Was the COVID-19 virus moving around the world in late 2019? And did we actually already have that information, but lacked the collective framework to make lay epidemiology coherent?
In terms of seeding and propagating the COVID-19 virus within the English population, we will probably never know precisely how disastrous the mass public gatherings of early March 2020 actually were. That those events (racing at Cheltenham, football at Anfield, rugby at Twickenham etc) gave a significant boost to the epidemic is now widely accepted and they will, no doubt, be closely examined in the series of public inquiries that will inevitably follow the United Kingdom COVID-19 disaster.
One interesting aspect of this regrettable chapter of the whole sorry saga is that, to function as fully-fledged ‘super-spreader’ events, quite a large number of the punters and sports fans must have been carrying the virus (and capable of passing it on) when they turned up. That being the case it is only logical to assume that they had all previously caught it from someone they’d been in close proximity to in the preceding days or weeks.
It is possible (but rather unlikely) that all of those ‘one-step-before-Cheltenham’ cases had just returned from a ski resort, or were fresh back from Lombardy, or recently got off a plane from Wuhan or Teheran. Or, of course, they could have picked it up via so-called “community transmission”. In other words from a person resident in this country without any obvious link to a foreign source. Officially, though, when the horses were galloping at Cheltenham, community transmission had only just started – the first case having been identified in Surrey in the last days of February.
That timeline raises a couple of important questions. If lots of transmission happened at the early March mass events, just how long had the SARS-CoV-2 virus been circulating in England? And if the Surrey case was actually quite late to the party and many others had already been infected, wouldn’t those people have known that they had it? The answer to the first question might well be “… a lot earlier than you think” and the answer to the second could well be “they did – they just didn’t have a name for it”.
Since the epidemic started, I have been struck by how many people I have spoken to think that they or someone they know actually had the illness we now call COVID-19 several weeks before the first cases in England hit the news in late January 2020. So, could it be the case that Lay Epidemiology (the normal cultural practice of noticing and discussing illness, death and their surrounding circumstances) was actually picking up this slightly unusual illness long before its scientific/professional counterpart?
Typical of the Lay Epidemiology accounts are these two. The first is an extract from the ‘Slugger O’Toole’ blog, the second is a description of my own experience :-
“My mum was badly ill back at the start of February. The illness had all the symptoms of Covid 19, the fever, the bad chest etc. Thankfully she pulled through, but it has taken her weeks to get over it. Many of us will have similar stories of unusual seasonal illness that we are now wondering, could it have been Covid 19?”
“On the day after Boxing Day I felt a bit shivery and, in the course of a couple of hours that afternoon, I lost all sense of taste and smell. I assumed that New Year would be a bit ruined by my condition but by the following day the shivery feeling had subsided and by the next day taste and smell quickly began to return. I had barely ‘felt ill’ at all. A few days later, in the first days of January, I very suddenly developed a dry tickly cough. It was so sudden that I thought I must have breathed in a dust particle or an insect. But the cough continued (as a minor but persistent irritant rather than a great big phlegmy hauk) until the end of January.”
During the early days of lockdown I spent some time musing over whether the ‘Lay Epidemiology’ accounts of very early suspected COVID-19 (including my own) were all about some kind of imaginary hindsight and/or a fervent desire to have anti-body protection. On balance I felt that this wasn’t the case, especially as the epidemiological situation in England in late March seemed to fit better with a narrative of early, undetected circulation than with the ‘official version’ of the Brighton super-spreader and a couple of Chinese visitors to York.
The idea that the virus was circulating in Europe during the latter months of 2019 has now begun to exit the ‘cookie conspiracy theory’ world and move towards the mainstream. Several pieces of investigation around the world now appear to support the unofficial version. A French team who analysed samples from past cases of apparently influenza-linked pneumonia found that a patient (with no apparent links to China or any other known hot-spot) had died with COVID-19 in a hospital in Paris on December 27th 2019. This of course indicates that the virus had been in the area from mid December, if not before. The issue is also attracting attention in the USA, where strangely early cases in California are coming to light. In the UK the existence of a “death spike” in the third quarter of 2019 has not yet been either confirmed or formally ruled out.
Perhaps the strongest indication that the story about the virus emerging in the Wuhan wet market in December is not true is to be found in a recent paper published by the scientific journal called Infection, Genetics and Evolution. This reports that “Phylogenetic estimates support that the COVID-2 pandemic started sometimes around 6 October 2019 – 11 December 2019.” One of the pieces of research reported in the article has May 2nd 2019 as the earliest date that would still be consistent with the virus’s later genetic evolution.
If a general national or international tracking system (like the ZOE symptom reporting app for COVID-19) had been in existence during 2019 it may well have picked up the rather odd combination of symptoms that make COVID-19 visible. The ZOE app, for example, was able to identify the importance of the loss of smell and taste symptoms long before they were officially recognised and included in Public Health messages. Given that the public can now readily do things like app-based reporting, might the time have come for the basic processes of Lay Epidemiology to be formally included in future early warning systems?
Sonia Griffiths on Jun 8, 2020
I read an article yesterday that said as early as December 1st it could have been in England. It was written by a lady who had a mother with the illness.
Charlie Davison on Jun 8, 2020
Hi Sonia – thanks for your comment. If you have a link to that article, could you post it here? It sounds like an interesting account of a (nother?) possible early case in England.
Paul Chant on Jun 11, 2020
The idea of the early emergence of coronavirus raises the question why it didn’t become apparent earlier. The graphs in the BBC explanation of the R figure are consistent with a mean transmission time of 5 days for successive generations of infection spread. If, conservatively, we take an R value of 2 and a 90 day period from August then 1 case would multiply to a very noticeable 250k or so. Conversely, if the case number became just 1,000 (the noticeability threshold?) then R is/was about 1.4, which I suppose is possible though I think the figures generally quoted are higher. There remains the possibility that the virus mutated in humans. The nature of the autumn bulge in Wuhan hospital cases, if it indeed occurred, might elucidate this.
Another puzzle is the shape of the infection curves. Exponentiality has been drummed into us, and with good reason, but curves I’ve looked at, on Bing and elsewhere aren’t exponential. They have quite long nearly straight segments before bending down, from lockdown effects rather than herd immunity or mutation I suppose. The implication of a straight segment is that R=1 – why? The bendy bits aren’t very exponential either. I’ve looked at successive weeks on the Bing graph of UK cumulative cases from 13/02 (13 cases) to 8/04 (63,377). The increases week on week show ratios of 5.15, 5.85, 3.18, 2.98 and 1.61. Under-reporting in the early weeks may explain the early high ratios but nonetheless there seems to be a decline. One explanation might be that with multiple sources of spread due to multiple arrivals of travellers then we have a sum of exponentials, and this generally is not an exponential. Whether it would give the shape we have I’m not clear.
So having failed to clear up the past let’s look into the future as we emerge from lockdown. The R number, we’re told is betwen 0.6 and 0.9. If we reckon that the last 7 days of reported cases are active, i.e. infectious, there are about 11,000 spreaders in the UK. The number of infections will decline as an infinite series with multiplier R. The total of new cases is given by multiplying the current number by R/(1-R). For R=0.9 the total will be 99,000. To reduce the number of spreaders to less than 1,000 will take 23 generations of spread, 115 days, nearly 4 months. For R=0.6 the corresponding figures are 16,500, 5 generations, less than a month. Of course easing the lockdown in itself will lift the R number but successful track, trace and isolate will reduce. Let’s hope the above calculations are wrong or that track, trace and isolate will work really well. Personally I’m going to stay pretty well isolated.
If we take R=0.89 rather than 0.9 the figures become 89,000, a 10% drop, and 21 generations, not much different. It seems plausible that wearing face masks could easily make the difference between these 2 values of R and thus effect a significant difference in casualties. If so then the WHO seems to have come to this rather late.
I find it astonishing that Matt Hancock should say that the ‘virus is in retreat’, as if it is a besieging army which can be counted on to slink away as soon as it’s taking unacceptable losses. What doesn’t he understand about exponentials and the mindlessness of viruses?