Thursday, March 6, 2014

Reusing data

The National Health System (I'm pretty sure that's what NHS stands for) of Britain has a rebuttal to the article about a high protein diet being bad for you. I posted about it yesterday. Some of the points the NHS makes:

* That thing about high protein being as bad as smoking was not in the original article. Some reporter made that comparison and other news sourced copied it.

* As far as I can determine the study did something like this: Back in 1988 over 6000 people were asked to record what they ate for 24 hours. They were also asked whether this food consumption was typical for them. More questions helped place them in various categories, including age. In 2006, 18 years later, researchers assessed the health of the subjects. If the subject had died researchers recorded the cause of death.

* The level of physical activity was not recorded.

* The data was not recorded for study being discussed. This is a case where later researchers use data that was originally designed and gathered for some other purpose.

I had reported the dieters ate 51% of their calories in carbs. That is the average, not the same for every subject. There was analysis to correlate health outcomes with protein consumption. It sound like the researchers did not correlate health outcomes with carb consumption. That would have been a much different paper.

If I had been on the review team for that paper I would not have approved it. I see all sorts of holes in the conclusions and consider the study worthless. Bad science. I'm sensitive to bad science for two reasons: (1) bad science is plaguing the American diet and I'm living the consequences and (2) bad science continues to be used to deny rights for sexual minorities (coming up in a later post).

No comments:

Post a Comment