David S. Seres: Why nutrition ‘experts’ actually have no idea
These new guidelines have, like most dietary guidelines in the past, done little to solidify our understanding of the cause and effect relationship between diet and disease. Even worse, they’re likely to discredit nutritional science. Moreover, guidelines like these tend to suggest, without proper research as proof, that people have control over whether or not they develop certain diseases, and it is therefore their fault if they become ill.
A large part of the issue comes down to funding, and a lack of resources to do the kinds of studies that would help answer some of our most pressing questions. Our insistence on making recommendations — without having the proper research to back them up — has unintended, harmful effects.
Nutrition research, like all medical research, is extremely difficult and expensive. But nutrition research is even more complex than research in other areas of medicine. First, many well-designed studies have failed simply because the subjects couldn’t adhere to the diet: asking subjects to eat a diet lower in salt, for example, can initially make food seem tasteless.
Second, changes in diets have the potential to change dozens of other variables at the same time. When trying to determine whether a low-protein diet decreases cancer, for example, a study may involve reductions in grilled or processed meat — which may contain cancer-causing substances. To control for these coincidental effects, and subtle differences between the subjects, studies of dietary changes often require tens of thousands of participants.
Third, the impact of diet is subtle and researchers may need to watch a subject for decades in order to observe any changes.
Within nutrition, and medicine as a whole, researchers rely on two broad types of studies: randomized and observational. In a randomized study, we recruit a group of subjects with a desired set of similarities, and randomly assign them to a treatment, which in this case is a diet. Researchers then monitor the subject to see how the different treatments have affected him or her. Because the subjects are relatively similar, and the treatments randomly assigned, researchers can establish cause and effect.
Observational research takes a snapshot of a population, and looks to see whether two things occur together with a high frequency — like smiling and happiness. While less difficult and less expensive — and therefore much more popular — this type of research can only generate hypotheses about cause and effect. In this example, does smiling cause happiness? Most of our dietary guidance is based upon this kind of research.
Not only does observational research not establish cause and effect, it’s often misleading and wrong, and the recommendations that come from the conclusions of these studies can lead to harm. When two things occur together in an observational study, their relationship is actually more likely due to a third, often unknown factor than to what the researchers conclude.
For example, in observational studies there is an association between eating diets high in antioxidant vitamins and a lower rate of cancer. However, in proper randomized trials a diet supplemented with antioxidant vitamins has been shown to increase the risk of cancer. This result shows how far off the conclusions of observational research can be, and may help explain the inconsistencies in our dietary recommendations.
Because contradictory guidelines are so frequent, the public has every right to question the credibility of dietary recommendations. By failing to educate the public about the weaknesses in our research, or even acknowledge those weaknesses to ourselves, nutrition experts continue to erode what public trust remains. I worry that as long as we continue to make our recommendations sound definitive, we send the message that we do not need more research or the funding to do it.
There’s another problem that occurs because experts act with unjustified confidence in their research. The conclusion one could draw from these guidelines becomes: If you develop cancer, or another “preventable” disease, it’s your fault. I have treated thousands of cancer patients, many of whom feel failure, remorse, and/or regret about steps that they think they could have taken to avoid getting sick. People with cancer and other diseases shouldn’t be made to feel as though they were the cause when the correlation between nutrition and disease is so poorly proven.
I don’t believe that we should throw out all dietary recommendations. In fact, I feel fairly comfortable with the recommendations made in these recent guidelines. But I also know that they are not definitive — and so, intuitively, does the general public. When we experts change our minds so frequently, we send the message that all dietary guidelines should be dismissed as the latest fad.
Until and unless the funding for very large, effective, long-term randomized studies of the effect of different diets on preventing disease becomes available, nutrition experts must educate themselves and the public about the strengths and weaknesses of the data on which their opinions are based.
David S. Seres, M.D., is associate professor of medicine in the Institute of Human Nutrition.