The COVID-19 evidence base

July 15, 2020 Jon Williamson

Medical science has clearly struggled with COVID-19. There are plenty of sophisticated epidemiological models, yet they have been very inaccurate: a child who joined the dots of weekly excess mortality figures would have been much more accurate with regards to where the UK is on the infection curve. New treatments and risk factors have been touted on the flimsiest of grounds. Public health advice has been tardy and inconsistent. Governments have been keen to ‘follow the science’ but the science has been taking a random walk in the park. Why?

Medical science is reliable where, and only where, there is plenty of relevant evidence. Moreover, it is prone to take a very narrow view of evidence. The scope of present-day evidence-based medicine (EBM) is restricted to certain kinds of association study, such as randomised controlled trials (RCTs) and prospective observational studies. EBM offers little when such evidence is not available, as has generally been the case so far with the COVID-19 pandemic.

EBM+ offers more. EBM+ is a development of EBM which systematically takes mechanistic studies into account alongside association studies. There are good reasons for taking mechanistic studies into account: they can enable reliable decisions in the absence of high-quality RCTs, and they are crucial when it comes to extrapolating results from one population to another.

This recent paper argues that coronavirus research needs to be situated within an EBM+ framework:

Jeffrey K. Aronson, Daniel Auker-Howlett, Virginia Ghiara, Michael P. Kelly and Jon Williamson: The use of mechanistic reasoning in assessing coronavirus interventions, Journal of Evaluation in Clinical Practice, 2020.   doi: 10.1111/jep.13438

Medical science will only become worth following if it starts making best use of the evidence that is available. We have the methodological tools, but we need to start using them.

Jon Williamson
University of Kent