MORE DOUBTS REVEALED ABOUT WOMEN'S HEALTH INITIATIVE STUDY DESIGN

REcent reports in the news about conclusions drawn from study's conducted as part of the Womens Health Initiative have caused considerable confusion. Most recently it was reported that the WHI study on Reduced Fats Consumption showed that reduced fats intake in the diets of the women studied did NOT show reduced risk of certain types of Cancer.

Well, more and more doubts are coming to light about these so-called studies. An article in the Current issue of Newsweek reveals some of these:


(emphasis my own_JW)



The study followed the womens eating habits (as reported) for seven years. THAT IS NOT ENOUGH TIME TO DRAW ANY CONCLUSIONS RE CANCER RISK. THAT REQUIRES AT LEAST 10 YEARS OR EVEN BETTER 15 YRS TO REALLY BE ABLE TO DRAW MEANINGFUL CONCLUSIONS.

Yet the study's authors released conclusions on cancer based upon an inadequate time period.

As it turnms out the women in the control group ate almost as much fat as the test group! You have to have enough of a difference in fats intake to be able to draw conclusions.

http://www.msnbc.msn.com/id/11225530/site/newsweek/





In the world of business making claims for a product which you know are not true (this little baby was only driven on Sundays by a little old lady. Would I lie to you?) that's considered engaging in fraudulent business practices.

When you announce conclusions from a 'study' which wasn't really a study (no significant difference between study group and control group in consumption of substance being studied, insufficient time allowed for symptoms of disease to be detected) are you not engaging in fraud. You are saying you have a legitimate basis for drawing certain conclusions when you really do not.

Sounds pretty fraudulent to me.

my mother-in-law was one of the participants in the dietary study, and nearly drove us ALL crazy with the low-fat crap. She wouldn't eat hardly anything with any micro-measure of any sort of fat at all, which I don't believe is healthy anyway, and her poor husband had to stay on the diet with her! When we had dinner at her house, she wouldn't serve anything off her diet, which meant we all had to eat her stuff. Personally, I am SO glad that this study didn't prove what it set out to, because I have been saying all along, that you shouldn't deprive yourself of anything, just eat it in moderation. I must admit that when I read the results, my heart leapt and I was particularly happy, because I could kinda say 'I told you so' to the MIL!!

She (and your father) must have lost a good deal more than the rather pathetic 3-4 pounds (over 7 years!) that was the average.

By the way, do you know if she was in the 'target' group or the 'control'?(they admitted women in the 'control' reduced fat intake too.)

It's kinda interesting how she demanded your father diet along with her. But, I guess that's the subject for another study. :-)


I'm glad your happy, but somewhat confused, since the "study" didn't prove anything, one way ... or the other. You can't draw any conclusions pro or con from this "study".

Don't be confused: yes, the study didn't prove anything one way or the other. That's the point. Her insistence that the extreme low-fat lifestyle was THE way to go, was kinda nullified by the results of the study. She was on the darned thing for over 7 years. For heavens' sake, if being on that extreme doesn't do SOMETHING in 7 years, what's the point? I still stand by a modified approach: nothing in extreme, but enjoy your diet.

She drove us all crazy with how strict she was about everything, and then to find out that it was all moot actually made me smile....

ps. My MIL and FIL (my husband's parents) maintained their weights. My MIL lost some, but it was because she found out that she has celiac disease, and was hospitalized for awhile with extreme diarrhea. Also note: her hair is very brittle and dry looking, and she has large, dark brown very dry, scaly patches on her face and arms. IMHO, a little more fat in her diet would do her good.

some by known variables and probably by still unknown variables.

Understanding and public awareness change. And in a longitudinal study they can effect the ability of the study to stratify the results enough to keep up with changes in awareness that occur during the study. Just consider the issue of fats that you raised. Across the study period public awareness of total fat vs. transfats, triglycerides vs sterols changed. Good vs Bad cholesterol came into common parlance vs. total cholesterol.

Let's let the experts toss this around for a while. Let them grind the results against what is accepted understanding and see where we end up. I think by summer the solid results will be clear. The areas that need cleaning up will be identified and studies will be designed to address them.

of diet (such as types of fats being so important : OMega 6 vs Omega 3's and trans-fats). I only spoke about the design and control aspects (or lack thereof).

If your 'control' group was also reducing their fats intake (an unknown amount) and you claim to be comparing a group who supposedly reduced their fats intake a known amount (the weight loss statistic seems to cast doubt on how much reduction of fats really occurred) to a 'control' group who you assumed were not going to reduce their fats intake - but they, in fact, did - Do you have a study? Can you make a meaningful conclusions?

I think it should be obvious you cannot.

And drawing conclusions about CAncer incidence after 8 yrs ??? Come on!! (and this is still conclusions based upon observations of two groups whose differences in fats consumption is not really known (they do not even know how much the 'control' group reduced their fat intake by). This isn't science, it's just a waste of time. - actually it's worse than that because some people who didn't read the fine print will think these so-called conclusions (which were pretty qualified, by the way, if you read them) are valid.

going to be obvious, or easy.

Everyone needs to think about how studies are done. There certainly are assumptions and weaknesses.

Yes, cancer takes a long time to develop.

BUT

The number of participants GREATLY increased the statistical sensitivity of the study. Consequently, a very small difference in 8 years might be (and BTW _IS_) argued as significant enough to be meaningful even without the passing of 20 or 30 years.

Clearly cancer and heart disease are more common in OLDER participants and as participants get older MORE cancers and more heart disease will show up. But the reason to involve thousands and thousands of women is to get TREMENDOUS statistical sensitivity. Some of these statistically significant signals have biological significance and some do not.

Similarly the dose response which is typically something like e^(exposure rate * time) where exposure and time are linearly related. Consequently higher doses (assumedly) can reveal the same result as longer time. Combined with very high sensitivity, looking for increased rates of disease among higher "risk/dose" exposures can _sometimes_ provide insight beyond the time frame one might associate with typical onsets of illness.

Making decisions on the biological/medical relevance is ONLY GUIDED by the decision aids. It's going to take time to sort out what is biologically/medically important and what is simply statistically interesting, what is confounded within that and the ever present phenomenon that what is biologically meaningful may yet not be quite statistically meaningful.

Understanding the results of the WHI is well beyond even half decent popular medical reporting like NPR's Science Friday. In half a year scientists and epidemiologists (after they have read, reread, considered other scientists opinions and raised their own questions) will have a much better grasp of how to address the confusion. It takes time. This study NEVER promised to produce results with the speed and efficiency of fast food outlets.

weightings of factors does not imply any contretemps. I appreciate your insights (I'm always ready to learn from others!). Your statements re "higher doses (assumedly) can reveal teh same result as longer time" - I recognize this as a principle FDA, among others, rely upon, but of course you realize for some deseases this is issue which is subject to debate (and I suppose endlessly so).

I appreciated what you had to say about dose response relationship and larger sample sizes producing greater statistical sensitivity. However, I think all this sensitivety is useless when you don't even know the (relative) dose your dealing with (They didn't know what the difference in fat intake was between the target group and the 'control' group. (keep in mind they're not even sure what the test group actual dose was (3 pounds lost over 8 years - makes claims of dietary fat restriction, of 20% of total caloric intake, kind of suspect- referring to Brownell and Ornish here- and thay have no idea what the 'conrol' group was consuming.). And some of the researchers they talked to involved in the study said there is room for concern about whether the amount of fat reduction (assuming the purported or sought amount of fat reduction was actually known to be the actual reduction and you knew what the reduction for the 'control' group was) was even enough to produce strong enough results to be significant (in an 8 yr time period). I didn't show that quote but it is in the article the first link goes to(I think).

You know if you have sloppy data no amount of statistical 'filtering' can clean it up.