4 ms·

17 minutes in, dismissing the R value that the UK scientific advisors calculated based on unpublished data, that we now know was correct: "you can't use epidemi

17 minutes in, dismissing the R value that the UK scientific advisors calculated based on unpublished data, that we now know was correct: "you can't use epidemiological data to prove a biological effect of an amino acid..." Yeah, that's a virology point of view all right. That something has to have a proven biological mechanism before it can be taken as fact, or even considered as a risk. That's bad public policy though. With his reasoning there would have been no reason to increase restrictions in the UK in Dec/Jan, but in fact the restrictions were extremely necessary and the variant caused tens of thousands of extra deaths.

native_samples 5y ago
The R numbers coming out of UK epidemiology are garbage, Racaniello is right about that. They aren't physically grounded in any way, they're just arbitrary fudge factors brute forced to make their equations plot graphs that look like reported government stats so far. The same model will routinely calculate totally different values of R for different areas of the same country at the same time, without variants or anything like that. Also the claims about variants being "more infectious" are - when you dig in - hopelessly confounded by seasonal effects that many epidemiological papers were at that time still ignoring.
- jsnell 5y ago
  The seasonal effects are trivially controlled for by comparing the growth rate of the variant to the growth rate of the wildtype at the same time. And that's exactly how the original analysis was done, because the people doing it were not incompetent. For your seasonal effect explanation to make any sense, the effect would have to have only applied to Alpha but not to baseline Covid at the same time. And then it's not a seasonal effect, is it? It's an actual difference in the behavior of the two variants. The same goes for any other similar confounder that you try to manufacture.
  - native_samples 5y ago
    Not the analysis I saw, where Delta's growth rate was compared to Alpha in the same time window, not the start of alpha's own growth period. Because yes, they are totally incompetent. You can look at graphs of the changing proportions to see visually that that's not much difference in how fast they took over.

native_samples 5y ago

The R numbers coming out of UK epidemiology are garbage, Racaniello is right about that. They aren't physically grounded in any way, they're just arbitrary fudge factors brute forced to make their equations plot graphs that look like reported government stats so far. The same model will routinely calculate totally different values of R for different areas of the same country at the same time, without variants or anything like that. Also the claims about variants being "more infectious" are - when you dig in - hopelessly confounded by seasonal effects that many epidemiological papers were at that time still ignoring.

jsnell 5y ago
The seasonal effects are trivially controlled for by comparing the growth rate of the variant to the growth rate of the wildtype at the same time. And that's exactly how the original analysis was done, because the people doing it were not incompetent. For your seasonal effect explanation to make any sense, the effect would have to have only applied to Alpha but not to baseline Covid at the same time. And then it's not a seasonal effect, is it? It's an actual difference in the behavior of the two variants. The same goes for any other similar confounder that you try to manufacture.
- native_samples 5y ago
  Not the analysis I saw, where Delta's growth rate was compared to Alpha in the same time window, not the start of alpha's own growth period. Because yes, they are totally incompetent. You can look at graphs of the changing proportions to see visually that that's not much difference in how fast they took over.