Most Nutrition Research Is Bunk

Implausible estimates of benefits or risks associated with diet reflect almost exclusively the magnitude of nutrition researchers' cumulative biases.


Charnalak Suwannate/Dreamstime

Government nutrition advice based on decades of "research" by nutrition epidemiologists has now been shown to be mostly unwarranted scaremongering, writes Stanford University statistician John P.A. Ioannidis, who has been at the forefront of criticizing the misuse and abuse of statistics to justify the publication of shoddy and just plain wrong research in numerous disciplines.

In his justly famous 2005 PLoS Medicine article, "Why Most Published Research Findings Are False," Ioannidis concluded that "for many current scientific fields, claimed research findings may often be simply accurate measures of the prevailing bias." As the co-director of the Meta-Research Innovation Center at Stanford (METRICS) Ioannidis has turned his attention to what passes for nutrition science in a recent analysis, "The Challenge of Reforming Nutritional Epidemiologic Research," in the Journal of the American Medical Association.

As an example of how badly nutritional research violates good scientific principles, Ioannidis parses the results of a recent meta-analysis of nutritional studies that aimed to "synthesize the knowledge about the relation between intake of 12 major food groups, including whole grains, refined grains, vegetables, fruits, nuts, legumes, eggs, dairy, fish, red meat, processed meat, and sugar-sweetened beverages, with risk of all-cause mortality."

In his critique of the meta-analysis, Ioannidis points out, "Assuming the meta-analyzed evidence from cohort studies represents life span–long causal associations, for a baseline life expectancy of 80 years, nonexperts presented with only relative risks may falsely infer that eating 12 hazelnuts daily (1 oz) would prolong life by 12 years (ie, 1 year per hazelnut), drinking 3 cups of coffee daily would achieve a similar gain of 12 extra years, and eating a single mandarin orange daily (80 grams) would add 5 years of life. Conversely, consuming 1 egg daily would reduce life expectancy by 6 years, and eating 2 slices of bacon (30 g) daily would shorten life by a decade, an effect worse than smoking." These inferences are implausible to say the least.

So what is going on here? Most nutrition research are observational studies that often rely on surveys in which participants unreliably recall what they eat. And since eating is a complex activity researchers are very likely to miss confounding data that would call their epidemiological speculations into question.

Consider the notorious 1981 Harvard study that found that drinking coffee was associated with a higher risk of pancreatic cancer. The effect entirely disappeared when the confounder of smoking was taken into account. As it happens, a 2016 meta-analysis found that "high coffee consumption is associated with a reduced pancreatic cancer risk." Ioannidis is surely right that pervasive nutritional research flip-flops "may have adversely affected the public perception of science."

Instead of performing yet more dodgy observational studies, Ioanndis suggests that "large-scale, long-term, randomized trials on nutrition may be useful." And yet he immediately follows up with a devastating critique of a prominent recent study that purportedly identified beneficial effects from eating a so-called Mediterrean diet.

The Prevención con Dieta Mediterránea (PREDIMED) study supposedly compared three randomized groups: the first was given free supplies of extra virgin olive oil; another a supply of mixed nuts; and the third a bit of advice on what constituted a Mediterranean diet. The researchers were aiming to see if there were any significant differences with regard to the incidence of heart attacks and strokes between the groups.

The initial publication did find some beneficial effects from consuming olive oil and nuts. However it had to be withdrawn and re-analyzed after outside researchers showed that it was actually not randomized. The rejiggered study still found that eating nuts and olive oil reduced by a tiny amount the risk of experiencing a cardiovascular event. On the other hand, there was essentially no difference between the groups with respect to the risk of dying from any cause.

Ioannidis calls for reforming the field of nutritional epidemiology by adopting such measures as requiring that researchers make all of their data available for re-analysis by independent investigators and that results should be presented in their totality for all nutritional factors measured.

Until nutritional epidemiology is radically reformed, we should all keep in mind Ioannidis' observation that the "implausible estimates of benefits or risks associated with diet probably reflect almost exclusively the magnitude of the cumulative biases in this type of research, with extensive residual confounding and selective reporting."

In the meantime, it's probably best to follow your parents' advice with respect to diet and health: Eat and drink in moderation.