Many diets work about the same, U.S. study finds

[Demi]

Reuters News Service
Published: 2 July, 2007



Many diets work about the same, U.S. study finds

By Maggie Fox, Health and Science Editor

WASHINGTON (Reuters) - Looking for that perfect diet? Researchers have bad news -- all diets have just about the same result, and none of them are great, U.S. researchers reported on Monday.

A typical diet helps people lose an average of 6 percent of their weight, typically 10 to 15 pounds (5 to 7 kg), and most people put it all back on after five years.

Weight loss drugs are similarly ineffective in the long run, said Dr. Michael Dansinger of the Tufts-New England Medical Center in Boston.

"It's disappointing but I am optimistic that we can do better in the future. We are learning some of the factors that improve the effectiveness (of diets)," said Dansinger, whose study is published in the Annals of Internal Medicine.

The news is bad for those who hoped a gentler approach to dieting might be more effective over the long-term. Programs that made people eat fewer calories worked better, as did those that involved more frequent visits to either diet groups or to a counselor's office.

But there is good news -- even a small, temporary weight loss can benefit health, Dansinger said.

"A modest weight loss of six percent that is partially maintained for five years is likely to have important health benefits such as delaying the onset of diabetes," he said in a telephone interview.

Dansinger and colleagues looked at the results of 46 trials that included nearly 12,000 people.

About half were on diets. Dansinger said it was difficult to find good studies that included a control group not on a diet. It was also hard to find studies that followed people for more than three years.

The only commercial program included in the study was Weight Watchers. Most were government or university-sponsored programs.

No studies that included food or shakes were included because they did not include a non-dieting group for comparison.

"The results we found, 6 percent weight loss after one year, is in the same ballpark as most of the studies of weight loss, including studies of weight loss medications," Dansinger said.

"We also found the weight loss gradually goes away so that about half the weight loss was gone within three years and almost all the weight loss was gone within five years. That that is also similar to what has been found with weight loss medications."

Dansinger said some of the studies included exercise, but his analysis was not designed to tell whether exercise helped weight loss last longer.

Nearly two-thirds of U.S. adults are overweight or obese, with a higher risk of diabetes, heart disease, high blood pressure, arthritis and cancer.





http://www.reuters.com/article/healthNews/idUSN0236678720070703

[Abd]

What is consistently ignored in these studies is that it is a population which is studied, and, presumably, the population has been made relatively inclusive. The diet works for some and not for others. Indeed, it might work well for some and not at all for others, hence the low overall results.

However, what causes it to work for some and not for others? Are there differences between these two subpopulations? Wouldn't it be interesting to find out what factors make diets work and what factors make them fail?

Instead of real work like this, what we get is study after study which concludes that "diets don't work," though what they generally find is not that they don't work, but that when the results are averaged over a large population, they only work a little.

Suppose that the overall population consists of ten classes of people, each one of which would have an ideal diet, one that would cause them to lose weight and keep it off, but which is not effective for all the other subpopulations.

(This is not an actual proposal, just an idea that would need to be ruled out if you really want to claim that "diets don't work.")

So, take any one of the diets and test it on the entire population. For nine-tenths of the population, the diet has no effect except to irritate and torture the dieters. They have poor compliance and they don't lose weight. The other ten percent do very well. Suppose this is a quite obese group and they actually lose sixty percent of their body weight. Average this together with the others, and you get a six percent average loss. "Terrible," the authors pronounce! Yet, for ten percent of the population, the diet was fabulously successful.

Bad science. Taubes, when is your book coming out?

(Now, in any population there is going to be variance. A diet could be totally irrelevant and a certain percentage of people, quite possibly, could lose weight on it. So if you took the diet study imagined above and then looked at the group that actually lost weight, do they share a common characteristic that could be identified. If you can identify such a characteristic, this does not yet prove much of anything, unless the characteristic is so clear and so distinct that it *must* be related to the result. What you'd normally do, then, is to create a new study group that share this characteristic and see what the diet does with *them.* If the results are very good, then one has indeed identified a subpopulation that will generally be helped by the diet. Otherwise, it was a fluke in the original study.)