How more reproducibility can improve productivity
Opinion from Dr Stewart Hay, CEO of Therapeutic Innovation Australia
"In Australia $5-6 billion is invested in health and medical research each year – however it is becoming increasingly clear that a significant amount of this research may not be reproducible.
Several years ago, Bayer HealthCare reported that for a total of 67 oncology, women’s health and cardiovascular projects, only 20–25% of the published data was supported by in-house data1. Similarly, scientists at Amgen tried to reproduce published preclinical studies.
Of 53 papers describing innovative cancer therapeutics, scientific findings were confirmed in only 6 (11%) cases.
So how can we address this large systemic issue?
The problem we face is still largely undefined and we do need to better understand the drivers behind the lack of reproducibility. Data will be critical in developing our understanding of this issue. There is little doubt that the systems for data handling and analysis will play a key role in constructing a solution.
A collaborative study between Therapeutic Innovation Australia (TIA) and ANDS will be examining this significant systemic issue, with the hope of identifying meaningful interventions for improving the productivity of the health and medical research sector.
Whilst the prospect of over 80% of research being irreproducible is sobering, conversely the potential for substantial improvements to productivity is compelling."
Irreproducibility in the US
Global Biological Standards Institute (GBSI), a non-profit organisation based in Washington DC, has been studying the problem of irreproducible research. The infographic, right, accompanied a GBSI article published in PLOS Biology, June 2015.
Their study indicates that approximately US$28 billion of preclinical research in the United States annually is not reproducible.