If you’ve ever found yourself wondering why so many nutrition studies seem to contradict each other – and why studying nutrition and cancer is so difficult – you are in good company. I recently returned from the American Society for Nutrition meeting at the 2013 Experimental Biology conference in Boston and as usual, this was a prominent topic of conversation.
This year at the conference a session that particularly drew my attention was sponsored by ILSI North America. Presentations and videos from this session are available on the ILSI website.
One of the speakers in the session, statistician David Allison, PhD, suggested provocative actions that challenge the way research is traditionally conducted and publicized. According to Dr. Allison, all too often published research is distorted and misleading.
Some of the reasons for this include the common practice of not reporting all the analyses that are conducted in a given study, not publishing (or being able to get published) results from studies don’t find effects, and overstating the importance of a study’s findings.
Actions he proposes to address these concerns include:
• Researchers could publish a description of their data analysis plans before publishing studies and make their data available for others to analyze. This practice would discourage the practice of conducting numerous analyses of data sets but reporting only the analyses that find effects. It would also allow researchers to check and replicate each other’s work.
• Press releases from journals could be reviewed by independent scientists in the same way journal articles undergo peer review. Writers who report on findings from journals often rely on the journal’s press release to describe a study and its results. If the press release is distorted the writer’s article will be, too.
• Funders could require publication of all studies conducted, even if the publication is a simple technical report published on the web. When studies are funded and conducted but results are never reported, other researchers and the public have no way of knowing the work has ever been completed. We cannot hope to fully understand any particular research question if much of the research addressing it has never been published.
These proposals would undoubtedly be helpful – but only if the research community embraces them. I hope to see a lot more discussion about these and other potential solutions in the coming weeks and months.
Stay tuned – but in the meantime remember that there is a lot we do know about how nutrition and lifestyle are important determinants of cancer risk. See AICR’s evidence-based recommendations and comprehensive reports for more information.
Susan Higginbotham, PhD, RD, is AICR’s Director of Research.