D. And so they go from what's logically correct to taking this illogical step. And it's that illogical step that then gets you into the newspapers and gets you funding.

BRAND: Gary Taubes, thank you for joining us.

Mr. TAUBES: It's been a pleasure.

BRAND: That's Gary Taubes. His article on the problems with epidemiology is the cover story in this coming Sunday's New York Times Magazine, and he has a new book coming out. It is called "Good Calories, Bad Calories: Challenging the Conventional Wisdom on Diet, Weight Control and Disease."

Copyright © 2007 NPR. All rights reserved. Visit our website terms of use and permissions pages at for further information.

NPR transcripts are created on a rush deadline by Verb8tm, Inc., an NPR contractor, and produced using a proprietary transcription process developed with NPR. This text may not be in its final form and may be updated or revised in the future. Accuracy and availability may vary. The authoritative record of NPR’s programming is the audio record.