Low carb, Hypothyroidism, and How to Lie with Statistics

I'm just an artist. An artist that read Darrell Huff's "How to Lie with Statistics". It's an old book, published sixty years ago now. Apparently people have been using statistics to lie about things for a while now.

I had heard something about low-carb diets causing hypothyroidism (repeat after me, correlation is not causation). As it stands, it seems to me that there is no way to know definitively whether they do or not, because the subject has never been studied. Oh, but you say, what about all those studies that are cited by so many bloggers. One of the first posts I found when I delved into this notion, was by blogger Anthony Colpo, and I read his post on the matter. Not withstanding the rudeness (which is not just on his side of the argument but rather all about the blogosphere as it were [did y'alls mothers not teach basic manners?]), I did think he might-could-be on to something. Until of course I looked at the studies he cited.

I never understood exactly why an underpowered study was, well, underpowered until I read Huff's book. You see, if you don't have enough of a sample size, what you have is chance. Yes that's right, pure unadulterated chance. The same sort of chance that you have flipping a coin. The explanation that Huff gives is crystal clear, and I urge you to buy his book and read it. If you flip a penny ten times (let's say to represent the outcome of ten patients in a study), you might get that it comes up heads eight out of the ten times. But you might get that it comes out heads two of the ten times. You would have to flip thousands of times (or maybe study 100k participants because humans have more than two variables) to come out with any sort of reasonable data to study.

"The importance of using a small group is this: With a large group any difference produced by chance is likely to be a small one and unworthy of big type...How results that are not indicative of anything can be produced by pure chance--given a small enough number of cases--is something you can test yourself at small cost. Just start tossing a penny... Only when there is a substantial number of trials [or participants] involved is the law of averages a useful description or prediction."--Darrell Huff, How to Lie with Statistics p.39-40

The two studies Colpo cites in his article, Dietary-induced Alterations in Thyroid Hormone Metabolism during Overnutrition and Isocaloric carbohydrate deprivation induces protein catabolism despite a low T3-syndrome in healthy men are woefully underpowered. The first one used three, yes THREE participants to conclude that no carbohydrates induces low T3 over a week. Yes, a week. The second study had a total of six participants over the course of eleven days (the abstract doesn't say if that's 11 days for each diet, or 11 days total, but whether it is one or the other it adds up to bull-hockey). I suppose that's twice as good as the first study. Bull-hockey times two equals twice as much bull-hockey for your grant money.

There could be something to this and there might not be. If there is a higher incidence of thyroid problems among people who eat low-carb (and I don't even know that that is true, but IF it is) how do you know that the problems weren't triggered by the diet they ate before they ate low-carb, the crappy Standard American Diet? Or that since most people who do low-carb want to lose weight, maybe the obesity triggered their thyroid to go out of whack. Or maybe it was a virus they were exposed to. It could be absolutely any of the above, or none at all. A week long study is not enough for your endocrine system to adjust to the changes of going from a Standard diet to a low-carb diet. Six males or three random participants tells you nothing other than chance. It seems like most of the medical studies I read are run by Vegas gambling addicts. Let's throw the dice and see what we come up with! This is not science. This is nonsense.