Clinical Trials In Trouble
In recent years, growing concerns in biomedicine about the reliability of published studies have been prompting open science reforms. They include keeping published research data open and accessible to the public, in order to ensure that studies and methods can be reproduced and repeated. The end-goal is to make science publicly accessible, and guard its credibility, which will benefit both the scientific and public communities.
And this week a major reform began: the Registered Report. Journals will commit to publishing this new kind of journal article publicizing the results from clinical trials, regardless of their outcome. This is far different from the current model, which is plagued by publication bias and hidden outcome switching despite the fact that trials put safeguards such as double-blind structures in place to ensure reliability.
In reality, the reliability of clinical trials is harmed by the prevailing academic “publish or perish” mentality. Publication bias describes the tendency in academic journals for positive results to be published more often than neutral or negative results, even when the quality of the methods underlying the research is the same. More publications means more status, money, and research funding. Negative results, however, are typically labeled “untrustworthy,” and cannot advance the careers of the scientists who achieve them so easily.
Researchers learn based on these incentives that positive results are publishable and negative results are bad. This is true even though negative results — like those that prove a treatment is ineffective — are essential to advancing medicine. In fact, between 16 and 66 percent of all clinical trials fail to report results, many of them due to publication bias.
Hidden outcome switching is another problem that sees researchers changing the goals of their research so that they can find a publishable outcome, even if it isn't as relevant, or if it was never hypothesized in the study until the results themselves suggested it. Recent research estimates that one in three clinical trials change their primary outcomes once the research in complete, and a different recent report concluded that 58 of 67 trials published in first tier medical journals secretly altered their outcome measures.
Registered Reports are a simple yet radical approach to solving these problems. They are reviewed in two stages, unlike conventional reports. Researchers submit their protocol to the journal before their data is even collected. This allows for peer review of the study rationale and more robust methodology. After any necessary revisions, the journal can accept the protocol in advance, guaranteeing publication of whatever outcomes come to pass, so long as the researchers stick with their registered protocol. Final publication also requires the public availability of any anonymized study data.
If journals as a kind of third-party participant are blind to results when they assign or reject, they can make fairer choices about what to publish and give the community all of the relevant information. In addition, if peer review of the published literature and protocol is necessary, these reports could ensure that researchers stick to the trial protocols they originally registered. BMC Medicine just became the first major medical journal to offer Registered Reports, and hopefully their participation will start a trend of more accurate reporting on clinical trials.