Tuesday, February 07, 2006

No Butter No Better

Catchy enough? The post refers to the new study which seems to suggest that a low fat diet does not protect women against breast or colon cancer or heart disease:

The largest study ever to ask whether a low-fat diet keeps women from getting cancer or heart disease has found that the diet had no effect.

The $415 million federal study involved nearly 49,000 women aged 50 to 79 who were followed for eight years. In the end, those assigned to a low-fat diet had the same rates of breast cancer, colon cancer heart attack and stroke as those who ate whatever they pleased, researchers are reporting today.

"These are three totally negative studies," said Dr. David Freedman, a statistician at the University of California at Berkeley, who is not connected with the study but has written books on clinical trial design and analysis. And, he said, the results should be taken seriously for what they are — a rigorous attempt that failed to confirm a popular hypothesis that a low-fat diet can prevent three major diseases in women.

And the studies were so large and so expensive that they are "the Rolls Royce of studies," said Dr. Michael Thun, who directs epidemiological research for the American Cancer Society. As such, he said, they are likely to be the final word.

Perhaps these studies are the final word on the specific question of the value of low fat in the diet, but as one those interviewed points out, now the fad is all about the "Mediterranean Diet": eating little or none of saturated fats but using other fats fairly freely. We could start a study on this and eight years later might find out that this was a wrong guess, too. Or a right guess, who knows.

Who knows, indeed. I'm not happy with the preaching and sure-as-certain way we are given dietary advice based on fairly flimsy studies, not happy, because changing the whole diet means changing much that has social significance for us, that denotes family roots for us, that gives us comfort and even meaning. Yet I seldom read anything from the medical popularizes that would acknowledge this and how hard it really is to change the foods one routinely eats. Instead, there is an unspoken assumption that people who can't change their diets are spineless and greedy and almost deserve to get colon cancer or something similar.

Having said that, I must also point out that this particular study should not be interpreted as meaning that any amount of butter gorging is perfectly fine healthwise.

As my regular readers might know, I like to criticize studies as a public service. I have only two critical comments about the article describing this study, and only the first one is about the study itself. It is the problem of relying on self-reporting of study participants in determining how much fat their diets contain. The study found out that there was no average weight difference between the group who was assigned the low fat diet and the group that was allowed to eat whatever they wished, and this suggests to me that perhaps the low fat group didn't actually follow a low fat diet that religiously. If this suspicion is true the results would be comparing two similar levels of fat in the diets of these women and would mean nothing.

The second critical comment has to do with this statement of one of the people discussing the study results:

Dr. Rossouw, however, said he was still intrigued by the breast cancer data, even though it was not statistically significant. The women on low-fat diets had a 9 percent lower rate of breast cancer — the incidence was 42 per 1,000 per year in women in the low-fat diet group, as compared with 45 per 1,000 per year in women consuming their regular diet. That might mean that fat in the diet might have a small effect, Dr. Rossouw said, perhaps in some subgroups of women or over a longer period of time. He added that the study investigators would continue to follow the women to see if the effect became more pronounced.

I see this done a lot. A lot. There is nothing wrong with Dr. Roussouw suggesting that further research should be done. That is fine. But statistically non-significant results are just that: We have not shown that the two groups differ on average on whatever measure we are analyzing here. Given this, people should not pay so much attention on statistically non-significant findings.