Coffee, Tea Might Stave Off Diabetes - US News and World Report

[dracodraconis]

Coffee, Tea Might Stave Off Diabetes - US News and World Report

A recent Australian study noted that increased consumption of coffee and tea, with or without caffeine, correlated with a decreased risk of developing type 2 diabetes. Earlier studies had linked coffee consumption to decreased risk of prostate cancer, liver disease, stroke, Alzheimer's, and Parkinson's disease. Unlike previous studies, however, this one discovered that the health benefits of coffee and tea include decaffeinated versions.

Comments

[nightprincess]

http://archinte.ama-assn.org/cgi/content/abstract/169/22/2053

Every time I see these sensationalistic food studies, I wish I could get more information. Do iced coffee and iced tea work too? How about hot water? Not that it'd be easy to find subjects who are willing to drink four cups of plain hot water per day for the length of the study, but maybe risk of diabetes is reduced because the liquid fills the stomach, and the temperature of the liquid prevent most people from drinking it very quickly. There are so many variables. How are they coming up with these conclusions?

[dracodraconis]

That's why it's important to stress a correlational (not causal) relationship. The results were based on historical data so there may be many other factors that are also involved but not available in the data set they used. What is needed are studies designed to minimize the presence of interference variables. And these studies should consider other commonalities such as those you mention.

Another thing to note is that there are other health problems associated with excessive caffeine intake. Four cups a day, if I recall correctly, is about the upper limit of healthy consumption, with 1-2 cups being a recommended safe level.

As to how they draw the conclusions they do, some is speculation, but often based on some knowledge of field. Humans are pattern-matching animals and we like to explain things. Saying "I have no idea why" doesn't cut it in a research paper. In the long run, speculation is used to drive further studies that confirm or deny the speculation, or suggest other possible interpretations. Also, by presenting the speculation, their peers can agree or disagree. As much as people talk about scientists "knowing", oftentimes we work at the edge of what is known and can only make educated guesses to suggest what should be studied next to expand the boundaries of knowledge.

[nightprincess]

It's good to know that some of it is acknowledged as just speculation. The way things are reported to the public, it does seem to me that science is very drastically broken. I wonder if the Renaissance artisan scientists would've come up with everything they did if they were tied down with a "publish or perish" mentality. In the area of nutrition especially, there seem to be a lot of propagated half-truths that seem more destructive than simply acknowledging "I have no idea why, but this correlation is potentially interesting."

[dracodraconis]

The other problem is news services who post these articles are also attempting to get your attention so they take some liberty in how the results are presented. The headline in this case made a claim, but the original study only indicated a correlation worthy of further study with some speculation about possible causal relationships.

[dracodraconis]

Oh, I just noticed this:

"Owing to the presence of small-study bias, our results may represent an overestimate of the true magnitude of the association."

They note in their conclusions that the data is based on a small number of data points so should be interpreted as a potential link. In other words, the results are interesting enough to warrant further study, but the size of the current study may make the association appear stronger than it really is.

[nightprincess]

I'm still not sure I understand how they have a "small-study bias". The abstract seems to suggest that they had data from "457,922 participants"? How much data do they need for something to not be a small study?

[dracodraconis]

Generally in studies of this type it means that, of the 400K+ participants, when you extract only those who consumed decaffeinated tea and coffee you only had a small batch of people to examine for diabetes prevalence.