Ten exact whole years have happened since I published The China Study: Fact or Fallacy?—a critique intended as a passion project, written for my dozen-person audience of fellow dogma-averse misfits, funded “between jobs” by insurance money from an accident, with a URL typo that haunts me to this day (verily, I weep).
I’ll be honest here. If I’d known ahead of time how much the whole thing would blow up—the stress of highly public back-and-forth debates, the sudden appointed role of dietary rabble-rouser, the bizarre permutations of my name I’d be greeted with for years to come (Diane Minger? Denise Miller? Dennis Munger? Who even am I)—I probably would’ve psyched myself out and never posted it. Or at least been much more careful about the “liberal use of adjectives and cutesy expressions” that flagged me as a meat industry shill (cattlemen sure do love their modified nouns!). I would’ve told Denise the Younger not to get seduced by the paleo community, replete with its own ideological circle-jerking, just because they welcomed me so readily. And I definitely would’ve deleted the “Professional Sock Puppeteer” title I’d listed as a joke-job on Facebook. The internet, as internets are wont to do, took it seriously, and I must forever bear the shame of having faked such prestigious credentials. I repent.
But life was kind enough to blind me from what was to come. So on July 7, 2010, I hit “publish” on the longest chunk of writing I’d pixeled out since college. And, due in no small part to the paleo and low-carb communities’ warp-speed sorcery for making anti-vegan things go viral, I suddenly had a career.
On one hand, I wouldn’t change any of this for the world. It could be argued that I owe the last decade of my life to this flippin’ study. Certainly, the critique and all that followed carved a path radically different than anything I could’ve dreamt up for myself. THANK YOU CAMPBELL!
But on the other hand—the mightier, stronger, dominant hand that’s really tired of punching dead horses—I would absolutely love to never have to talk about the China Study ever again, ever.
So on this honorable day, I thought it’d be nice to tuck away the few loose China Study ends that keep flapping in the wind, address some questions I still get asked pretty regularly, and buy myself another decade before having to discuss The Study That Shall Not Be Named (apart from naming it, like, 48 more times in this post) in any appreciable way.
ON WE GO!
Now that I’m older and wiser and new knowledges have happened, do I still agree with my critique?
In a word…
So, my original idea for this post was to do a line-by-line “critique of my critique” and tear holes, wherever possible, in the arguments I made way back when.
But then I re-read it all. And read it once more. And apart from being slightly nauseated by my early-20s snark and undertones of self-importance, I didn’t find much that was technically unsound. At least not enough to warrant a brand new blog post. The numbers hold up, the logic’s still logical, and my core gripes—particularly Campbell’s use of extra variables to form illusory links between animal foods and chronic disease—remain gripe-worthy.
Perhaps my chief regret is how important I made any of it seem.
You see, over these past TEN WHOLE ENTIRE YEARS, I’ve become somewhat of a China Study nihilist. I don’t think the original study can prove, disprove, imply, suggest, hint, damn, shout, or whisper any relevant nutritional truths. Nor do I think my critique can do any of that either. Ultimately, what I did was battle Campbell’s unadjusted correlations with more unadjusted correlations. All drawn from data that was observational to begin with. This is the weakest of weaksauce. The China Study is like a ghost of nothingness that I spent many hours poking with a stick that was also made of nothingness.
Even Richard Peto, one of the China Study’s very own lead researchers, is on record expressing his doubts about the data producing anything meaningful (correspondence forwarded from reader Jane Karlsson [thank you, Jane!]; PDF here):
“Having spent about 25 years trying and failing to make much sense of the geographic variation in cause-specific mortality rates, I’m pretty sceptical of anything definite coming out of it all.”
That said, I do think it was important for the nothingness battle to happen. After all, The China Study book went pretty gangbusters, and as far as its namesake study was concerned, lots of people thought the nothingness was somethingness—including 23-year-old me. Sometimes we need to entertain bad data for the sake of clearing it away.
Have any REAL math people with REAL math credentials backed up my findings?
Why, yes they have!
Luckily for us all, the good folks over at Red Pen Reviews combed through The China Study book with glorious rigor, and even hired a professional statistician—Karl Kaiyala, PhD—to analyze the same data I did. Their review is online here, and the relevant-to-this discussion bits are under “Scientific Accuracy.” If you don’t trust my number-crunching, you’re welcome to trust theirs:
In particular, the large observational study in China the book is named after does not support the central claims of the book. We confirmed this by consulting the original data at the University of Washington medical library and analyzing it with the help of a professional statistician, Karl Kaiyala, PhD. In addition, The China Study omits important evidence that undermines its claim that animal protein but not plant protein increases cancer risk in rodents.
However, as Campbell has pointed out, these figures are “unadjusted”, meaning they are simple analyses that don’t control for potential confounding factors. To address these concerns, we digitized data from the China Study on the total cancer mortality rate in people younger than 65, plant protein intake, animal protein intake (total protein minus plant protein), smoking rate, latitude, agricultural and industrial output (a marker of wealth), literacy (a marker of education), and age. We gave the data to a professional statistician, Karl Kaiyala, PhD. He analyzed the data in multiple ways (multivariate regression), none of which supported the book’s claim that people who ate more animal protein died of cancer more often.
On this point, Kaiyala’s findings reach the same conclusion as those of Denise Minger, who extensively analyzed China Study data beginning in 2010. Academic researchers have come to similar conclusions regarding the China Study data.
Lastly and most notably, they confirmed the plant protein/heart disease association I mentioned in my critique (as best I can tell, Kaiyala didn’t look at wheat—but if he had, it would’ve shown up as the reason plant protein looked bad):
Kaiyala did find in his multivariate models that higher cardiovascular mortality was independently associated with higher Apolipoprotein B, latitude, and plant protein intake. The first two of these were expected, while the third was not.
What about the wheat and heart disease thing?
For those who missed it, one of my critique’s surprise findings—which I confess I was initially whatever about, having not yet been exposed to the anti-gluten brigade overtaking cyberspace—was a strong correlation between wheat intake and heart disease. It caught so many folks’ eyes, in fact, that I ended up writing a follow-up post trying to figure out what was up: The China Study, Wheat, and Heart Disease; Oh My!
In that post, I ran a bunch of multiple linear regressions testing every potential confounder I could think of. Smoking? Industry work? Vegetable oil use? Latitude? Vitamin C intake? Sodium intake? Saturated fat? Alcohol, doggone it? In they went into the statistical fray!
Amazingly, nothing I adjusted for got rid of the correlation. NOTHING. The wheat/vascular disease link held up like a loofah by the foreman of the night.
Pretty damning, right?
Well… maybe. You see, there was one particular variable—folate—that I just couldn’t get out of my head. And in my head it remains, all these years later, like a crayon that got shoved a little too far up one’s nostril.
Here’s why I can’t get over this. In the China Study data, vascular diseases clustered heavily with neural tube defects—a well-known consequence of folate deficiency. Folate is also super involved in vascular health, and too-low levels increase the risk of heart disease and stroke. Meanwhile, back in 1980s-rural China, the wheat-eating regions tended to be notoriously slackin’ in the folate department.
And guess what? There are entire papers written on how cardiovascular disease and neural tube defects coincide, solemnly bonded by folate deficiency (and by consequence, elevated homocysteine). ENTIRE PAPERS. This is not a Dennis Munger pipe dream; it’s a real live true science thing.
So, even though I stuck folate in the statistical models I was running and wheat still “won” as the heart-killer, I’ve long suspected that I just couldn’t model the data adequately using linear regressions. Folate’s proclivity for U-shaped-curvage in relation to disease makes it a slippery sucker. My hunch is that a better model, accounting for folate’s curviness, might pardon wheat.
But what do I know? I’m not even qualified to put socks on my hands. If anyone out there has the skillz and the interest, I invite you to play with the data yourself and tell me what you find! (But not until the year 2030. Please recall, I’m off China Study duty ’til then.)
Likewise, stats wizard Ned Kock ran his own, much fancier analysis of the wheat and vascular disease data and published a peer-reviewed paper of his findings (PDF here: Wheat Flour Versus Rice Consumption and Vascular Diseases: Evidence from the China Study II Data). He used software that allowed for nonlinear analysis—something that was, and continues to be, above my pay grade—and in so doing, confirmed the association between wheat consumption and vascular disease mortality. (As far as diet went, he focused on the variables of wheat flour, rice, and total calorie intake.)
However, he ultimately landed on a different hypotheses: that wheat culture, defined as the constellation of social and lifestyle factors accompanying wheat consumption, was really the deadly element; not wheat itself:
Personally, I join him on the side of “it’s probably not actually the wheat”—in this data and elsewhere. Though I think plenty of people do better without wheat in their diets, the mechanisms linking it specifically to heart disease have seemed less and less plausible the more I’ve learned about human health. More compelling is its use as an ingredient in high-reward foods and subsequent contribution to chronic energy surplus—something that does have solid mechanistic backing.
With only a handful of Pacific Standard Time minutes left of my critiqueaversary before the clock strikes midnight [EDIT: clock struck; carriage is now pumpkin], this seems like a fine time to begin the transition to What Will Come Next.
I won’t be blogging much about nutrition from here on out. In fact, this might be my last food-related post for awhile [EDIT #2: fine, fine; In Defense of Low Fat Part 2 will still happen]. I’ll explain why soon. But in the meantime, I need to thank every single one of you for your presence on this journey. I don’t care if you’ve never commented, liked, shared, Tweeted, emailed, or otherwise made your existence known; your eyes on this page mean something. I truly believe that. And I hope you will stay for the next chapter. Thank you, thank you, and adieu.