Combating Dissemination Bias in Clinical Research: Recommendations

Tip of the iceberg: Dissemination bias includes publication bias, where published papers that are exposed to readers are biased in favour of particular characteristics (e.g. positive findings). Papers with other characteristics (e.g. negative findings) are not published and, as a result, remain out of view from readers.

This week we have been highlighting some of our favourite articles from 2015, which were selected by our staff editors in December. Today’s pick looks at what can be done to combat dissemination bias in clinical research.

Dissemination bias is a problem that extends beyond clinical trials to include all clinical research. It is estimated that 50% of research findings are not published at all or only present partial information, which means healthcare professionals and other stakeholders can’t make informed decisions. The focus of our third editors’ pick, by Joerg Meerpohl and colleagues from the OPEN project consortium, was to address how this problem could be resolved. The research group drafted a comprehensive set of evidence-informed recommendations for combating dissemination bias that was discussed by a variety of external experts and stakeholders. These recommendations included the need to raise awareness among stakeholders, e.g., patients participating in clinical trials who may need to be informed that their participation only contributes to scientific knowledge and progress if the results are published or at least publicly available. Other important recommendations included the need to promote trial registration and posting of results and to support initiatives that help trials to be easily searched and identified across registries.

A large number of more specific recommendations were proposed that targeted particular stakeholder groups. For journal editors/ publishers, these included (1) the consideration of research papers for publication regardless of the direction of the findings; (2) making trial registration a compulsory pre-requisite for publication; (3) checking manuscripts next to study protocols to identify selective reporting and; (4) checking for redundant publication of results.

Combating dissemination bias in medical research is an important issue for BMJ Open and its editors. We welcome these proposals and further studies addressing this vital issue.