Tuesday, June 09, 2015

A Large Swath of Nutrition Research is Based On Lies

A Sample Completed Memory Based Dietary Assessment (Not Mine)
Well maybe "lies" is too strong a word as not all of the omissions are wilful, but the fact of the matter is, as a whole, we're nearly useless at memory-based dietary assessments, and yet they serve as the underpinning of a tremendous amount of dietary research which in turn is used to craft public health policy and dietary recommendations.

In a paper published today in Mayo Clinic Proceedings, Edward Archer and colleagues thoroughly review the futility of memory-based dietary assessments, and while I don't share Archer's overarching belief that the bulk of obesity is explicable on the basis of maternal obesity's impact on developing fetuses, or his contrarian viewpoint that diet is not a major risk factor for disease, his paper does an admirable job of destroying the notion that 24-hour dietary recalls and food frequency questionnaires (FFQs) are useful.

Before getting to the meat of the paper, back to Archer's assertion that diet is not a major risk factor for disease. To illustrate that point Archer notes,
"This hypothesis is supported by multiple lines of evidence, such as a 40% decline in the age-adjusted mortality rate from 1969 to 2010, a progressive decades long reduction in age-adjusted cardiovascular disease incidence and mortality, and a 1.5% per annum reduction in age adjusted mortality rates from all major cancers as well as significant reductions in lung cancer incidence in men and women between 2001 and 2010."
Given that cigarette smoking has decreased dramatically in that same time period (see graph below),


and the fact that smoking is America's number one cause of death and lung cancer, and is a major cause of cardiovascular disease, that rather than point to smoking cessation as the improvements' probable cause, he instead suggests, absent of any pathophysiological explanation or hypothetical mechanism, that the standard American diet is responsible for the observed decrease in death and lung cancer, is plainly bizarre. All the more so when considering the impact the introduction of Westernized diets have been seen to have on non-communicable disease risks in populations the world over.

That aside (and truly, I wish it weren't part of the paper as it detracts from its impact), the paper's a worthwhile read. With an expansive review of the literature, Archer explains not only how far off the mark dietary recall is (up to 80% of some recall studies include dietary patterns that are "physiologically implausible"), but why, covering studies that explain how,
"24hr and food frequency questionnaires can be most accurately defined as mere attributions based on mental experiences that are strongly influenced by the respondents' idiosyncratic qualities (ie, education), previous memories and information, knowledge and beliefs, motives, goals, habitual behavior, and the social context in which the memories are encoded or reported."
He also asserts that the design of memory based dietary assessments itself inspires false reporting and posits that,
"it is not a question of whether FFQs induce false reporting, but to what extent".
Finally Archer discusses the criteria for scientific research and he suggests that memory based dietary assessments are,
"akin to creation science in that they fail to meet the basic requirements of scientific research"
in that they are not independently observable or measurable (reporting you've eaten an apple, and how much of that apple you ate, isn't the same as having your apple eating independently observed and the amount of apple you consumed measured), they are easily and regularly falsified, they lack validity, and they are unreliable. This leads Archer to call the discipline of nutrition epidemiology "pseudoscience", stating that,
"when a person provides a dietary report, the data collected are not actual food or beverage consumption, but rather an error prone and highly edited anecdote regarding memories of food and beverage consumption"
and hence,
"do not meet the basic requirements of the scientific method and, by definition, are pseudoscientific when presented as actual estimates of energy or nutrient consumption."
Notwithstanding the non-evidence based ease with which Archer dismisses the potential impact of diet on health, this paper soundly illustrates the need for the rapid development and adoption of an objective means to track dietary choices - something the technologies of today are well suited to provide.