7 ms·

> and the small difference in R0 is crude thinking. Given the reproduction number appears as the base of an exponent, I would not consider the difference "smal

> and the small difference in R0 is crude thinking. Given the reproduction number appears as the base of an exponent, I would not consider the difference "small." If influenza has a typical R0 of about 2, reducing contacts by 50% would make its growth non-exponential. If covid has a typical R0 of about 2.5, doing the same would require a further 20% reduction of contacts, to 40% of baseline rather than 50%. Since population behaviour and legal restrictions have diminishing returns (eliminating casual contacts first at lower cost, then progressively higher-cost and more essential contacts), this could easily correspond to a large difference in effected (necessary) policy. > Second, R0 is an estimated value. There are big error bars on all of these estimates -- the overlap is far greater than the 0.5 you're using to claim causality. I have seen no credible evidence that typical (non-pandemic) influenza has an R0 greater than that of SARS-CoV-2. If you have a citation handy to the contrary, I'd be interested in seeing it. In the meantime, I tried to be generous in my estimation of influenza's R0; even your reference [2] places its central estimate at 1.68. > Third, rhinovirus has not gone away (estimated R0 < 2). Nor has adenovirus (estimated R0 ~2.3) [1,2] Rhinovirus and adenovirus both have mean-estimate R0 greater than that of influenza in your reference [2], so if policy is going to eliminate any of these disease from general circulation then influenza would be the first. Furthermore, some analysis does show limited evidence of a reduction of rhinovirus cases (https://www.nature.com/articles/d41586-020-03519-3 https://www.nature.com/articles/d41586-020-03519-3). A confounding factor here, however, is that we don't really have comprehensive like-against-like monitoring for rhinovirus in the way developed nations monitor for influenza. Adenoviruses as a broad category are also not limited to respiratory transmission, so air-focused restrictions would have less effect on these other routes. Finally, I reiterate that flu has a pronounced seasonal variation, whereas rhinovirus is somewhat less variable. A set of social responses only needs to prevent the flu season from starting to make it "disappear."

timr 5y ago
> Given the reproduction number appears as the base of an exponent, I would not consider the difference "small." The difference in measurement is small, relative to the error in measurement. Look at the error bars on the R0 estimates in that paper. Influenza runs from 1.06 to 3.4. You are trying to make micrometer inferences from a parameter where our best estimates are measured in libraries of congress.
- Majromax 5y ago
  I'm not a clinician or public health policymaker, so I'm operating on the balance of probabilities rather than a p=0.05 standard. If you consider the influenza estimate independent of equivalent estimates for covid, then there's a substantially greater than 50% chance that influenza's R0 is meaningfully smaller than that of the flu. If the estimates are partially dependent (i.e. methodology that gives a larger estimate for influenza would also give a larger estimate for covid), then that chance goes up further. Moreover, we also have evidence in the actual observations of epidemic dynamics. Over the southern hemisphere fall of 2020 (northern hemisphere spring), covid cases increased more rapidly than influenza cases in countries that were entering their flu seasons. That's compelling-to-me evidence that covid in fact has a larger basic reproductive number, whether that be due to a fundamental difference in the virus or just the effects of low baseline immunity. Now, your idea that the particular set of policy restrictions may be more effective against influenza than covid is very interesting. There's a basic horse sense to it, as well, since policymakers were largely operating from a playbook written for a flu pandemic. However, I do not think that "more effective" is a necessary component of the argument; "at least as effective" will do just as well given influenza's low baseline prevalence (out of season). Remember, my original point upthread was not to explain every bit of the relative epidemic curves, but instead to claim that we really ought not be surprised by the difference. I see little need to resort to specialized theories to explain the broad outline, even though I'm sure they will be helpful in the details.
  - smolder 5y ago
    Your 2nd paragraph seems to have a mistake: influenza vs flu?
  - timr 5y ago
    > we also have evidence in the actual observations of epidemic dynamics. Over the southern hemisphere fall of 2020 (northern hemisphere spring), covid cases increased more rapidly than influenza cases in countries that were entering their flu seasons. That's compelling-to-me evidence that covid in fact has a larger basic reproductive number Yes, we know that SARS-CoV2 increased, and influenza went away. That's trivially observed. We don't know why flu went away. To turn that observation around and say "Covid grew faster than flu, and therefore it has a higher R0, and therefore, the flu went away because {insert your favorite NPI} was just good enough to stop it" is begging the question. It could just as easily be any number of other factors that you're not accounting for. There's also a number of other faulty assumptions inherent in your logic. For starters: there's no particular reason, a priori, that a person couldn't have both flu and SARS-CoV2 in the same year.
samatman 5y ago
Small but important correction: nothing can make epidemiological spread non-exponential. Driving base reproduction below one additional infection per infected patient means the exponent is less than one, so the incidence of the disease rapidly decreases. This decrease is also exponential by definition: infection is simply an exponential process, and an exponential equation doesn't stop being what it is when the exponent is less than one. The steady state is exactly 1, which is like balancing a pencil on its point. It's plausible that two things happened here: the various local restrictions reduced the R0 of flu below 1, and the border closures kept mutations from traveling: and influenza is notoriously mutation-prone. These can combine to practically extirpate influenza. Unfortunately, the effect is temporary, and the next flu season is likely to be a real doozy. Which will be "fun" because our societies now have a hyperactive immune system, and there will be considerable political pressure to "do something" about the next flu wave. This must be resisted but I'm not optimistic.
- Majromax 5y ago
  > The steady state is exactly 1, which is like balancing a pencil on its point. At exactly one, it's indeterminate; you could see a subexponential growth or decay. With no exponential terms remaining, lower-order terms that are ordinarily hidden by exponential growth become the dominant observables. (Since epidemic spread is also random, you're much more likely to see the stochastic terms show up in the overall statistics in these cases. You really can be lucky or unlucky with Rt ~= 1, but there's no persistent luck with Rt >> 1.) This is also important before community spreading sets in, and it is the mechanism by which contact tracing can work. Rather than change the general susceptibility of the population to the disease (changing the effective reproduction number), contact tracing tries to operate at the level of individual cases in ways that cannot be scaled to uncontrolled community spreading (even if Rt is close to 1). > and the border closures kept mutations from traveling: and influenza is notoriously mutation-prone. Perhaps, but are influenza mutations an important effect within a single flu season? I know the determination of to-be-dominant strains has a great effect on the composition of each season's flu vaccine, but as far as I am aware we don't see declining flu vaccine efficacy over a season.
  - 5y ago

timr 5y ago

> Given the reproduction number appears as the base of an exponent, I would not consider the difference "small." The difference in measurement is small, relative to the error in measurement. Look at the error bars on the R0 estimates in that paper. Influenza runs from 1.06 to 3.4. You are trying to make micrometer inferences from a parameter where our best estimates are measured in libraries of congress.

Majromax 5y ago
I'm not a clinician or public health policymaker, so I'm operating on the balance of probabilities rather than a p=0.05 standard. If you consider the influenza estimate independent of equivalent estimates for covid, then there's a substantially greater than 50% chance that influenza's R0 is meaningfully smaller than that of the flu. If the estimates are partially dependent (i.e. methodology that gives a larger estimate for influenza would also give a larger estimate for covid), then that chance goes up further. Moreover, we also have evidence in the actual observations of epidemic dynamics. Over the southern hemisphere fall of 2020 (northern hemisphere spring), covid cases increased more rapidly than influenza cases in countries that were entering their flu seasons. That's compelling-to-me evidence that covid in fact has a larger basic reproductive number, whether that be due to a fundamental difference in the virus or just the effects of low baseline immunity. Now, your idea that the particular set of policy restrictions may be more effective against influenza than covid is very interesting. There's a basic horse sense to it, as well, since policymakers were largely operating from a playbook written for a flu pandemic. However, I do not think that "more effective" is a necessary component of the argument; "at least as effective" will do just as well given influenza's low baseline prevalence (out of season). Remember, my original point upthread was not to explain every bit of the relative epidemic curves, but instead to claim that we really ought not be surprised by the difference. I see little need to resort to specialized theories to explain the broad outline, even though I'm sure they will be helpful in the details.
- smolder 5y ago
  Your 2nd paragraph seems to have a mistake: influenza vs flu?
- timr 5y ago
  > we also have evidence in the actual observations of epidemic dynamics. Over the southern hemisphere fall of 2020 (northern hemisphere spring), covid cases increased more rapidly than influenza cases in countries that were entering their flu seasons. That's compelling-to-me evidence that covid in fact has a larger basic reproductive number Yes, we know that SARS-CoV2 increased, and influenza went away. That's trivially observed. We don't know why flu went away. To turn that observation around and say "Covid grew faster than flu, and therefore it has a higher R0, and therefore, the flu went away because {insert your favorite NPI} was just good enough to stop it" is begging the question. It could just as easily be any number of other factors that you're not accounting for. There's also a number of other faulty assumptions inherent in your logic. For starters: there's no particular reason, a priori, that a person couldn't have both flu and SARS-CoV2 in the same year.

samatman 5y ago

Small but important correction: nothing can make epidemiological spread non-exponential. Driving base reproduction below one additional infection per infected patient means the exponent is less than one, so the incidence of the disease rapidly decreases. This decrease is also exponential by definition: infection is simply an exponential process, and an exponential equation doesn't stop being what it is when the exponent is less than one. The steady state is exactly 1, which is like balancing a pencil on its point. It's plausible that two things happened here: the various local restrictions reduced the R0 of flu below 1, and the border closures kept mutations from traveling: and influenza is notoriously mutation-prone. These can combine to practically extirpate influenza. Unfortunately, the effect is temporary, and the next flu season is likely to be a real doozy. Which will be "fun" because our societies now have a hyperactive immune system, and there will be considerable political pressure to "do something" about the next flu wave. This must be resisted but I'm not optimistic.

Majromax 5y ago
> The steady state is exactly 1, which is like balancing a pencil on its point. At exactly one, it's indeterminate; you could see a subexponential growth or decay. With no exponential terms remaining, lower-order terms that are ordinarily hidden by exponential growth become the dominant observables. (Since epidemic spread is also random, you're much more likely to see the stochastic terms show up in the overall statistics in these cases. You really can be lucky or unlucky with Rt ~= 1, but there's no persistent luck with Rt >> 1.) This is also important before community spreading sets in, and it is the mechanism by which contact tracing can work. Rather than change the general susceptibility of the population to the disease (changing the effective reproduction number), contact tracing tries to operate at the level of individual cases in ways that cannot be scaled to uncontrolled community spreading (even if Rt is close to 1). > and the border closures kept mutations from traveling: and influenza is notoriously mutation-prone. Perhaps, but are influenza mutations an important effect within a single flu season? I know the determination of to-be-dominant strains has a great effect on the composition of each season's flu vaccine, but as far as I am aware we don't see declining flu vaccine efficacy over a season.
- 5y ago