During coronavirus, science is moving too fast to always be accurate. We should all be concerned

The Lancet's concern about published papers on Hydroxychloroquine show that media interest does not necessarily promote good science

Many aspects of life are amplified in the pandemic. We see great acts of generosity and kindness and appalling acts of greed, selfishness and, with people cooped up together, violence.

This amplification also affects what and how science is being explored. Extraordinary efforts to find and test a vaccine continue. Randomised trials of possible treatments are being conducted around the world, with many of the normal, legitimate barriers to such trials being lowered in terms of the time taken to organise and approve them to enable a vaccine to be tested rapidly. Significant observational studies have also taken place in record time, in which large volumes of data have been analysed to produce results at a speed that has never been seen before.

At the same time, poor science has surged, and the growth of pre-print servers has led to findings being publicised without the usual formal peer review. The New England Journal of Medicine and The Lancet have both published recent expressions of concern over papers about Hydroxychloroquine from one research group, demonstrating that even traditional peer review, while often leading to improvement in the final publication, is no guarantee that the results are reliable.

The website Retraction Watch confirms that the majority of papers published and subsequently retracted have been peer reviewed. This process is not a perfect filter for bad science, or even for outright fraud and misconduct. Currently Retraction Watch lists 13 papers related to Covid-19 that have been retracted in addition to three expressions of concern.

For most of this year, science has been frontpage news and discussed in other media as never before. However, media interest does not necessarily promote good science. While it is important in a pandemic to respond rapidly, the scientific community may need to reflect on whether it has at times been pushed, perhaps by media interest, into doing things too quickly.

A lot of media coverage of poor research never leads to good outcomes. We saw this with MMR and autism, but the media in general seems to have learnt a lot from that experience of over 20 years ago and has been highly responsible in the overall reporting of science.

Science as an enterprise tends to self-correct, but that process is sometimes slow and there is no guarantee that all problems will be detected.

The rewards for reproducing others’ research are not as great as those for being first in the field. This is partly a human problem and partly the structure of scientific reward systems that favour innovation over confirmation.

Stephen Evans is professor of pharmacoepidemiology at The London School of Hygiene and Tropical Medicine, and is a collaborator in OpenSAFELY, an analytics platform for electronic health records in the NHS