Stanford's John Ioannidis recently discussed why the design of most nutrition studies impedes progress in the field and suggested a new kind of approach.
July 16, 2018 - By Hanae Armitage
Pervasive and compelling though it may be, research on nutritional health frequently yields less-than-dependable results. This, at least, is the opinion of John Ioannidis, MD, DSc, professor of medicine and of health research and policy at the School of Medicine. He suggests that many of the studies that aim to collect information about what we put in our bodies and how it affects our health is flawed — too small, not randomized or otherwise biased in some way. Too often, findings from one study might contradict those of a similar study, and it stunts progress in the field, he believes.
In a perspective piece published online on July 16 in Advances in Nutrition, Ioannidis and postdoctoral scholar John Trepanowski, PhD, suggest that a change is in order. Instead of many small, finely curated nutrition studies, the pair argue for a more robust approach in which scientists pool resources to answer big questions about nutritional health.
To learn more about this vision, science writer Hanae Armitage spoke with Ioannidis about the current state of nutritional research; the feasibility of conducting large, randomized trials in the field; and what Ioannidis sees as the most important questions to pursue.
Q: There seems to be a lot of contradictory findings related to nutritional research. What do you think is the main reason for that?
Ioannidis: We still largely depend on nonrandomized studies to assess questions of nutrition. These studies are notoriously incapable of giving reliable answers due to confounding factors. In nutrition, the situation is made even worse because our ability to measure diet is still limited in accuracy, and recall biases, in which study participants remember something incorrectly, can be severe. In addition, dietary intake of a single nutrient probably has small or even tiny effects on major health outcomes, even if diet as a whole is important. Therefore, any potential finding is largely shaped by the noise from errors and biases of observational studies.
Q: What is the biggest inherent problem in the current designs of nutritional studies, and how do you think it could be remedied?
Ioannidis: The biggest problem is that the vast majority of studies are not experimental, randomized designs. Simply by observing what people eat — or even worse, what they recall they ate — and trying to link this to disease outcomes is moreover a waste of effort. These studies need to be largely abandoned. We’ve wasted enough resources and caused enough confusion, and now we need to refocus. Funds, resources and effort should be dispensed into fewer, better-designed, randomized trials.
Q: Do you have a sense of how nutrition researchers feel about this approach?
Ioannidis: As you might expect, it’s double-sided. Many doctoral and postdoctoral students are being trained to continue this pandemic of flawed designs and unreliable results. But at the same time, I think many other scientists in the field see the need for a paradigm shift.
Q: How feasible would it be to implement a new approach to nutrition research in which resources are pooled for larger studies and a handful of major questions are pursued?
Ioannidis: It should be very feasible to implement this new approach. The cumulative cost would not be higher; it may even be less expensive. We would ask fewer questions, but we would get far more solid answers. We may be able to start getting some reliable evidence to inform nutritional guidelines rather than have them be battlefields of opinion.
Q: What are some of the more pressing questions that you would hope to answer through these large, randomized trials?
Ioannidis: Any question that is relevant to pragmatic, real-life handling of diet and nutrients can be addressed with such trials. We can at last get some sense of whether one diet is best, or if they’re all the same in terms of caloric intake. We can better assess specific nutritional or dietary strategies and more carefully parse the impact of health-policy decisions that are otherwise guided by special-interest parties and studies that are open to interpretation. In some cases, however, we don’t simply want to see what people eat in real life, but we seek to answer questions of physiology or mechanism. In these situations, randomized trials with direct observation of participants in experimental in-house settings could support those goals.
Other options include registry-based designs, in which randomized trials are embedded in registries and are linked to all the information that is collected, and N-of-1 designs, in which single participants sequentially get randomized into different dietary options to get personalized assessments. Mendelian randomization designs may also help generate randomized trial equivalents using information from genetic markers. In all, we have a wide array of experimental design options that can address almost any question in nutrition far more reliably than we do now with nonexperimental data collections.
About Stanford Medicine
Stanford Medicine is an integrated academic health system comprising the Stanford School of Medicine and adult and pediatric health care delivery systems. Together, they harness the full potential of biomedicine through collaborative research, education and clinical care for patients. For more information, please visit med.stanford.edu.