Thanks for your transparency and for updating this report; I think it is tremendously valuable. I have only skimmed so far, but I will hopefully read through it completely soon.
I also tried to do a back-of-the-envelope calculation of the cost-effectiveness of 254 nm upper-room GUV per eACH a while ago. I estimated higher numbers in my model, more on the order of $100/eACH/year (For context: the report estimates ~$14/eACH/year). Note that the model is pretty rough, but I’ll post it anyway in case it proves useful for others.
I think this difference comes mainly from having a much lower point estimate of eACH that is realistically achievable. In general, I am rather skeptical of eACH estimates for GUV that go into the hundreds.
As far as I understand, the way eACH are calculated depends on the specific pathogen in question. Pathogens vary widely in their susceptibility to GUV, and coronaviruses are unusually susceptible IIRC. Since many of these eACH estimates were calculated based on measurements with coronaviruses, this inflates the values, and they’d presumably be significantly lower for other pathogens.
Another point is that AFAICT, the higher eACH estimates mostly (all?) come from computational models and not real-world measurements. I assume that real-world environments are messy and will be less ideal for achieving very high eACH rates.
I think I had another reason for my skepticism, but unfortunately, I can’t recall it right now—will update this comment should I remember.
Thanks so much for the kind feedback and comparison calculation! Your skepticism about the eACH estimates is warranted—I was unaware that coronaviruses were unusually susceptible (compared with other viruses, you mean?); the estimates we saw were all based on either SARS-CoV-2 or tuberculosis (also quite susceptible). It’s useful to know how other people are approaching this question, and ultimately the problem calls for much more extensive real-world observations.
Yea, the way I recall it is that coronaviruses are more susceptible than other viruses. I first tried to recheck this in Appendix B of Kowalski (2009), but the values provided there vary extremely widely. I suspect the experimental quality varies a lot between those estimates, and coronaviruses were, of course, of much less interest back then.
[...]
IMO, the easiest way to read these values is the D90 (J/m^2) value, the dose required to inactivate 90% of viruses in a sample. This is equivalent to speaking about a “1 log reduction”.
Blatchley et al. (2022) have more recent data and provide better evidence for coronaviruses being especially susceptible:
“At 254 nm approximately 1 log10 reduction of coronaviruses is achieved for each 2 mJ/cm2 delivered UV-C fluence (dose). For comparison, other human pathogenic viruses, such as poliovirus and rotavirus require about 4–5 times that amount (i.e., 8–10 mJ/cm2) for each log10 unit reduction (Masjoudi et al., Citation2021).
UV222 irradiation is at least as effective as UV254 irradiation for inactivation of viruses, with approximately 1 log10 reduction of coronaviruses achieved for each 1 mJ/cm2 of delivered UV-C fluence or less. In other words, irradiation at 222 nm provides roughly twice the rate of inactivation as observed at 254 nm.”
On the other hand, when considering catastrophic biorisk, we mostly don’t care about, e.g., poliovirus, but other viruses with more pandemic potential, such as Influenza. So the more productive comparison could be restricted to these high-profile pathogens. I haven’t taken the time to compare all their UV susceptibility values. If they turn out to be similar to those of coronaviruses, my skepticism of high eACH estimates might not be as valid.
Thanks for your transparency and for updating this report; I think it is tremendously valuable. I have only skimmed so far, but I will hopefully read through it completely soon.
I also tried to do a back-of-the-envelope calculation of the cost-effectiveness of 254 nm upper-room GUV per eACH a while ago. I estimated higher numbers in my model, more on the order of $100/eACH/year (For context: the report estimates ~$14/eACH/year). Note that the model is pretty rough, but I’ll post it anyway in case it proves useful for others.
I think this difference comes mainly from having a much lower point estimate of eACH that is realistically achievable. In general, I am rather skeptical of eACH estimates for GUV that go into the hundreds.
As far as I understand, the way eACH are calculated depends on the specific pathogen in question. Pathogens vary widely in their susceptibility to GUV, and coronaviruses are unusually susceptible IIRC. Since many of these eACH estimates were calculated based on measurements with coronaviruses, this inflates the values, and they’d presumably be significantly lower for other pathogens.
Another point is that AFAICT, the higher eACH estimates mostly (all?) come from computational models and not real-world measurements. I assume that real-world environments are messy and will be less ideal for achieving very high eACH rates.
I think I had another reason for my skepticism, but unfortunately, I can’t recall it right now—will update this comment should I remember.
Thanks so much for the kind feedback and comparison calculation! Your skepticism about the eACH estimates is warranted—I was unaware that coronaviruses were unusually susceptible (compared with other viruses, you mean?); the estimates we saw were all based on either SARS-CoV-2 or tuberculosis (also quite susceptible). It’s useful to know how other people are approaching this question, and ultimately the problem calls for much more extensive real-world observations.
Yea, the way I recall it is that coronaviruses are more susceptible than other viruses. I first tried to recheck this in Appendix B of Kowalski (2009), but the values provided there vary extremely widely. I suspect the experimental quality varies a lot between those estimates, and coronaviruses were, of course, of much less interest back then.
[...]
IMO, the easiest way to read these values is the D90 (J/m^2) value, the dose required to inactivate 90% of viruses in a sample. This is equivalent to speaking about a “1 log reduction”.
Blatchley et al. (2022) have more recent data and provide better evidence for coronaviruses being especially susceptible:
“At 254 nm approximately 1 log10 reduction of coronaviruses is achieved for each 2 mJ/cm2 delivered UV-C fluence (dose). For comparison, other human pathogenic viruses, such as poliovirus and rotavirus require about 4–5 times that amount (i.e., 8–10 mJ/cm2) for each log10 unit reduction (Masjoudi et al., Citation2021).
UV222 irradiation is at least as effective as UV254 irradiation for inactivation of viruses, with approximately 1 log10 reduction of coronaviruses achieved for each 1 mJ/cm2 of delivered UV-C fluence or less. In other words, irradiation at 222 nm provides roughly twice the rate of inactivation as observed at 254 nm.”
On the other hand, when considering catastrophic biorisk, we mostly don’t care about, e.g., poliovirus, but other viruses with more pandemic potential, such as Influenza. So the more productive comparison could be restricted to these high-profile pathogens. I haven’t taken the time to compare all their UV susceptibility values. If they turn out to be similar to those of coronaviruses, my skepticism of high eACH estimates might not be as valid.