Note: this article is very speculative. Its main goal is to attract attention to important recent research on influenza transmission and to evidence on Covid-19 spread modes, not to make confident conclusions. If you don’t like my speculations, just read the research cited.
Much of the modelling and public health response to the pandemic of Covid-19 has been based on the research on the spread of a disease which has been with us for much longer, the good old influenza. You would think that, given how long standing the research on influenza spread is, everything of importance would have been known about it by now. And you would be forgiven for that, given the absolute certainty with which the authorities of all kinds have been claiming to base their actions on crystal clear science. But the science of influenza transmission is far from clear-cut.
It is, by the way, often the case in medical and other sciences that macro models and public measures are based on outdated (shall we call it) “first-order science”. By “first-order-science” I mean actual empirical research on the ground, rather than exercises in curve-fitting that most modeling seems to involve.
Recall, for instance, how many times have you heard from doctors that salt is bad for your heart. Only recently have researchers actually bothered to assess this conventional wisdom empirically, and the evidence for it is weak at best, even though the debate continues.
Empirical research on influenza transmission
The key unknowns about how influenza spreads that are hugely are important for any response, are in fact the same as for Covid-19. Does it spread outside at all? In which spaces does it primarily spread? What is the primary means of transmission? Is it the so-called large respiratory droplets? Or aerosols? Or fomites (viral plaques on surfaces)? Do masks (especially surgical ones) significantly mitigate transmission? Why does spread seem to exhibit seasonality? And so on.
None of these crucial questions has been resolved for influenza. In fact, serious attempts to address them seem to have started only relatively recently. I cannot even attempt to study and summarize the whole literature but here are some of the most interesting glimpses of understanding that seem to have emerged.
The abstract of a 2013 meta-analysis of the empirical research on the spread of influenza by Killingley and Nguyen-Van-Tam opens with the following stunning admission, “Remarkably little is known definitively about the modes of influenza transmission. Thus, important health policy and infection control issues remain unresolved.”
What is also interesting about that meta-analysis, however, is that it seems to imply that the recent empirical research on influenza spread casts doubt on the common understanding that influenza spreads primarily through large respiratory droplets. The importance of another, aerosol-driven mode has increasingly been recognized in the literature (see here, here and here). It was discovered that virus-containing aerosol emission is possible even through innocent breathing.
- One of the most promising avenues in influenza spread research has been human challenge experiments. Those mean that experimenters deliberately infect part of the participants with flu, usually through intranasal inoculation. Then, they look at how many of the inoculated develop influenza-like disease (ILI), and how many of the remaining participants they will manage to infect in a confined space with controlled ventilation and other conditions.
- Intranasal inoculation seems to be the experimental super-powered equivalent of large respiratory droplets, as the latter are too large to reach the lower-respiratory tract (LRT). However, there is substantial experimental evidence that the viral dose needed to produce infection in the upper respiratory tract (URT) is much higher than for the LRT. Furthermore, only a small part of the aerosol particles have stayed in the nose. These findings led some researchers to hypothesize that the aerosol spread to the LRT could actually be the preferred route of infection for influenza, at least when it comes to aerosol transmission. An alternative fascinating conclusion could be that there are different forms of illness depending on the mode of transmission.
- Some studies have shown that influenza could spread through aerosol at a distance of up to 4m. However, as researchers note (see the aforementioned meta-analysis) this does not imply that aerosol transmission does not matter. In fact, there is just no way in such studies to distinguish between large-droplet and short-distance aerosol transmission.
- The aforementioned meta-analysis also casts doubt on the spread of influenza through surfaces.
- Nguyen-Van-Tam et al. (2019, Lancet preprint) summarize purportedly the largest known human influenza challenge study by the time of publication. With intranasal inoculation, they succeeded in infecting 80% of the target group, most of whom developed an ILI. However, astonishingly, all those infected people only managed to infect one person in the secondary-attack group. Despite the purposefully created low-humidity conditions thought to favor transmission. Importantly, the infected volunteers also shed very little virus-containing aerosol compared to measurements from more natural transmission settings.
After excluding the other potential explanations, the researchers zeroed in on two alternatives. The first one is that influenza needs really poorly ventilated environments to spread, and the experimental setting was too well ventilated. The second hypothesis is even more fascinating. What if influenza is primarily spread by a small subset of the infected with significant LRT involvement and aerosol production? Intranasal inoculation does not create LRT involvement, and could explain the poor spread observed.
The implications of either of the hypotheses at which influenza human challenge studies are hinting are dramatic. If poorly ventilated confined spaces are where most of the spread happens, then ventilation — not masks, quarantines, lockdowns, school closures, etc. — becomes the elephant in the room for the spread of respiratory disease.
If the primary spreaders are those with LRT involvement and significant aerosol production, then, first, trying to measure nasal viral loads (say, in symptomatic and asymptomatic patients) will be misleading at best. Could some people who are asymptomatic at a given point actually have significant LRT involvement and matter more than most symptomatic people?
So what about Covid-19?
What does influenza spread research imply for Covid-19? First, an important question may be whether that research is relevant to a different virus. I think it probably is. Both viruses are single-strand RNA viruses, causing often similar symptoms, to the point where both can lead to death through ARDS. Both may even cause much of the horrible damage through downregulating the same ACE2 receptors (here and here) that are crucial for maintaining vascular function and that only SARS-CoV-2 uses for cell entry. True, there are also significant differences. SARS-CoV-2 seems to be almost harmless for children, has a different approach to entering cells, has a longer incubation period, does not need to enter the cell’s nucleus, etc. But those differences do not seem to be very relevant to the physical mode of transmission.
The spread of Covid-19 may actually be hinting at very similar issues as influenza. There is now a mountain of evidence that it essentially does not spread outside, that it has trouble to spread even in restaurants and supermarkets. That the risk of spread via casual contacts is very low. Similar findings exist for influenza. German researchers studying the spread in the heavily-hit area of Gangelt also could not isolate any viable virus from surfaces suggesting the risk of spread through them is also extremely low.
The places and contexts where it does spread well, sometimes infecting hundreds of people in a short amount of time, seem to involved prolonged interaction in confined spaces, are cruise ships, war ships, nursing homes, churches, mosques, crammed dance halls, Lombardian osterias with congregations of old people, Alpine chalets, after-ski parties in crammed bars. And the most dominant mode of transmission is through banal family households.
And it is not just close interaction that seems to matter. Take the tragic gathering of French Catholics in Mulhouse. In a matter of several days, probably, just one infected participant seems to have spread it to hundreds of the around 2000 other participants who came and went as the event went on. Or take the church singing superspread in South Orange, New Jersey. Singing and loud speech have been known to significantly increase the exhalation of small particles from the respiratory tract, with the emission increasing by an order of magnitude for some people. So, is it plausible to invoke just close-contact spread here or is short-range aerosol spread staring us in the face?
Most fascinatingly, the fact that Covid-19 spreads like wildfire in nursing homes and on cruise ships may be hinting at the huge importance of people with LRT involvement because the average age of the participants is significantly elevated.
Covid-19 also has at least three forms of manifestation. The majority, perhaps the vast majority of those infected seem to develop no symptoms or very slight symptoms that do not even bother them. Some people, like Boris Johnson and Chris Cuomo, seem to develop a “moderate” form in which there is prolonged weakness and high fever, and some manageable breathing difficulty. Finally, a minority of cases have critical illness with ARDS and have to be intubated.
Could influenza spread research suggest that those with the latter two forms of Covid-19 have primarily LRT involvement and have been primarily infected through the aerosol mode? Could some seemingly asymptomatic spreaders have early LRT involvement and the capacity for aerosol superspread without knowing it? Could the severe, and thus, actually important form of Covid-19 be primarily spread through aerosols, and potentially, only peopel who themselves have a severe form? These questions, and others related to the mechanics of spread, are way more crucial for handling Covid-19 than modeling, and they are sadly receiving very little attention.