Big thank you to @EricCarroll for pointing out this new WHO document on SARS-CoV-2 transmission.
This document is pretty complex, in-depth, dense, and I still expect it to evolve as we learn along the way. They have some of the correct people to be working on this, for once. Hello Lidia Morawska signing off on it at the beginning of the forward.
First, a tldr. If you don't care about how it came to be, or the science, and just want to know the outcome, here it is:
https://partnersplatform.who.int/tools/aria
Go to the calculator, enter your data, and come out with a probability of infection in a given situation along with the number of expected secondary infections from that interaction.
Here's the document itself if you want to follow along:
https://iris.who.int/bitstream/handle/10665/376346/9789240090576-eng.pdf?sequence=1&isAllowed=y
Disclaimer - This is evolving science.
I'm going to split this up in a thread, because I took a lot of notes of what stood out to me on a first read, and I hope to come back to it, and use it as a general reference moving forward.
I was playing around with this calculator and so I put in a scenario that, pre-pandemic, was fairly common in my old workplace. We had ~12 people and there were times where we had to sit in a conference room from 8-5, with a lunch break for some training or certification.
The results were pretty interesting, in my opinion. I put in the size of the room, and that 1 person would be infected. As would be typical these days, I said zero masks.
It came up with a 45% chance that I would be infected were I in that room if I had short-range interactions with the infected person. 13% with long-range interactions.
You further get a graph showing Mean Concentration of Infectious Respiratory Particles(IRP) vs. Time of Day(aka, how long you've been in the room.)
The next graph it shows you is the Probability of Infection vs Viral Load. This is one of the biggest variables discussed above. While the estimate was 45%, interestingly based on how infectious the person who was COVID positive actually is, the 75th percentile is a 94% probability, for instance.
In this scenario at the end of the 8-5 work day you end up with an estimated 5 new cases, and 6 people who were exposed but not infected on average.
It then goes through the ways that you can reduce your risk, including wearing a well fitting mask, reduce interactions with people, and ventilating the space better.
Finally, it gives you the hierarchy of controls from most effective to least effective:
Elimination
Substitution
Engineering Controls
Administrative Controls
PPE
Giving you one last hint that maybe just you masking isn't the most effective thing, overall.
@BE This is great! Thank you for posting it! I just started plugging in average values for a school classroom. As you observed, the number of air changes per hour has only the smallest effect on the numbers. 1 goes in sick this delivers 6 new infections all because of the short range interactions….
@BE And that is for a 1 hour class with everyone unmasked. With 6-8 classes per day, plus the bus, and the halls, that one infection is going to generate some 50-100 per day, with the finest of air cleaning. Ouch!
@BE "How low can you reasonably go given this room's scenario? Respirators on everyone, mechanical ventilation(no open windows, because there weren't any in this room), and HEPA filtration. Now we're down all the way to less than 1%. Period."
This would be close to, say, a hospital operating theater, albeit the patient likely would be unmasked and on a ventilator. IIRC, COVID infections in these sorts of clinical settings aren't much of a concern with the expected precautions.
I took the same overall data, but added respirators to everyone. Now it's a 1.3% chance of personal infection. Only with the infectious person being 95th percentile did the chances get much above 1%, to 8%.
If I take the masks back off, and increase the room ventilation from 3 ACH to 6 ACH the estimate is essentially the same as the original 45% for short-range interactions, while the long-range interactions only went down to 11%.
Bumping that up to 12 ACH leaves your short-range still at 45%, but now your long-range is down to 8%.
What did we just learn? Their model shows that the amount of air changes per hour doesn't matter if you are face-to-face with someone who's got COVID. Makes sense, right? But people who only have long-range interactions fare better the more air changes.
So then I tried putting in what would be an average CADR for HEPA filtration in the room. Guess what? Still 45% if you're going to be face-to-face. But, now we're all the way down to 6% for people only having long-range interactions.
If I bump that CADR up to twice what's usually recommended for a room that size, the long-range interactions percent now comes down to just 5%.
Their model clearly shows that masks are effective given the data available. It's, in fact, the only way you're going to seriously reduce your short-range interaction risk.
But, ventilation and filtration absolutely have a fairly dramatic effect on long-range interactions according to this model.
How low can you reasonably go given this room's scenario? Respirators on everyone, mechanical ventilation(no open windows, because there weren't any in this room), and HEPA filtration. Now we're down all the way to less than 1%. Period.
0.03 expected new cases amongst the 12 people in the room all day.
Stay safe out there!