Here's what the dietary guidelines of the government really should say

I have a confession. When the call was launched to recruit members of the advisory committee on the 2020 dietary guidelines, I thought of throwing my hat in the ring. I have a few things to say and I thought there might be room in the committee for a reporter.

But then, I looked hard and hard on my lack of graduate degrees, as well as all school reports that indicated I was not playing well with others, and I thought about it. Also, who needs all these tedious meetings where you can just use your column to tell everyone what you think he should be eating?

So this is it. If I wrote the dietary guidelines, I would give them a radical overhaul. I would go up to radically reform our way of evaluating our diet. Here's why and how.

[Did the government’s dietary guidelines help make us fat?]

The reason we know so little about what to eat despite decades of research is that our tools are terribly inadequate. Lately, as scientists attempt, and fail, to replicate results, all science looks closely at funding bias, statistical shenanigans, and group thinking. All these criticisms, even some, apply to nutrition.

John Ioannidis, professor of research and health policy at Stanford University, holds an important position in the transformation of science. In 2005, he published "Why Most Research Results Are Wrong" in the journal PLOS Medicine and since then he has made the headlines of science (though not always friends). In an editorial of the British Medical Journal published in 2013 by the British Medical Journal, titled "Un plausible Results in Human Nutrition Research," he wrote: "Almost every conceivable nutrient has peer-reviewed publications. associating with almost all the results. . "

Ioannidis said that establishing a link between diet and health – nutritional epidemiology – posed a considerable challenge, and that "the tools we are developing are not commensurate with the complexity and difficulty of the problem. is an observational research in which we collect data about what people eat and what happens to them.

Vegetables and other plant-based foods are part of a healthy diet, says Harvard's Frank Hu, but he acknowledges that we do not understand enough to prescribe combinations or a specific number of servings. (Christian K. Lee / The Washington Post)

The problem starts with this "data collection" part. There are many ways to do it, but none of them is particularly effective. You can use a 24-hour reminder, which gives respondents a chance to remember what they have actually eaten, but does not give you a representative sample of the diet in general. Food diaries over a long period do this better, but people tend to eat differently when they follow their diet for researchers. Most large population studies use meal frequency questionnaires (FFQ, in industry jargon), in which people are asked to count the portions they have consumed from a wide range of foods. , often during a year.

[Why we worry about preservatives — and why we shouldn’t]

Give FFQ tens of thousands of people and you'll end up with a ginormous repository of possible correlations. You can target a vitamin, a macronutrient or a food and go to town. But not only do you start with erroneous data, you have a zillion of possible confounding variables – food, demographic, socio-economic. Statisticians call it "the exploitation of noise" and Ioannidis is equally skeptical. "With this type of data, you can achieve the desired result," he said. "You can align it with your beliefs."

Ah, beliefs. Approximately every week, a new study on a food funded by the people who benefit from it is presented. (Marion Nestle of New York University has been monitoring this for years, and her 2018 book Unsavory Truth details her findings.) But funding bias is not the only type. "Fanatical opinions abound in the field of nutrition," wrote Ioannidis in 2013, and they also have a biased power.

So, what are we doing about it? "The definitive solutions will not come from another million observation items or small randomized trials," reads in the subtitle of Ioannidis' article. His mind is an arsonist.

In studies, the consumption of red meat is as much correlated with death in an accident as with the death of a heart disease. (Tom McCorkle for the Washington Post, Lisa Cherkasky's culinary style for the Washington Post)

Frank Hu lives in the house and is naturally less enthusiastic about the incendiary approach. He heads the T.H. Nutrition Department at Harvard. School of Public Health Chan, it can be said that nutritional epidemiology is on the floor zero. While acknowledging research gaps in its field and respecting Ioannidis' criticisms, nutrition researchers have contributed much to our understanding of healthy eating and can solve problems.

He pointed out that data collection was improving, with new tools to better evaluate food control and reality with measurable biomarkers (testing the urine of respondents in a study on the salt for sodium, for example). And he does not believe that prejudices undermine the credibility of the field. Often, they cancel, he says, and the most authoritative recommendations, such as the dietary guidelines (he was part of the 2015 committee), are the consensus of major groups concerned with the preponderance of evidence. He acknowledged, however, that there is still work to be done. "If we have no challenges, our life will be very boring."

[To lose weight, don’t just resist temptation. Remove it.]

But if the nutritional advice is not supported, a little degradation is in order. Ioannidis is in favor of fruits, vegetables and whole grains, but "the evidence behind them is rather sweet," he wrote in an email. Older observational studies showed large reductions in cancer risk, but new studies showed small, if any, benefits. "When the benefit of studies published in the literature decreases over time, multiplies by 10 or 100," you have every reason to worry about whether this type of research effort can give you an answer "Reduction has been important," noted Ioannidis, "but it's still observational data, and confusion and data-sharing issues mean we can not link the diet to the health, as Hu notes in his own research.

Great differences in what people eat align with other differences. Large plant eaters differ, for example, from big meat eaters (income, education, physical activity, BMI). The consumption of red meat is correlated with the increased risk of dying in an accident as much as dying from heart disease. The amount of faith we put in the observational studies is a judgment.

In the two decades I wrote about nutrition, my confidence in what we know about food and health has eroded and I find myself in Ioannidis camp. What did we learn, unequivocally enough, to reach a consensus within the nutrition community on the effects of diet on health? Well, trans fat is bad. Something else, and you get a refoulement from one camp or another.

In 2003, FDA Commissioner Mark McClellan introduced a new labeling label indicating trans fats. (Akira Ono / AP)

Many population studies are cut into pieces and it is almost impossible to determine the signal and noise. Researchers are trying to do this with controlled trials to test connections, but they are also problematic. They are expensive, so they are usually small and short term. People have a hard time sticking to the diet at the study. And scientists are usually looking for what they call "ultimate parameters," such as increasing cholesterol rather than death from heart disease, as it is not practical to continue a trial until the patient dies. Although I hope we can improve, it will take a while.

In the meantime, what are we doing? Hu and Ioannidis actually have similar suggestions. For starters, they both think we should look at dietary habits rather than unique foods or nutrients. They also want to view the datasets. Ioannidis insists on transparency. He wants to open the data to the world and analyze all data sets in the same way to see if "all signals survive". Hu is more cautious (partly to protect confidentiality) but believes in wider access to data and verification of results against multiple sets will help identify the actual effects.

I do not think anyone would be against that – I certainly am not – but do you remember those FFQs? You are still working with wrong data and I am not optimistic that we will get many useful tips. Ioannidis either. When I asked him if this approach would be more likely to solve the problem of dietary advice or tell us how much we know about it, he answered "probably this last one".

The important question – what are we supposed to eat already ?! – is always on the table and I have a suggestion. Abandon the evidence-based diet. It only got us into trouble and conflict. Our tools can only find the most obvious links between diet and health, and we have already found them. Instead, recognize the uncertainty and eat to protect ourselves from what we do not know. We have two excellent hedges: variety and foods containing intact nutrients (describing diets such as the Mediterranean, touted by researchers). If you severely limit your food (vegan, keto), you may miss something. Ditto if you eat foods with low nutritional value (sugar, refined grains). Oh, and pay attention to the two things we can say for sure: Reduce your weight and exercise.

When I started writing about nutrition, I used to say that I could tell you everything about the diet in 60 seconds. Over the years, my spiel became shorter and shorter as truisms disappeared, and my confidence declined in a field where we know less than more over time. I only have five seconds left: Eat a wide variety of foods with their nutrients intact, limit your weight and get some exercise.

Oh, and playing well with others is highly overestimated.