Abstract

Fixing Evidence-Based Medicine requires transparency, strategic campaigning, and education When patients ask their doctor about the benefits and harms of a treatment, the doctor can only say what is likely to be true based on the volume and quality of evidence available. Poor research integrity misleads doctors, meaning what we tell our patients might simply be wrong. In 2001 a single misreported trial, study 329, led to millions of adolescents being prescribed an antidepressant that didn’t work, and caused suicidal behaviour. This was entirely avoidable. I believe there are three fixes to such problems in evidence-based medicine: transparency; strategic campaigning; and education. Transparency so we know about missing data and poor methodology Half of all clinical trial results are never reported, and positive trials are twice as likely to be published as negative trials. This means we often don’t learn about harms data, and the data we have are biased. Hidden documents mean we don’t know when publication bias is happening: readers accept biased results, because they trust journals and authors to maintain reporting integrity. Transparent reporting fixes publication bias, allowing us to see the most accurate picture of the evidence available. Transparent trial registration, protocols, and editorial practices allow us to detect poor methodology and interpret the evidence accordingly. Such transparency has very real consequences: If all data had been transparently reported, study 329 would have shown the lack of efficacy and increases in harm of paroxetine, and thousands of lives may have been saved. Strategic campaigning to effect real-world change Outcome switching and publication bias remain prevalent, showing decades of prevalence studies do not in themselves solve these problems. Guardians of evidence-based medicine must be more strategic in fixing its flaws. There are strong financial, academic and personal incentives for trialists to comply with requests from journal editors. Appealing to editors to require full registration and transparent reporting is therefore likely to motivate trial authors in good practice. Incentives such as CONSORT and ICMJE do just this and provide a standard to which journals can be held accountable. Funding is arguably the biggest barrier to research. Plan S sets out 10 principles to improve research funded by public grants, including funders’ research integrity requirements. Future projects should learn from this: researchers are likely to comply with such requirements to secure funding. Education so people know what to do Responses to correction letters in the COMPare project showed widespread misunderstandings around the reporting of trial outcomes. Education is needed to show doctors and patients what good trial reporting looks like, so they can make informed clinical decisions with the evidence available. In COMPare we did this through journal correspondence, leading to policy changes and extensive discussion in the clinical trials community. Education should be considered more broadly: teaching EBM at schools sets standards for the next generation of leaders; educating the public creates appropriate intolerance to poor research practice; and open discussion of existing research educates doctors and patients about poor research practice affecting clinical decisions.