Read on Twitter
New preprint on how temperature and humidity affect the infectiousness of SARS-CoV-2 and other viruses.
Key finding: the virus remains infectious longer at low temperatures and extreme relative humidities. https://doi.org/10.1101/2020.10.16.341883
Key finding: the virus remains infectious longer at low temperatures and extreme relative humidities. https://doi.org/10.1101/2020.10.16.341883
What do I mean by "extreme" relative humidities (RH)? There's a U-shaped pattern: the virus does well at low RH and high RH, and less well at intermediate RH
We didn't just want to report this; we wanted to understand it. We took mechanistic principles identified by @linseymarr, @Lakdawala_Lab, and others ( https://doi.org/10.1098/rsif.2018.0298) and encoded them into a simple mathematical model that we could fit to our data
Details of the model are in the paper, but the key *idea* is the following: virus inactivation follows the principles of chemical reactions
Reactions proceed faster in hotter solutions (since more collisions between particles get over the "energy barrier") and in more concentrated solutions (since collisions are more frequent, all else equal).
Ambient temperature sets the temperature of the solution in which a virion finds itself—no surprise there!
But ambient relative humidity sets the concentration. Why? Water evaporates from the solution until a (quasi-)equilibrium with the ambient environment is reached. Lower RH --> more evaporation...
...but below a certain threshold relative humidity, the efflorescence relative humidity (ERH), an electrolyte solution like cell culture medium or human respiratory fluid can crystalize. Concentrated solutions are favorable for reactions. Crystals may be less so
So we expect that U-shaped pattern. Above the ERH, more humidity --> better virus survival. Below the ERH, virus survival should rise again. And that's exactly what we see!
A simple (4 main parameters) mathematical model encoding these ideas fits our data extraordinarily well. Again, this is not a regression; it's our mechanistic model!
And using that mechanistic model, we can extrapolate to unobserved conditions and predict observations from other human coronaviruses—with success!
Lots more details in the paper, but some key epidemiological and basic science takeaways:
Epi takeaways: outdoors, you've got sunlight and ventilation to help you, even if it's cold.
In typical indoor settings, climate control keeps things cool and dry. Add in poor ventilation and limited UV, and virus does pretty well
In typical indoor settings, climate control keeps things cool and dry. Add in poor ventilation and limited UV, and virus does pretty well
And in specialized indoor settings like a food processing plant, it should do even better https://www.cidrap.umn.edu/news-perspective/2020/04/us-food-processing-plants-become-covid-19-hot-spots
Further underscores what we already knew: workers need better PPE, better ventilation, testing, and paid sick leave
Basic science takeaway: simple chemistry can explain a lot about virus environmental stability, and parameters from SARS-CoV-2 predict other human coronaviruses well.
Conclusion: shared simple mechanisms may affect virus environmental stability for a number of enveloped viruses
Conclusion: shared simple mechanisms may affect virus environmental stability for a number of enveloped viruses
That's it for now. Thanks very much to anyone who made it all the way through the thread, and hope you'll check out the paper!
P.S. As always, our code and data are open for anyone who wants to play around https://zenodo.org/record/4093265
P.S. As always, our code and data are open for anyone who wants to play around https://zenodo.org/record/4093265
Addendum: speaking of open code and data, we used a number of wonderful free and open source software tools, including @mcmc_stan, #Rstats, the #tidyverse (esp #ggplot2), and @mjskay's wonderful tidybayes and ggdist packages. Grateful to the communities that maintain them!
Another footnote, since people are wondering about humidities: the ERH of 45% that we use in the figures is a best guess for NaCl dominated electrolyte solutions. But practical ERHs can vary a bit. So don't assume 45% = efflorescence and 46% = solution
So what does this tell us to do? By itself, not much. A good rule of thumb is that no single study should make policy, peer reviewed or not. But in light of our work and the work of others, I think we can say the following:
Ventilate! humidify! https://twitter.com/dylanhmorris/status/1321566413291724802
Ventilate! humidify! https://twitter.com/dylanhmorris/status/1321566413291724802
Thread FAQ:
Q: This studies the virus on plastic. What about aerosols?
A: We're working on that! But in the mean time, inert surface data can be qualitatively informative RE: aerosols. As we say in the paper:
Q: This studies the virus on plastic. What about aerosols?
A: We're working on that! But in the mean time, inert surface data can be qualitatively informative RE: aerosols. As we say in the paper:
Q: Very high RH might be good for virus stability, but doesn't it also cause particles carrying the virus to settle out of the air faster?
A: Another effect we're studying now! I'm comfortable saying "dry air's bad for us / good for the virus" but don't comment on v wet air 1/2
A: Another effect we're studying now! I'm comfortable saying "dry air's bad for us / good for the virus" but don't comment on v wet air 1/2
Because with dry air, known effects (inactivation, settling, effects on our antiviral defenses) point in the same direction. Since they potentially don't for wet air, we need careful modeling 2/2
