Carl Heneghan and Tom Jefferson:

Poor quality research is being used to justify a policy with enormous consequences for us all

Queuing outside shops, dodging each other once inside, and not getting too close to other people anywhere: social distancing has become the norm. The two-metre rule, however, is also seriously impacting schools, pubs, restaurants and our ability to go about our daily lives. The Government is reviewing the policy, but what evidence is there that keeping our distance makes any difference to catching Covid?

The influential Lancet review on this matter provided evidence from 172 studies in support of physical distancing of one metre or more. However, all the studies were retrospective and suffer from biases that undermine the reliability of their findings. Recall bias arises in research when participants do not remember previous events accurately, and it is problematic when studies look back in time at how people behaved, including how closely they stood from others.

More concerning was that only five of the 172 studies reported specifically on Covid exposure and proximity with infection. These studies included a total of merely 477 patients, with just 26 actual cases of infection. In only one study was a specific distance measure reported: “came within six feet of the index patient”. The result showed no effect of distance on contracting Covid.

In another study of 121 healthcare workers exposed to a patient with unrecognised Covid-19, three workers went on to test positive. Yet all three had unprotected patient contact: two never wore a face mask, respirator, eye protection or gown, making it impossible to identify the specific effects of distance. In the one study with the most substantial effect, no distance measure was reported. It was designed to test any association between sleep quality, stress and risk of infection, not distance.

On further independent inspection of 15 studies included in the review, we found multiple inconsistencies in the data, numerical mistakes and unsound methods in 13 of them. When assumptions of distance were made, we could not replicate any of them.

Might evidence from so-called super-spreading events show more promise? Following a choir practice in Skagit County, Washington, 32 cases of Covid were confirmed among the attendees. But how Covid was transmitted has not been established. Several opportunities for droplet transmission occurred, but also fomite transmission via objects which are likely to carry infection. The choir members did sit near to each other, but they also shared cookies and oranges. No one reported any physical contact. Vital information is missing.

The Lancet review says “robust randomised trials are needed to better inform the evidence for [social distancing] interventions”. We do not disagree, and neither do the Centers for Disease Control Prevention in the US. In a supporting review of nonpharmaceutical measures for pandemic influenza, they found the evidence base was mainly derived from observational studies, was poor quality and that there was a need to undertake controlled studies to clarify the effectiveness of social distancing.

The evidence does show that the risk of catching infections is higher in healthcare settings than in the community, and higher indoors than outdoors. What the evidence cannot say is that there is any measured distance that reduces your risk. The tendency to develop, interpret and report information confirming one’s prior beliefs has distorted conventional methods for creating guidance based on the best available evidence. Confirmation bias warps our thinking and can have a significant effect on the proper functioning of society by misrepresenting the evidence.

GP consultation data, on the other hand, has shown that simply encouraging social distancing and handwashing reduced transmission of acute respiratory tract infections by about half pre-lockdown. Handwashing and encouragement are what we need, not formalised rules. This means trying to keep a distance from each other where possible and avoiding spending time indoors in crowded places. Much of the evidence informing policy in this outbreak is poor quality; let us hope that evidence-informed decisionmaking will at some point resume.