I have great respect and appreciation for the New York Times, and a fair portion of both for the customary insights of foodie/journalist, Mark Bittman, as well. But I have just about none of either for the combination of the two represented in an editorial by Mr. Bittman in the Times last Thursday (Feb. 27, 2013) declaring sugar toxic.
In that column, Mr. Bittman chose to interpret for us all a scientific study of the association between sugar intake and diabetes around the world. As described by the authors, a team including Dr. Robert Lustig, who has made the vilification of sugar his signature cause, the study used "econometric models of repeated cross-sectional data." You probably don't know that that means -- and that's fine. My point is that, with all due respect, neither does Mr. Bittman.
The study itself, published in the journal Plos One, is excellent overall. The authors describe sophisticated methods in considerable and cogent detail. Just as importantly, they detail the many limitations of their study, and draw very reasonable and measured conclusions.
While the modeling methods in this paper are admirably robust, it remains an ecological evaluation when all is said and done. Such studies look at variables occurring together in a population, and attempt to reach reasonable inferences about how one may be influencing another. In this case, the variables of particular interest were variation in per capita availability of sugar calories, and the prevalence of diabetes.
Mr. Bittman looked at this analysis, and asserted, "This is as good (or bad) as it gets, the closest thing to causation and a smoking gun that we will see." But this is emphatically wrong.
This kind of ecological evaluation -- looking at an association between X and Y at the population level -- is certainly of some value. But it is absolutely not the "closest thing to causation" -- in fact, it is among the least reliable forms of evidence. At the population level, the presence of a Bentley in the driveway, or a high-speed Internet connection in the home, is profoundly associated with reduced likelihood of malaria (or tuberculosis, or leprosy, etc.). This is absolutely not because Bentleys or the Internet protect against leprosy or malaria. Rather, affluent people with expensive cars and high-speed Internet access are much less likely to encounter malaria or leprosy than the poor for reasons having nothing to do with horsepower, or bandwidth. Ecological studies are so notoriously subject to findings that are "true, true, but unrelated" that the phenomenon has a name: the ecological fallacy.
I am by no means suggesting the association between sugar and diabetes is ecological fallacy. Dr. Basu and colleagues did a truly masterful job of accounting for many other potentially confounding variables. And, as the authors correctly point out, the association between sugar intake and diabetes is supported by other evidence, and makes biological sense -- unlike Bentleys defending against leprosy. But still, the study falls far below the prevailing standard of causal proof.
Fundamentally, Dr. Basu's team found that the greater the availability of daily calories from sugar around the world, the more diabetes. I'm having some trouble figuring out what is supposed to be surprising about this. If people add sugar calories to their diets, they are getting both more sugar, and more calories, and those exposures alone and together are well known to contribute to diabetes risk. If they are adding sugar calories in place of other calories, then the percentage of total daily calories coming from sugar is going up, and the overall quality of diet is going down. Again, this pattern is pretty well linked to diabetes risk (and other chronic disease risk, for that matter) already.
To some extent, the Basu study was a self-fulfilling prophecy. It's true the researchers looked at other associations between diet and diabetes. But consider that one food category was "meat," which presumably lumped together everything from salmon to salami. Some such "meats" likely defend against diabetes, while others would likely increase risk. Lumped together, nothing much would be visible. Similarly, there was a category of "cereals," but with no distinction between whole grains and refined starch. Putting everything from steel-cut oats to Cap'n Crunch in one category would not facilitate keen insights. The authors did not look at nuts and seeds, despite numerous studies -- including one just published in the Journal of Nutrition -- demonstrating that higher intake of walnuts and other nuts is associated with reduced diabetes risk.
Dr. Basu and colleagues note that several countries with high diabetes prevalence rates have low obesity rates, and vice versa. It is well established that some people are heavy but metabolically healthy. It's also well established that some whole population groups, such as ethnic Indians, are especially prone to accumulate the most dangerous kind of fat -- visceral fat. With even very modest weight gain, some people are genetically predisposed to put that extra fat right where it does the most damage. The result is that some people are "thin" in their dimensions, but metabolically obese. This weakens the apparent association between obesity and diabetes, but that's misleading. The reality is that in general, gaining body fat increases diabetes risk -- and the results reported by Basu, et al. corroborate this. But weight, per se, does not account for all of the variation in diabetes risk, nor should we expect it to.
Dr. Basu and colleagues assert that if weight contributes to diabetes risk, we should focus on calories, but if sugar is an important contributor, we should focus on it. What is not at all clear is why these should be mutually exclusive. Efforts to reduce calories and control weight should focus on reducing the least valuable calories, and switching to more foods that foster vitality. Any such effort would include a focus on sugar, which has long been used by food industry elements to enhance the palatability of food and increase overall eating.
Since Dr. Basu's methods involved looking at sugar dose available in the food supply per person per day, and the prevalence of diabetes in the population, there is nothing in the data to prove that those who ate the most sugar were most likely to develop diabetes -- although one may infer that. Rather, as countries make more sugar calories available each day (i.e., become more like the U.S.), their populations have more diabetes. Again, this is not terribly surprising.
But it's more than a little wrong to conclude that since sugar intake contributes to diabetes risk, nothing else matters. In fact, Dr. Basu and colleagues state explicitly that increasing sugar availability accounts for roughly one-quarter of the global rise in diabetes rates between 2000 and 2010. This clearly implies that something other than sugar accounts for almost 75 percent of that increase!
Finally, the authors highlight the fact that "obesity does not appear to account for the major part of the impact of sugar on diabetes." The statement is true, but so is the converse: Sugar does not appear to account for the major impact of obesity on diabetes. The article includes two figures, and the one depicting the association between obesity and diabetes appears to show a stronger relationship than the other showing the association between sugar and diabetes. By asserting the one and ignoring the other, the study authors appear to be tacitly acknowledging the biased nature of their mission.
Still, the paper is very good despite that bias. It simply does not support the conclusions reached by Mr. Bittman.
For example, Bittman states* that "according to this study, obesity doesn't cause diabetes: sugar does." The study, however, says no such thing, and the statement is patently wrong. At most, the study indicates that each can contribute to diabetes risk independently of the other.
Mr. Bittman also tells us that the takeaway message is, "It isn't simply overeating that can make you sick; it's overeating sugar. We finally have the proof we need for a verdict: sugar is toxic." But again, the study does not actually prove anything -- it merely characterizes an association at the population level. Nothing in the paper refutes a role of overeating, or of obesity, in the development of diabetes. And as noted, variation in sugar intake accounts for roughly one-quarter of the variation in diabetes rates -- leaving fully three-quarters unaccounted for. Mr. Bittman would, apparently, have us ignore 75 percent of the relevant risk.
We have two problems here, each compounding the other.
First, there is a tendency for even very good scientists to place an exaggerated emphasis on one small piece of a large puzzle. A journalist asked me this week why that is, and while I don't know for sure, I have a theory. I think it's the allure of Nobel prizes, and the paparazzi.
The most valued, admired, and recognized scientific advances -- the kinds that result in a Nobel Prize on occasion -- tend to be very reductionistic, deep rather than broad. Most of the world's leading scientists have highly cultivated expertise in a very well-defined niche. This may be particularly true in the biomedical sciences, where specialization has given way to sub-specialization. This is conducive to great erudition about parts, but at the risk of failing to see the whole. That's a time-honored dilemma.
The rewards for reductionism within science are compounded by the rewards of popular culture. Our culture doesn't want to hear that the active ingredient in broccoli is broccoli -- it wants to know what supplement it can take. The fast pass to fame and fortune in our society is a magical promise of effortless success based on one simple thing. As fate would have it, just today I heard a radio commercial asserting that some particular teeth-whitening method was "like magic -- because it works!" When magic has become the standard of what really works, you're in serious trouble.
The reality, alas, is that the state of our health is not about any one thing. We can cut fat, and get fatter and sicker -- by eating more starchy, sugary junk. We can cut carbs by switching from beans to baloney, and get fatter and sicker. And we can cut sugar and consume ever more artificially-sweetened, starchy, fatty junk -- and get fatter and sicker. There are many more ways to get this wrong, than right.
The second problem -- pervasive in the era of cyberspace and blogosphere -- is pseudo-expertise. Every opinion seems to count as much as every other.
Which brings us back to Mr. Bittman. I like him and his writing in general. But he lacks any relevant expertise in epidemiology, endocrinology, or the interpretation of biomedical research studies. That he chose to interpret for us a study he did not understand was a transgression on his part, and that of the New York Times for ceding its prime real estate to someone in this instance impersonating an expert.
Mr. Bittman is knowledgeable in his field, writes well, and is clearly intelligent. But he is, in my opinion, duty-bound to differentiate between his opinion as a citizen, and genuinely expert commentary. I might opine about recipes -- but the only thing I'm really qualified to say is how much I like them. I cannot critique them with culinary expertise, and would not presume to do so. Mr. Bittman -- I stay out of your kitchen; please stay out of mine!
But we are all complicit in this. For not only do we fall in love with active ingredients, silver bullets, quick fixes, conspiracy theories, and magical promises -- we fall in love with the same ones over and over again. The book Sugar Busters came out some 15 years before Dr. Lustig's anti-sugar diatribe went viral. Déjà vu, all over again.
We live in an incredibly dumbed-down world. We are a culture that uses the technological products of science to transmit messages denigrating whatever science we choose to disbelieve (nutrition, evolution, climate change -- take your pick) while asserting that "magic works," because we wish it did. Maybe sugar is the least of our problems.
We clearly like little bits of truth we find easy to digest. But none of these is the whole truth, and when bits of truth are mistaken for the whole -- they might just as well be falsehoods. The history of public health nutrition is redolent with examples of how our indulgence in such sweet nothings has cost us dearly. The bitter truth is, it can do so again.
Dr. David L. Katz; www.davidkatzmd.com
*my column addresses Mr. Bittman's column as originally published in the New York Times on February 27th. He or others have revised the on-line column since, apparently for having recognized some of the very excesses of interpretation I chronicle here.
For more by David Katz, M.D., click here.
For more healthy living health news, click here.
For more on diet and nutrition, click here.