Dubious findings about food from the ‘Nutrition Researchers Guild’: How can we learn from this unscientific manipulation of statistics?

Are you confused about conflicting “analysis” findings on sure meals’ results on our well being? It would hardly be stunning. First, butter is the enemy; then, it’s stable margarine. Is caffeine good or bad to your coronary heart? For a time, beta-carotene dietary supplements are thought to forestall most cancers — till they’re discovered to extend the chance of lung most cancers in people who smoke. And lastly, does a woman’s diet at conception determine the sex of her fetus

When you do a deep dive into the methodology of research that produced such conclusions, it’s not stunning that they’re inconsistent or implausible.

Let’s take into account who performs such analysis. 

A department of pseudo-science has turn out to be a form of throwback to a phenomenon that started centuries in the past — a “guild,” a formal group with a standard curiosity, or a group of craftsmen or retailers which have mutual pursuits and requirements. (Or, as mentioned beneath, a lack of statistical requirements.) 

Entry right into a guild initially was by apprenticeship the place pay was low whereas a craft was being discovered. There have been goldsmiths’ guilds, and in Holland, even painters’ guilds. Sometimes, commerce guilds ranged over massive areas and infrequently have been aligned with or protected by governments. Medieval cities supported guild monopolies.

Today, some scientists — or, as they’re extra precisely described, “pseudoscientists” — have shaped what quantities to a guild of researchers whose intent, it appears is to undermine the science of meals.

Here’s what they do: They carry out what are known as “cohort studies” that embody the outcomes of questionnaires surveying responders’ meals consumption habits. A cohort examine circumscribes a gaggle of individuals and follows them longitudinally over time, serially asking questions and testing associations, or correlations — for instance, whether or not sure patterns of meals consumption predispose to coronary heart illness or dementia or alter longevity. 

England’s “The Life Project,” funded by the U.Ok. authorities, was one of many first cohort research. Over a few years, it adopted all the youngsters born in a window of time. Eventually, new cohorts have been shaped and adopted. The knowledge from questions requested and answered within the 1958 The Life Project cohort has given rise to some 2,500 papers, and continues to be ongoing.

Credit: Alamy

The variety and amount of meals consumed arguably have an effect on well being, so it’s logical that cohort researchers need to accumulate details about meals that may very well be associated to subsequent well being outcomes. To facilitate this, dietary researchers at Harvard University developed a semi-quantitative “food frequency questionnaire,” or FFQ, within the mid-Nineteen Eighties. 

The FFQ is a self-administered dietary questionnaire which asks folks how usually they eat particular meals. The preliminary FFQs had 61 meals. The selection of that exact quantity was in all probability intentional, as a result of there’s a 95% likelihood of a minimum of one optimistic outcome in case you ask 61 (impartial) questions. 95% is usually set as the extent mandatory for a examine to statistically make its case.

 Over the years, many cohort research have been began; and FFQs with increasingly meals represented turned a regular factor of lots of them. In a research project of the National Association of Scholars that examined a set of 105 papers involved with the consequences of purple meat’s results on well being, the variety of meals included ranged from 14 to 280, with a median of 51. 

meat caution imgcache rev ed cdfbfd c fbc dc d bdb x
Credit: Health Initiative

The questionnaires may ask, for instance, how a lot citrus, broccoli, avocado, and many others. the topic consumed, and what well being issues — hypertension, despair, angina, bronchial asthma, weight problems, erectile dysfunction, glaucoma, and so forth — she or he skilled. The variety of entries in each classes is massive and, due to this fact, the doable associations are huge. 

That presents one thing of a statistical dilemma — actually a quagmire — as a result of dietary epidemiologists hardly ever, if ever, modify their statistical evaluation to account for the variety of questions examined and the probability {that a} sure fraction of supposed correlations will come up by likelihood. Putting it one other approach, conducting massive numbers of statistical checks — that’s, looking for correlations in a examine — produces many false positives by likelihood alone.

What’s regarding now?

An additional complicating issue is that the supposed correlations could be sliced and diced based on gender, age, race, and many others., and likewise at a number of time factors. To make clear this conundrum, the just-released landmark report from the National Association of Scholars (NAS) challenge, Shifting Sands: Unsound Science and Unsafe Regulation, examined how these sorts of likelihood correlations yield false, irreproducible conclusions that have an effect on varied areas of presidency coverage and regulation at federal businesses. The report applies Multiple Testing and Multiple Modeling (MTMM) to evaluate whether or not a given physique of analysis has been affected by such flawed practices, wherein case the declare could be decided to be thought-about unreliable.

Non-statisticians won’t admire the importance of this discovering, nevertheless it presents a basic drawback: Without acceptable adjustment for the variety of claims examined, the analysis findings of dietary researchers are sometimes irreproducible. In different phrases, nevertheless persuasive or probably necessary the findings appear, they’re actually meaningless.

What does this flawed strategy of researchers who use dietary meals frequency questionnaires should do with their constituting a guild? They maintain their analysis knowledge non-public; they don’t criticize each other over questionable statistical strategies; they promote their monopoly “experience” to the federal government in return for grants; they usually police their commerce themselves (i.e., they referee one another’s papers). They additionally defend their very own: Several years in the past, for instance, fourteen of the “guild members” demanded that an article criticizing the analysis of one of many members be retracted. 

screenshot am

Nutritional researchers have constructed a really snug, worthwhile, and handy pseudo-science guild whose members know, or ought to know, that, within the curiosity of padding their article depend, they’re typically publishing irreproducible findings. And, within the course of, they’re filling the scientific literature with tens of 1000’s of nugatory meals frequency questionnaire-based research that confuse and frighten the general public.

Consider, for instance, the 2008 article, “You are what your mom eats: proof for maternal preconception weight loss plan influencing foetal intercourse in people.” It generated numerous consideration by making the genetically implausible declare that ladies who eat extra breakfast cereal usually tend to have a boy youngster. That result’s simply defined by likelihood. On what was the article primarily based? On observational (cohort) research.

That’s not a lone instance. Claims from cohort research (e.g. FFQs) in regards to the results on human well being from varied interventions equivalent to vitamin consumption or low-fat diets have not been supported by large randomized clinical trials.

What can we do about this example? It’s tough, as a result of the established order represents collusion among the many dishonest researchers themselves, predatory journals that may publish something for a hefty payment, and tutorial departments that overlook skilled misconduct. They are all ‘benefiting’ in a method or one other from simply generated however usually specious analysis. Until a number of of them breaks the cycle, nothing will change. 

Perhaps funding businesses, which have the facility — and accountability — to police scientific integrity will start to do their job and crush the guild. We aren’t optimistic. 

Henry I. Miller, a doctor and molecular biologist, is a senior fellow on the Pacific Research Institute. He was the founding director of the FDA’s Office of Biotechnology and a Research Associate on the NIH. Find Henry on Twitter @henryimiller

Dr. S. Stanley Young is a Fellow of the American Statistical Association and the American Association for the Advancement of Science. He is an adjunct professor of statistics at North Carolina State University, the University of Waterloo and the University of British Columbia the place he co-directs thesis work.


Recommended For You

About the Author: Adrian

Leave a Reply