Action on reproducibility would benefit patients and the economy, say Stephen Bradley and Marcus Munafò
The UK’s strong universities and centralised health system should make it a magnet for medical research. But the evidence says otherwise: a major drug trial was recently abandoned due to low patient recruitment, part of a pattern of falling international pharmaceutical investment relative to other countries.
In response to concerns that the UK is losing out, the government commissioned former innovation minister James O’Shaughnessy to conduct a review into commercial clinical trials. His report, published in May, proposes a range of actions, including a 60-day target for regulatory reviews, ringfenced industry funding to give clinicians time to support research, and incentives for general practices to engage with R&D.
Last year’s Goldacre review on harnessing health data for research and innovation presented similarly clear-sighted recommendations, including rationalising and streamlining access arrangements, secure sharing through trusted research environments, and investing in analytics skills capacity.
Both reviews present sensible, ambitious recommendations to make UK life sciences more productive and deliver transformational innovations in healthcare. Both also draw inspiration from initiatives during the coronavirus pandemic, including the Recovery Trial of potential therapies and the OpenSafely research platform for data analysis.
Endless possibilities
Such milestones, achieved at speed during a crisis, hint at what is possible when the agility and creativity of our health and research systems are unshackled from delays and bureaucratic processes. Yet achieving a step-change in the efficiency of scientific discovery for patient benefit also requires improving the quality of research.
Here, there are deep problems. Health research generally is afflicted by systemic biases and perverse incentives. Estimates suggest most research funding is wasted due to widespread practices that mean research findings go unpublished, are distorted or can’t be reproduced. Such issues have been widely acknowledged, not least in May’s report from the House of Commons Science, Innovation and Technology Committee into reproducibility and research integrity, to which we both gave evidence. But the remedies so far have only been partial.
Requirements to register clinical trials, for example, so that negative results are not hidden in bottom drawers, have improved transparency. But researchers and journals often publish different outcomes from those that were pre-registered. For example, one study of nine trials funded by the National Institute for Health and Care Research found that eight published different outcomes from those that had been pre-registered.
For non-trials research, there is usually no expectation that methods should be prospectively registered, and so even less transparency. Researchers are free to retrospectively frame questions to match their results and massage statistical analyses to fit newsworthy conclusions.
Other problems include the failure to routinely share study protocols along with supporting data and computer code, and cultural issues such as insufficient recognition and support for statistical expertise within research teams compared with clinical researchers.
The Commons report sets out actions to address these problems. Some may require government to step in, for example challenging publishers to scrutinise research better and adopt practices to reduce publication biases—and so deliver value to the taxpayers who prop up their business model.
Similarly, keeping the UK in step with global progress towards greater transparency, such as the US government’s requirement that all federally funded research is made freely available, and EU proposals to eliminate author fees for open access publication, requires government leadership.
Where this is not forthcoming, the MPs’ report makes clear that individual organisations and sectors have the power, and therefore responsibility, to tackle systemic problems.
Harnessing the NHS’s formidable data assets and removing needless impediments to research are crucial if the government is to fulfil its ambition of making the UK a life sciences superpower. Lacking the scale and clout of competitors such as the United States, the UK cannot afford to overlook low-cost measures to refine the research system to generate consistently credible, transparent and reproducible research.
If the UK were to embrace the recommendations set out in the O’Shaughnessy, Goldacre and Commons’ reports, it has the opportunity to set the bar for medical research worldwide.
Stephen Bradley is a lecturer at the University of Leeds School of Medicine and a general practitioner. Marcus Munafò is professor of biological psychology at the University of Bristol and chair of the UK Reproducibility Network Steering Group.
This article also appeared in Research Fortnight