Best known for fathering modern nutrition in the United States, for pioneering nutrition research and education in the country, and for having developed the system of measuring energy in food that is used throughout the world today, a lesser known fact about USDA’s first chief of nutritional investigations Wilbur Olin Atwater was that he also fought for our right to enjoy a good stiff drink.
In the early 1880s, the US temperance movement was in full force and widely promoted the myth that alcoholic beverages were “poison,” destructive to families, and that drinking them served only to fill the greedy pockets of saloons and breweries. One of the most influential women in the movement was Mary Hunt, who in alliance with the Woman’s Christian Temperance Union (WCTU) had managed to convince politicians to pass laws that required physiology textbooks in schools to encourage complete abstinence of alcohol and its prohibition.
Vehemently protesting the measure was Atwater, arguing that alcohol was not “poison” and that children should only be taught the “simple facts” supported by science and that those facts be free of “exaggerated theories” and “errors.” But, for his position, he suffered personal criticism by Hunt, as well as attacks on his career and on his funding from the USDA, even his ability to publish scientific papers.
Biostatistician David Allison, professor at the University of Alabama-Birmingham, and 2014 recipient of the W.O. Atwater Lectureship awarded by the American Society for Nutrition (ASN), drew parallels between Atwater’s and some of his own experiences, such as his publications regarding the science of sugar-sweetened beverages, in fighting for rigorous research in obesity and nutrition.
In his lecture given on Tuesday, April 29, at Experimental Biology in San Diego, Dr. Allison said, “As you see, these things are repeating today. I think we need to try to try to learn those lessons from history, try to come out in favor of truthfulness in science, saying what we believe, and being able to engage in dialogue without trying to shut each other down.”
When we take the long view, Dr. Allison said, there has been positive progress made across the field of nutrition science at a practical level. For example, we have effectively eliminated lead and other toxins from our wine and foods. We have made many advances in agriculture and have come to understand much about nutrition, which has allowed us to radically reduce hunger and nutrient deficiencies in most of the developed world.
There has also been a lot of positive process in understanding the basic underpinnings of obesity and energetics. It has occurred through several historical intellectual lineages, Dr. Allison said, including through advances in areas of body composition measurements, genetics, endocrinology, nutritional physiology and biochemistry, and the design of randomized controlled trials.
Nevertheless, Dr. Allison said, there are questions being raised about the progress, scientific quality, and integrity of the field of nutrition. It’s most apparent in the popular press, but also exists among well-established people in the scientific field, particularly in regards to research in obesity because of lack of progress in eradicating it. “What we have to acknowledge is that at the population-level, our progress in reducing obesity levels has been nil to negative,” he said.
Why do we struggle so much to make progress in obesity and nutrition? There are probably many contributing factors, Dr. Allison said. But he identified a subset of factors and practices that he believes impede progress in both understanding and in reducing obesity levels. In addition, for each factor, he offered a solution.
One of the factors is choice of studies. “We have an over-reliance on observational studies,” Dr. Allison said. And he emphasized the word over-reliance so as not to suggest that we didn’t need any observational studies. But on average, he said, observational studies rarely “point us in the right direction.”
The problem with observational studies is not necessarily the execution, but is inherent, he said. Every good author knows that observational authors show association, not causation. But then they look the potential threats to show causation, he said. Even if one could effectively eliminate measurement error, genetic variation, smoking, socioeconomic status and other usual suspects, there are still confounding problems that don’t allow for the same level of confidence as other experiments such as randomized controlled trials.
“There’s just a fundamental concern that no matter how well you do an observational study, you can’t be guaranteed that it will give you the answers that you could get with a different experiment,” he said.
Another concern is that short-term studies are insufficient. We still need them and should do them, he explained, but they don’t account for learned compensation in biological organisms with adaptive systems. He said, “We’re not simple machines. We adapt.”
Other studies are simply conducted in ways that are “beyond the point where they are probative,” he said. They aren’t able to “move the needle” of our knowledge. The problem is that the mere exposure and frequent repetition of these lead people to believe in falsehoods. An example is the assumption studies make that breakfast promotes weight loss—something most of us have heard repeatedly—despite lack of evidence from randomized controlled trials that has shown this to be true. “We not only we need to stop doing studies that are a waste of time. We need to the proper studies that add knowledge,” he said.
Additionally, he said, there’s a failure in nutrition research to take measurements seriously as happens in other domains. An example of this is in use of self-reported data versus other data measurements. “In the age where physicists are investigating the Higgs Boson, we still use self-reported methods,” Dr. Allison said. “As a field we ought to strive for more.”
It’s also important to take design seriously, he said. The question one should ask when designing randomized controlled trials is, “Are you studying what you think you’re studying?” For example, during a study of a treatment, it may be unclear whether the treatment is providing the effect or the motivation for the goal such as weight loss during the study.
Dr. Allison also highlighted a common temptation for scientists to be complicit with the media and put out misleading statements. In several cases, a press release or article will use a headline and language that is completely unsupported by the study, Dr. Allison said.
Bias creeps in too whether in the form of industry bias or “white hat bias,” a term Dr. Allison coined. It can happen through distortion through statistical fiddling in obesity trials. Or, the results of the paper may be reported accurately, but causal language that is unsupported by the data is found in the conclusion. “It’s conclusion spinning,” Dr. Allison said.
Oftentimes in observational studies, authors will use causal language despite that the study only showed correlations. They also often use “emotion-raising language”. One example is the word implicated. “That’s a funny word. Does it mean causal or associated? Isn’t that usually reserved for criminals? I think it means I can’t demonstrate cause because all I have are correlations, but I can call it criminal,” he said.
If progress in the fields of obesity and nutrition science is to move forward, Dr. Allison said, then there are five points that should be considered in producing sound, meaningful research:
- We need to recognize that humans are adaptive systems.
- We need to treat obesity as a science as meriting the same rigor as any other science.
- We need to recognize that short-term studies, studies of intermediary endpoints, and observational studies all have their place, but should not be our stopping points or bases for overreaching conclusions.
- We need to develop a set of ‘meta-methods’ (e.g. clinical trials registries, CONSORT statements, public data sharing), which will collectively buttress/ensure the implementation of the fundamental scientific methods that already exist.
- We need to put resources in place to implement those meta-methods.
“Finally, and unfailingly, we need pursue truth through science not as a job, but as a discipline, a vocation, and a privilege,” Dr. Allison said
Like Wilbur Atwater, Dr. David Jenkins, Dr. James Hill, and others, the new recipient of the Atwater award deserves hearty recognition and praise for supporting the rigorous use of scientific research in obesity and moving the field of nutrition science forward.
Evolving Health 
Thanks so much for posting on this. Looks like Allison gave an excellent talk. He’s spot on about the problems in obesity/nutrition research. I wonder if he is getting pushback about being a shill for the soda industry? That seems to be what happens in this sort of situation. Mind if I reblog?
Yes, he is getting pushback. Of course I don’t mind, Adele.
This sounds pretty extreme. In what way is he a shill? It is precisely in fructophobia that all of the abuses he mentions play out.
What I’m seeing (Marion Nestle is one of the worst offenders) is that if anyone says “hmmm, perhaps we should put more science in nutrition science”–which is a message that, for some reason, does not seem to come from NIH-funded sources–the response is not to consider and evaluate the message, but to attack the messenger as being a “shill” for whatever industry or group did provide the funding for the researcher to do the investigations related to that idea. This was kind of rhetoric used to shut down opposition to the initial 1977 Goals; it is still the first line of defense against anyone criticizing the status quo.
See for instance: http://www.foodpolitics.com/2013/10/annals-of-nutrition-science-coca-cola-1-nhanes-0/ This example is not about David Allison, but about the rhetoric.
As a clinical trials biostatistician, it’s very refreshing to here for someone calling for more randomized studies rather than in how to avoid them. The era of “big data” is upon us, and the general trend is to try to avoid using clinical trials by trusting in what amount to very large sample size observational studies. Funding for randomized studies in many fields from heart disease to diabetes to oncology is less available than it was even ten years ago.
Saying that, I’d be curious to know what recommendations Dr. Allison has for practically carrying out such studies. You certainly aren’t going to be able to house people in institutions and monitor their food supply or rigorously control their diet for all but very short term studies. Any data collection on diet is going to be self-reported data, and inaccurate on quantity as well as nutrients, especially if sample sizes and time frames are large enough to conduct the end result studies he suggests are essential. At best, you can get “intention to treat” analyses of dietary recommendations and education, perhaps with some sort of structured support. You can learn what recommendations and what support interventions result in controlling weight. But actual long-term randomized studies of the biochemical effects of diet with the kind of rigorous data collection he seems to propose do not strike me as practical.
I can share his frustration with the poor reporting of the observational data, but I’m not sure randomized studies of the effects of actual nutrient consumption are going to be feasible.
Gah! Hear, not here.
Is his talk published? What is the pushback?
Sorry. There wasn’t any pushback at the conference. I was referring in general to pushback.
Its not just in nutrition science, its more widespread. Today it seems acceptable to claim consensus equals truth. Science is supposed to be about challenging consensus to find deeper truth. Consensus science is an oxymoron but how many in science understand that. Giving undue credibility to consensus has led us into the health problems we have today in which the misunderstanding of fat metabolism is at the root of the problem. Perhaps society has lost the skill for rational debate so that challenging a long-held opinion is viewed as a personal attack and is rebutted with a personal attack.