Sound advice or biased reporting? Breakfast as a strategy to reduce or prevent obesity or weight gain

Eating breakfast to control weight is a strategy presumed true without sufficient supporting evidence. We investigated whether biased reporting in primary literature exaggerates the protective effect of breakfast against obesity. From 92 identified studies of breakfast and obesity (all observational), ORs were calculated and meta‐analyzed from 88 study groups in 57 publications representing 30 countries. The resulting p‐value of p<10 −42 demonstrates gratuitous repetition of the breakfast‐obesity presumption. No publication bias was detected. Analysis of 88 of the 92 studies’ abstracts revealed that authors biased the reporting of their own findings: breakfast‐obesity associations were more likely to be mentioned in abstracts’ conclusions if results were pro‐breakfast, compared to associations that showed mixed, negative, or no relationships (65% vs. 39%, p=0.0492); and causal language was used in 26% of abstracts that mentioned breakfast and weight in the conclusions. Citing of others’ work was also biased: 67% of relevant citations of one breakfast RCT were overly positive toward breakfast (2% overly negative), while 48% of relevant citations of a case‐series paper inappropriately used associative or causal language. In conclusion, there is evidence of biases in the breakfast literature that misleadingly exaggerate the strength of the evidence for a presumed protective effect against obesity. Grant Funding Source : NIDDK P30DK056336 and NIDDK T32DK62710–01A1