* For the purposes of this note, social distancing means the instructions we are being given to stay one or two metres away from any person who is not from our own household at all times.
* This is arguably the measure that has had the most impact on lives and livelihoods and will have the most impact on any chance of returning to normal.
* The World Health Organization has changed its advice from “Avoid close contact with anyone who has fever and cough” (start of February) to “Maintain at least 1 metre (3 feet) distance between yourself and anyone who is coughing or sneezing” (end of February) to “Maintain at least 1 metre (3 feet) distance between yourself and others” (end of April). Different countries have imposed different rules using different distances of between one and two meters.
* There were two systematic reviews available when the WHO introduced its one meter advice. One review didn’t examine this practice (and was about influenza); the other did examine it but found no evidence for “spatial separation of at least one meter” (for respiratory viruses generally).
* Another systematic review was published in May 2020, but this did not examine spatial separation and was about influenza.
* The Lancet published a paper on June 1st which examined physical distancing for coronaviruses (along with masks and eye protection). The paper claimed that there was approximately a 1.3% chance of contracting a virus when 2 metres from an infected person, but halving this gap raised the risk to only 2.6%.
* It is important to note that no randomised controlled trials on spatial separation of at least one meter have ever been undertaken. None of the systematic reviews have been able to include any ’cause and effect’ studies, therefore.
* The Lancet paper contained nine studies relevant to social distancing among the general public (once studies with high-protection surgical masks and/or in healthcare settings had been removed). One of these examined the transmission of SARS on flights (from the 2002-3 outbreak). The other eight examined the transmission of MERS, SARS or COVID-19 from infected people to close contacts including family and/or household members. None of them examined social distancing, as we are being told to do today.
* In every one of these studies, the major, or only, risk factor was being in intimate contact with an infected person – being a spouse, being the carer for that person, embracing that person, sharing a bedroom with that person and so on.
* The Lancet paper has assumed that these household studies have provided evidence that people 0-1 meters away from an infected person have a greater risk of catching a coronavirus than someone more than 0-1 meters away. What these household studies really found was that the spouse/lover/carer had a greater risk than an extended family member (not in the same household), or a co-worker, or a random person in a waiting room.
* Politicians say that they are following the science. There is no science on social distancing; not as it pertains to what they are instructing us to do.
This week we’re looking at social distancing. The term social distancing has been used to capture a number of different behaviours, which are intended to reduce virus transmission. For example, working from home, or working different shifts, have been described as social distancing measures in research papers. The World Health Organization speech on March 13th gave “cancelling sporting events” as an example of social distancing (Ref 1).
For the purposes of this note, social distancing means the instructions we are being given to stay one or two metres away from any person who is not from our own household at all times (whether in our own garden, at work, or queuing outside or inside a shop).
For many people, this is the most draconian measure that has been adopted as a reaction to SARS-CoV-2. It is the one that has condemned people who live alone to be denied a single human touch for over 11 weeks (and ongoing in the UK). It is the one that has stopped us from hugging our loved ones, even as they are dying. It’s the one that has stopped low/no risk children from playing. It’s the one that will most impede any return to normality if we continue to adhere to it. It’s the one that will kill the travel, restaurant, sporting, and hospitality industries, if they are not dead already. It’s the one that will take away the livelihoods of so many people who work in those industries, if they are not unemployed already. It’s the one that makes shopping an endurance test, rather than a pleasurable experience. It’s the one that has broken societal interaction, as strangers treat each other as if we have the plague.
I looked at the evidence for social distancing when we were first instructed to do this, and I have kept up to date with any emerging evidence. On June 1st, 2020, a systematic review and meta-analysis was published in the Lancet, which is highly relevant to this topic and will be seen as the final word on the subject, but should it be? (Ref 2)
I think that “social distancing” will be one of the Oxford English Dictionary “words of the year” for 2020 (they do allow two words). Politicians and public health advisors have been telling us all along that they have been following the science. But when it comes to this new term, have they really been following the science?
World and country advice
At the start of February 2020, the World Health Organization (WHO) advice was “Avoid close contact with anyone who has fever and cough” (Ref 3). By the end of February, the WHO was advising people: “Maintain at least 1 metre (3 feet) distance between yourself and anyone who is coughing or sneezing” (Ref 4). This remained the advice until that page was updated on April 29th when it became: “Maintain at least 1 metre (3 feet) distance between yourself and others” (Ref 5). Notice the change from fever and cough, to coughing and sneezing, and then to anyone, but not until the end of April.
Since March 23rd, UK citizens have been instructed to stay two meters apart. The US still works in imperial measurements, so it’s been six feet for the US (approximately two meters). Australia, Germany, and the Netherlands went for one and a half metres. Austria, France, Italy, Sweden, and Singapore opted for one meter. As soon as you see different numbers, you know that the advice can’t be evidence based.
The evidence for social distancing at the time it was issued
When I first starting researching this topic, I found remarkably little evidence even claiming to examine social distancing. When I reviewed the literature for wearing masks (Ref 6), the major piece of evidence was a Cochrane systematic review called “Physical interventions to interrupt or reduce the spread of respiratory viruses” (Ref 7).
This (2011) paper also reviewed evidence on social distancing available at that time. It defined social distancing as “spatial separation of at least one metre between those infected and those non-infected” and so it was looking at our definition. There was not enough evidence for the Cochrane report to do a meta-analysis (that’s a technique to pool available data together). The extremely limited data were examined individually, therefore. The conclusion was: “There is insufficient evidence to support screening at entry ports and social distancing as a method to reduce spread during epidemics.”
We got “screening at entry ports” for free there. I hope that the UK government, which is introducing an air travel quarantine today (June 8th – more than four months after the first recorded UK case of COVID-19) takes note of this and abandons this non-evidence based policy before it begins.
A search of the literature for evidence since 2011 produced two systematic reviews – one by Ahmed et al (2018) (Ref 8) and one from Fong et al (2020) (Ref 9). The latter was published in May 2020 and thus would not have been available to inform social distancing instructions at the time they were introduced.
The Ahmed et al paper was called “Effectiveness of workplace social distancing measures in reducing influenza transmission: a systematic review.” This review only examined social distancing in the workplace – not queuing outside the grocery store – and it studied influenza, rather than COVID-19, or any coronavirus.
Ahmed et al found 12 modelling studies and 3 population studies to be included in their systematic review. Hence, we have no randomised controlled trials (RCT) to examine. Only these can tell us if social distancing works or not. We could stop there and conclude that there was no RCT (cause and effect) evidence as of 2018, but, for completeness, let’s take a brief look at the paper.
I started with the three population studies. One of them was a multi-factorial intervention and social distancing wasn’t even one of the interventions. (This was conducted in a military base and hand washing, temperature checking and having different mealtimes were the main interventions.) The other two studies were about working from home – not staying one or two meters apart at work. Hence none of the population studies examined our definition of social distancing.
I then looked at the 12 models. Models are only as good as their assumptions, so if a model concludes that social distancing is of benefit, the assumptions must have made it so. Notwithstanding this, none of the 12 models examined what we’re calling social distancing. They variously modelled possible workplace interventions such as working from home; working different shifts; some employees working during the week and others at weekends; a number of them modelled scenarios for employees who had received anti-viral treatment and vaccines vs those who hadn’t.
Nothing in Ahmed et al thus provides evidence for or against social distancing.
The Fong et al study was called “Nonpharmaceutical Measures for Pandemic Influenza in Nonhealthcare Settings—Social Distancing Measures.” Again, it studied influenza, not a coronavirus. Examination of the paper revealed that it had looked at many different behaviours under a banner of “social distancing”, but none of them matched what we’re looking at.
Social distancing in this paper covered: isolating sick people; contact tracing; quarantining exposed people; school closures; workplace closures; workplace measures (e.g. work from home); and avoiding crowding (cancel events, manage public transport etc). The paper concluded: “our review found some evidence from observational and simulation studies to support the effectiveness of social distancing measures during influenza pandemics.” I would not be surprised that isolating sick people and quarantining exposed people would help to contain the spread of flu, but this paper didn’t examine social distancing (as per the focus of this note).
Nothing in Fong et al thus provides evidence for or against social distancing.
The Lancet paper
Interestingly, The Lancet paper was commissioned and funded by the World Health Organization (WHO). Perhaps with the goal of finding retrospective evidence for advice from months earlier.
The Lancet paper looked at a number of different measures to prevent person-to-person transmission. It examined what it called physical distancing (ostensibly our definition of social distancing), wearing face masks and wearing eye protection. I’m just going to look at the physical distancing data in the paper.
This was a systematic review (search for all data) and a meta-analysis (pool that data together), so it was high quality research. The review only looked at studies of coronaviruses (MERS, SARS and COVID-19) and not influenza and so it was far more relevant. The researchers noted that there are no randomised controlled trials to review, so we still have nothing that can determine cause and effect. This is important.
Analysis of 9 studies, including 7,782 participants, found that there was approximately a 1.3% chance of contracting the virus when 2 metres from an infected person, but halving this gap raised the risk to only 2.6%. The paper reported that with a 1 meter distance between people, the virus would spread to fewer than 3 in 100 people, vs 13 in 100 without any social distancing at all. The authors noted that the certainty of the evidence was moderate. Even more importantly, the authors reported that none of the studies quantitatively evaluated whether distances of more than two metres were more effective. The researchers estimated the risks therefore (Ref 10).
The physical distance studies
We have no randomised controlled trials and we have no studies that actually evaluated distances beyond two meters. What evidence exactly did we have?
Figure 2 in The Lancet paper summarised the evidence for what they called physical distancing. It was a meta-analysis of 30 studies: 8 for MERS; 17 for SARS and 5 for COVID-19. I removed all studies (12) that examined people wearing an N95 respirator or similar. (This is a protective device designed to achieve a very close facial fit and very efficient filtration of airborne particles). I also removed all studies from a healthcare setting. We’re interested in the queue outside the grocery store and whether or not people can return to sports stadia.
This left nine studies: two studies (from Saudi Arabia) on MERS; four studies (from China, Canada, Hong Kong, and Vietnam) on SARS; and three studies (from the USA, China, and Taiwan) on COVID-19. We’re going to whiz through these studies – it’s important because you will soon see that social distancing has still not been tested:
1) Study 1 MERS (Van Kerkhove et al 2019) examined 828 residents in a women-only dormitory in Saudi Arabia (Ref 11). One woman was originally infected with MERS and 19 of the other women became infected in total. The study looked at transmission of the virus from the first case and concluded that the major risk factors were having direct contact with and/or sharing a bedroom with an infected woman. In the dormitory, 54 women came into direct contact with someone with the virus; 11 of these women contracted the virus. That means that 774 people did not come into direct contact with someone with the virus; 8 of these women contracted the virus. The Lancet paper reported that as a 95% risk reduction if you were more than zero meters away from an infected person (Note 12).
You may like to read that again, as the other studies continue in that way. Please also note the impact of the denominator (the bottom of two numbers – the 54 vs the 774). Direct contact included sharing a bedroom, kitchen, bathroom, meal, and transport. If there were 100 or 10,000 women in the dormitory, the number of direct contacts would likely have been similar, but the risk reduction for being further away is hugely impacted by the number of people assumed to have been further away. You may have no different risk if there are 10 or 10,000 people queuing outside a shop but – using this methodology – your risk will look substantially different.
2) Study 2 MERS (Arwady et al 2016) looked at 19 MERS cases among 79 relatives in the extended family in another Saudi Arabian study (Ref 13). This also found that sleeping in an infected person’s room was a risk factor while “Casual contact and simple proximity were not associated with transmission.” Using data from Table 3 in the Arwady et al paper, the Lancet paper reported this as 1 in 10 people were infected more than 1 meter away and 8 out of 20 were infected less than 1 meter away.
3) Study 3 SARS (Tuan et al 2007) looked at 212 household and close contacts of 45 SARS cases identified in Vietnam (Ref 14). Nine close contacts became infected. The study concluded that “Physically caring for a symptomatic laboratory-confirmed SARS case was the only independent risk factor for SARS transmission.” Table 2 in Tuan et al reported that 57 people had physically cared for a patient (6 of these had contracted the virus) and 123 had not physically cared for a patient (3 of these had contracted the virus). That’s where The Lancet paper got 3/123 were infected more than 0 meters away vs 6/57 less than 0 meters away.
4) Study 4 SARS (Rea et al 2007) was a Canadian study that used ‘track and trace’ style contact data to examine 8,662 exposures (including household exposures) in Toronto that resulted in 61 probable cases (Ref 15). Out of 647 of the closest contacts (e.g. caring for or embracing the patient) 41 contracted SARS. Out of 3,493 of the second level of contact (e.g. same waiting room in a medical centre for more than 30 minutes), 18 contracted SARS. The Lancet paper reported this as 18 out of 3,493 over 1m away got the virus vs 41 out of 647 within 1m. This is one of the more relevant studies but sleeping with and caring for a SARS patient vs sitting innocently in a waiting room not knowing a SARS patient is also waiting still doesn’t justify the advice that citizens are being given in garden centres, for example.
5) Study 5 SARS (Lau et al 2004) reviewed 1,214 SARS cases in Hong Kong and transmission from these cases among household members (Ref 16). It found that 8% of all household members got infected and spread occurred in 15% of all households. The major risk factors were being the main caregiver, sharing a room or a bed with the patient, frequency of dining with the patient and being coughed on by the patient. The Lancet paper reported this as 136 out of 1,124 people within 1m contracted the virus vs 39 out of 965 who were not that close.
6) Study 6 SARS (Olsen et al 2003) was a Chinese paper that reported on a flight with 1 person who had SARS and 119 other people on board (Ref 17). Lab-confirmed SARS was subsequently recorded in 16 other people on the flight and 2 others were given diagnoses of probable SARS. Of the 18 infections, 8 came from 23 people who were seated in the 3 rows in front of the SARS case; 10 came from the other 88 people not in these rows. The 6 crew and 2 pilots were not infected. The Lancet paper reported this as 11/35 people within 1.5m contracted the virus vs 9/84 people not within 1.5m. Apart from the fact the numbers don’t match, the flight was three hours long and in an extremely confined space. It doesn’t help the social distancing research question in this note.
As this is an important study for the aviation industry, the paper reported on two other flights. One carrying four people with SARS resulted in transmission to at most one other person and no illness was documented in passengers on the flight that carried one person who had pre-symptomatic SARS.
7) Study 7 COVID-19 (Burke et al 2020 – pre-print, not yet peer-reviewed) (Ref 18). This is the first study on COVID-19 and so is of particular importance. This paper reviewed nine early travel-related cases of COVID-19 in the US. Track and trace style monitoring revealed that 338 people were exposed to these 9 cases, leading to 2 transmissions (that’s a low infection rate for close contacts). Both people who contracted the virus were spouses. The Lancet paper reported this as 0 out of 13 people more than zero meters away caught the virus, while 2/2 people less than zero meters away did. Of the 338 close contacts, 15 were household contacts. The researchers appear to have assumed that none of the remaining 13 household contacts were spouses and none got within zero meters of the infected person. The researchers have also assumed that none of the other 323 close contacts were direct contact, despite 258 of them being among healthcare personal and/or in healthcare settings.
8) Study 8 COVID-19 (Liu et al 2020) is in Chinese, but I found an English abstract (Ref 19). It included data on 20 close household contacts exposed to the virus, of whom 2 were infected. The Lancet paper reported this as 2 out of 3 household members within 1m contracted the virus vs 0 out of 17 who were not that close. I don’t know how they know that none of the other household members got within 1m of an infected person.
9) Study 9 COVID-19 (Cheng et al 2020 – pre-print, not yet peer-reviewed) was another household transmission study – this time in Taiwan (Ref 20). The study involved tracing 1,043 people who had come into contact with 32 people with lab-confirmed COVID-19. Of the 1,043 contacts; 12 of these became infected (another low transmission rate). Among the 1,043 contacts, 36 were from the households of an infected person and 47 were from the families of an infected person. All 12 infections came from these household and family contacts – 7 from the household and 5 from the family. The Lancet paper reported this as 7 out of 36 people contracted the virus within 1m of an infected person while 5 out of 47 people infected were further away than 1m. This cannot be concluded, and it cannot be concluded that the other 960 contacts were more than 1m away.
That’s it. Thank you for bearing with me because that was important. The chance of returning to any semblance of normality relies on understanding the evidence for social distancing and there is simply no evidence that a one or two meter rule among members of the public has even been tested, let alone conclusive evidence to support it having been found.
Of the nine studies without high-protection surgical masks and outside healthcare settings, one was on a plane and the other eight involved examination of the transmission of MERS, SARS or COVID-19 to close contacts including family and/or household members. In every study, the major, or only, risk factor was being in intimate contact with an infected person – being a spouse, being the carer for that person, embracing that person, sharing a bedroom with that person and so on. The most relevant public setting study is Rea et al (SARS) and this reported that 18 out of 3,493 people at the second highest level of contact (e.g. being in the same waiting room for more than 30 minutes) contracted SARS. The third level of contact, perhaps a passing greeting or being in the same elevator as an infected person, saw 2 cases among 1,559 people.
The Lancet paper has assumed that these household studies have provided evidence that people 0-1 meters away from someone else have a greater risk than someone more than 0-1 meters away. What these household studies really found was that the spouse/lover/carer had a greater risk than an extended family member or co-worker or a random person in a waiting room.
Ref 1: https://www.who.int/dg/speeches/detail/who-director-general-s-opening-remarks-at-the-mission-briefing-on-covid-19—13-march-2020
Ref 2: Chu et al. Physical distancing, face masks, and eye protection to prevent person-to-person transmission of SARS-CoV-2 and COVID-19: a systematic review and meta-analysis. The Lancet June 2020.
Ref 3: https://web.archive.org/web/20200201021512/
Ref 4: https://web.archive.org/web/20200229135608/
https://www.who.int/emergencies/diseases/novel-coronavirus-2019/advice-for-public (last updated 29th April)
Ref 6: https://www.zoeharcombe.com/2020/04/covid-19-do-masks-help/
Ref 7: Jefferson et al. Physical interventions to interrupt or reduce the spread of respiratory viruses. Cochrane Database Syst Rev. 2011.
Ref 8: Ahmed et al. Effectiveness of workplace social distancing measures in reducing influenza transmission: a systematic review. BMC Public Health 2018. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5907354/
Ref 9: Fong et al. Nonpharmaceutical Measures for Pandemic Influenza in Nonhealthcare Settings—Social Distancing Measures. CDC Policy Review. May 2020.
Ref 10: https://www.eurekalert.org/pub_releases/2020-06/tl-pss060120.php
Ref 11: Van Kerkhove et al 2019 https://wwwnc.cdc.gov/eid/article/25/10/19-0130_article
Note 12: 11 out of 54 is 20%; 8 out of 774 is 1% – that’s the 95% difference.
Ref 13: Arwady et al 2016 https://pubmed.ncbi.nlm.nih.gov/27191038/
Ref 14: Tuan et al 2007 https://pubmed.ncbi.nlm.nih.gov/16870029/
Ref 15: Rea et al 2007 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2870656/
Ref 16: Lau et al 2004 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3322902/
Ref 17: Olsen et al 2003 https://pubmed.ncbi.nlm.nih.gov/14681507/
Ref 18: Burke et al 2020 – pre-print, not yet peer-reviewed
Ref 19: Liu ZQ, Ye Y, Zhang H, Guohong X, Yang J, Wang JL. Analysis of the spatio-temporal characteristics and transmission path of COVID-19 cluster cases in Zhuhai. Trop Geogr 2020; published online March 12th 2020. English abstract found here:
Ref 20: Cheng et al 2020 – pre-print, not yet peer-reviewed https://www.medrxiv.org/content/10.1101/2020.03.18.20034561v1