Top critical review
7 people found this helpful
Good biochemist morphed into bad epidemiologist
on July 21, 2014
Highly credentialed diet gurus, with oversize ego and extreme greed, battling each other with irreconcilable sect-like edicts and operating freely in a country where big agro-alimentary and pharmaceutical businesses have thoroughly corrupted not only governments and professional associations but also the actual process of knowledge acquisition through clinical trials: this is the painful situation each citizen now has to face in the western "developed" world.
Unfortunately The China Study, does not provide a solution. Reading it feels like being forced to attend a preacher meeting by a pushy friend trying to ensnare you in his sect.
The book start with a worthwhile description by the author of his biochemistry/nutrition research which establish that casein enables aflatoxin carcinogenicity in rats. This is good research that very likely was instrumental in propelling the author to a certain fame.
But there is a monumental step from showing that casein is bad for rat to wildly extrapolating that *all* animal proteins are bad to *men* and therefore a plant-based diet is required, a step which the too enthusiastic author unfortunately takes. Such ultimately irrational crusade by a well-meaning scientist is not without similarities with the late Linus Pauling ill-fated vitamin C crusade.
The author appears to lack the minimal undergraduate-level statistic/epidemiology knowledge to navigate the treacherous field of nutritional correlation. He explains to the reader what (he thinks) a statistically significant test means: he explains that if the test pass for example the commonly accept treshold of 95%, that means that the effect we see has 95% of being "real".
This is a classical immensely common rookie mistake.
The explanation requires some context that can be found in any good undergraduate level statistical/epidemiological but the main idea is this: a 95% significant test means that there would be only a 5% probability that we would observe the difference of results between the 2 groups if the treatment had no effect. It does NOT mean that there is 95% probability that the effect is real. This is a very subtle difference and typically requires to read a more detailed explanation to understand it.
But you can get a taste of it with this example: suppose you have a state lottery small price that you have a 1/20 (5%) change to win. Now you make the experiment with a friend: you are both going to buy a ticket, you will wear a red shirt and your friend any other shirt color, the goal being to see if wearing a red shirt is a factor in winning a lottery. Now suppose you did win. There would be only a 1/20 (5%) chance of this occurring by chance and therefore you could be tempt to say "Wearing a red shirt is a factor in winning a lottery (95% confidence)". However because you have *prior* knowledge that lottery are unrelated to red shirt, the correct statement would be something like "Winning this lottery was simply a great strike of luck and unrelated to my red shirt (100% confidence)" which is basically the opposite of the other statement! Applied to epidemiological research this has considerable importance, often not sufficiently appreciated.
The author almost never mention plausible alternatives to observed effect, again another common rookie mistake. For example, when a study shows a link between consumption of milk and prostate cancer, the author concludes this has to be related (not even just to the specific casein protein) but to "animal protein facilitating/causing cancer". Absent from the discussion, is some sensible alternative such as that perhaps milk is now produced by cows subject to hormonal treatment and these could affect the milk and ultimately have an effect on a possibly hormone-dependent prostate cancer.
In reality, the entire book if chock-full of unwarranted generalization and unfounded admonishment. For this reason, it was truly painful to read.
Finally the condescending tone of the book is very irritating: the author never miss a chance to mention how great he is and behaves like a common unpleasant diet-guru/preacher.
I suggest as an alternative an immensely better book: "Food and Western Disease" by Staffan Lindeberg. It is extremely well researched book. It lists the result of an incredible number of studies in a well organized way. Furthermore it is written in a very professional and scientific style (Note: I have never met and have no personal, professional nor financial relations whatsoever with Dr Lindeberg)