LAUSR

Last year, a nutrition scientist at a prestigious American university was charged with academic misconduct owing to the use of flawed research methods [1]. One of the methods that led to the misconduct charge was “P-hacking”, defined by Simonsohn et al. [2] as conducting multiple analyses but only revealing the results with P values < 0.05. Thus, Phacking is “data dredging” or a “fishing expedition” followed by selective reporting, with P values singled out as the driver of the selection process. Selective reporting can lead to a type of publication bias that can result in falsepositive findings that are unable to be replicated. In an effort to reduce this and other types of bias, in 1996 a group of experts that included medical journal editors, clinical trialists, epidemiologists, and statisticians published the Consolidated Standards of Reporting Trials (CONSORT) statement [3] with the goal of encouraging complete and transparent reporting of trial data. The original statement was followed by revisions with checklists [4, 5] and two publications that more fully explained the rationale behind the guidance [6, 7]. Since then, > 20 extension papers have been published that address specific types of trials. In 2018 over 600 journals endorsed the CONSORT guidelines, typically in the form of a supportive statement in the Instructions to Authors [8]. Some journals require submission of the CONSORT checklist [4] prior to publication. This checklist explicitly queries 37 points (numbered as 25 items), and the authors must list the page of the manuscript where each point is addressed. Thus, the checklist provides a detailed, concrete tool to guide the reporting of randomized trials. Nevertheless, a gap remains between theoretical guidelines and actual practice. The purpose of this commentary is to address that gap as it impacts selective reporting in community-based childhood obesity prevention trials. In the field of obesity, norms and practices regarding trial methodology differ depending on the setting or context, i.e., in a traditional research laboratory; part of a pharmaceutical development process; in medical model weight loss programs; or at the population level. In this commentary, I highlight a community-level study and show real-life examples of how CONSORT guidelines were interpreted in an obesity prevention study in children. The examples illustrate practical approaches that may help other researchers comply with the CONSORT principles and promote transparency in the reports of their research.