Breakfast, bias, and obesity: Distorted research reporting contributes to beliefs beyond scientific evidence

“Breakfast is the most important meal of the day” and other such maxims reflect a moral and health-related halo surrounding breakfast. Breakfast is purported to instill numerous health benefits to those who consume it, not the least of which is weight loss and weight maintenance. However, the scientific literature does not necessarily support a causative role between skipping breakfast and obesity. Yet, information sources including blogs, popular health icons, and government agencies have made statements that eating breakfast will help control weight. In our new study in the American Journal of Clinical Nutrition, we quantify ways in which the scientific community may be contributing to these premature causal claims.

We previously highlighted common misconceptions in obesity research, but in this article we wanted to evaluate factors contributing to the spread of such misconceptions. To do this, we analyzed research reporting behavior using breakfast as an example, which we believe will represent other obesity presumptions, as well.

We first confirmed that breakfast being protective against obesity is, in fact, widely accepted as true by looking at supporting statements from information outlets. We then reviewed the scientific literature to establish that the totality of research confirms it is only a presumption. Specifically, few randomized controlled trials have been conducted looking at breakfast and obesity, and those that have been done have equivocal results. Associations exist, but mechanistic studies do not necessarily support a causal relationship. For instance, one common assumption is that skipping breakfast will cause over consumption of energy later in the day, but mechanistic studies often show no difference in, or even a lower, energy intake after skipping breakfast.

This leaves us with many observational studies demonstrating that breakfast and obesity are associated. In fact, there are so many associations, that many of them likely did not substantively add to the scientific body of knowledge (something we labeled ‘research lacking probative value’). We found over 90 studies that reported associations between breakfast and obesity, of which 78 were cross-sectional. We cumulatively meta-analyzed 58 studies containing odds ratios, and demonstrated that after the second year of available association studies (1998), there was a highly significant association (p<0.001); by 2011 the association was p<10-42. After all of the effort spent from publishing these associations, we are still left with the question of whether a causal relationship exists.

So how did ‘breakfast prevents obesity’ come to be so widely believed? The common refrain is that lay media misunderstand and miscommunicate research, and that the benefits of breakfast on obesity is a conspiracy of breakfast-food companies. However, we demonstrated that the primary literature often overstates the known relationship between breakfast and obesity. In particular, we showed that there is a substantial use of causative language when describing observational evidence about breakfast and obesity, both when researchers cited others’ work as well as when summarizing their own work. In addition, there is evidence that researchers misleadingly cite others’ studies to support the presumed effect of breakfast on obesity. Specifically, when authors would cite a study that had results both supporting and refuting the breakfast-obesity hypothesis, authors often would exclude the information against breakfast. Even in their own abstracts, there was a tendency for authors to only make conclusions about breakfast and obesity when the results were in favor of eating breakfast.

Overall, we are not sure, on a population level, about the influence of breakfast on obesity. A common question we get after sharing this study is, “So does that mean we shouldn’t eat breakfast?” Uncertainty is not the same thing as evidence of no effect. It means that we do not know if getting everyone to eat breakfast would increase, decrease, or have no effect on weight. And uncertainty is okay. Furthermore, we used breakfast and obesity as the example; the relationship between breakfast and mood, cognitive performance, and glycemic control may be different. Finally, the study was meant to focus on bias in research reporting, using the presumed effect of breakfast on obesity as a vehicle. If we expect health writers and the public to understand research on health, we in the research community should make sure we are communicating it to the best of our ability.

Read what others have to say: