April 04, 2012
Flavonoids Cut Parkinson's Disease Risk In Men Only?

Berries, grapes, tea: get flavonoids to cut your risk of Parkinson's disease. If these compounds really do protect against Parkinson's they probably slow brain aging in general. But the weird result: The benefit was only seen in men. Why?

Men who eat flavonoid-rich foods such as berries, tea, apples and red wine significantly reduce their risk of developing Parkinson's disease, according to new research by Harvard University and the University of East Anglia (UEA).

Published today in the journal Neurology , the findings add to the growing body of evidence that regular consumption of some flavonoids can have a marked effect on human health. Recent studies have shown that these compounds can offer protection against a wide range of diseases including heart disease, hypertension, some cancers and dementia.

This latest study is the first study in humans to show that flavonoids can protect neurons against diseases of the brain such as Parkinson's.

Around 130,000 men and women took part in the research. More than 800 had developed Parkinson's disease within 20 years of follow-up. After a detailed analysis of their diets and adjusting for age and lifestyle, male participants who ate the most flavonoids were shown to be 40 per cent less likely to develop the disease than those who ate the least. No similar link was found for total flavonoid intake in women.

Again, why? Does estrogen already provide the protective effect that flavonoids provide for men?

Share |      Randall Parker, 2012 April 04 10:30 PM  Aging Diet Brain Studies


Comments
dlr said at April 5, 2012 8:39 PM:

Ah, the joys of epidemiology. If you only look at one disease, and one group you have 1 chance in 20 of getting a 'significant' result -- just by chance. Divide the test subjects into two groups, by nationality, or race, or sex, and voila, you double your chances of getting a 'significant' result, just like that. One chance in 10 that the results they discovered were just by chance. By definition. And that's if they just ran the analysis for one disease (what are the odds?). If they checked flavinoids in the diet against 10 diseases (and it's pretty easy to think of ten diseases for them to have checked against, say, Alzheimer's disease, Parkinson's disease, heart disease, breast cancer, prostate cancer...) and just 2 groups, that gives them 20 cases, heck, they now statistically -- just by chance -- have very high odds that one of those 20 events will look like a 'significant result' -- just by chance. After all, by definition it's 'significant' if there is only one chance in 20 that the results occurred just by chance. So, run twenty analyses on the same data for 20 different cases, and the odds are pretty poor that you won't hit pay dirt. Publish the one that is 'statistically significant', and throw the rest in your desk drawer.

See, it's almost like science, only not so hard.

All an epidemiological, ie, observational study is good for, alas, is generating ideas worth looking into, and TESTING, say, by a double blind randomized study. Until they do that you have to approach their results with extreme skepticism. The first question has to be: how many ways did they slice the data to come up with the one 'statistically significant' result that they reported? How many cases got analyzed and thrown away because they weren't 'statistically significant'?


dlr said at April 5, 2012 9:06 PM:

Here's a good link that discusses some of the (many) shortcomings of epidemiological studies. http://www.spiked-online.com/Articles/000000005513.htm

or

http://www.spiked-online.com/articles/0000000CA5AB.htm


or, in more depth, a very good book "The Epidemiologists: Have They Got Scares For You" by John Brignell. Extremely worthwhile read. He cranks the numbers on enough cases that you really see that an increase of just 40% in the odds, as in this study, is probably completely meaningless.

Nick G said at April 6, 2012 8:40 AM:

In college I did a statistical analysis of a bunch of sub-samples of my data, and got an unexpected positive result for one of them. I was pleased, but puzzled, so I looked again: turned out I had 20 groups, and I was using a 95% test for significance, so a positive result in one was exactly to be expected...

Ekoz said at April 6, 2012 10:48 AM:

As stated above, this is observational stuff. Hence, there are no "effects", just more or less arbitrary residual correlations. If they want real results, do a RCT.

the wolf said at April 6, 2012 10:50 AM:

It's the Flavonoids' "war on women."

Ezra Gonzalez said at April 6, 2012 12:09 PM:

If I had a flavonoid it would look just like Trayvon.

bbartlog said at April 6, 2012 6:55 PM:

Their p value may be pretty good, though (i.e. way better than 95%). Eyeballing the abstract and doing back of the envelope math: say that 500 of these Parkinson's cases were men (I'm assuming men are more likely to develop the disease). And let's say further that they split the men into quintiles when looking at dietary intake of flavonoids or anything else. Then to get their result, they'd have to have something like (say) 125/100/100/100/75 cases of Parkinson's, least to most, in the quintiles. That's +2.5 SD (in the low flavonoid quintile) and -2.5 SD (in the high flavonoid quintile). Very unlikely to occur by chance, roughly p~0.00004. You can make other plausible assumptions that would result in an even lower value.
That said... if they have *enough* different ways of slicing and dicing the data, enough different nutrients and diseases to look at or ways to partition the populations, even that good a value would still look dodgy. As everyone else is pointing out, it's not a great way of doing science.

PacRim Jim said at April 6, 2012 8:22 PM:

I'm breakthroughed out.
Decade after decade of wonderful discoveries.
Yet the killer diseases continue apace.
At this rate, the major diseases won't be controlled much less cured for another century.

Andrew Bailey said at April 7, 2012 2:58 AM:

Does anyone know any sites that I can find health programs??? in which i can found all health links such as mens health, womens health, general health, skin care, prostate health and much more in competition such as http://www.naturalherbaproductz.com got the best site for health in 2006 ?
1. Are health costs lower for people who maintain high levels of functional health?

2. Are guidelines that can help you avoid crime or threatening situations to let someone know where you are going and when you expect to return?

3. Is hypertension considered to be the #1 risk increasing the likelihood of strokes?

4. Is jogging 30 minutes an example of cardiovascular or aerobic activity?
5. What parts of the body does it benefit?
6. Why should we take time out to relax and rest?

Please number the answer your giving. This is really important.

Thanks
Visit: http://www.reviewphen375.co.cc

Post a comment
Comments:
Name (not anon or anonymous):
Email Address:
URL:
Remember info?

                       
Go Read More Posts On FuturePundit
Site Traffic Info
The contents of this site are copyright