Live data from Hacker News

Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

medrxiv.org

151–160 of 165 posts

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#151

Earlier quoted context omitted.

Well 20% of NYC had COVID and we are currently seeing outbreaks in communities that have relaxed social distancing measures, so it seems obvious that her calculations are incorrect.

That’s part of the thesis of the article. Some populations mix more than others. In the mixing populations the threshold will be higher. In populations with limited mixing, lower. 20% of NYC can have the virus and the general level for herd immunity can still be what is postulated. NYC is incredibly dense compared to the rest of the USA and as such will naturally have more mixing and a higher threshold than say Topek…

The USA as a whole is already at ~10-15% and the rate of infections increased since late summer, and we are not even in flu season

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#152

Earlier quoted context omitted.

FWIW they almost certainly didn't write on a cigarette package in this instance ;o) It's just the name of that type of maths because people used to use whatever small, available, piece of paper was around - so cigarette packs some time ago (you definitely could write on them with a ballpoint pen (ie un bic ). Getting tangential, napkins to me in the UK have always been cloth, and we have serviettes (a French word, me…

I can't decide whether to use the term cutlery (mainly UK) or silverware (mainly US). They're both inaccurate, especially when asking for plastic/disposable versions.

If it includes a knife then "cutlery" seems accurate? Sometimes I'll just make a compound of the items "spoons-and-forks" (when it's pasta for tea [tea meaning evening meal where I come from]).

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#153
post #138
post #64

Earlier quoted context omitted.

Are you seriously suggesting we lock over 110,000,000 people (34% of the population) in close quarantine for the duration? How do you plan to feed them? Get them medical care (be sure you don't overwhelm the system with the under 50s that get sick)? Keep them from rioting against their captors? What happens when you decide it's good enough and release them, and the residual infection sweeps through that population li…

No one is seriously suggesting we forcibly lock vulnerable people in close quarantine. Instead we should provide those at greatest risk with free hotel rooms if they want to quarantine on a voluntary basis.

Where they would have to stay. For the duration. With minimal contact. If they don't want to take a significant chance of dying.

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#154

Earlier quoted context omitted.

And I'm sure at some point one of them is going to publish an epidemiological model in support of their position, and not a press release. Until then, they're not even at Wired's level.

Here's a published epidemiological model that supports Kulldorff, Gupta, and Bhattacharya position. Though the author is not one of them. https://journals.plos.org/plosone/article/peerReview?id=10.1... One of the authors' twitter thread with the paper's summary: https://twitter.com/WesPegden/status/1288140129677332482

That paper doesn't consider reinfection risk or non-fatal outcomes.

There are many problems with the GBD, but the simplest is that we don't know who the high-risk groups are. Yes, we know age and certain categories of pre-existing condition make for higher risk of death. But we also know that perfectly healthy young people end up with strokes, heart damage, and lung damage, and we're not really sure why. We don't know why some people end up with debilitating symptoms months after infection.

We don't even know if herd immunity is actually possible, or if we'd be committing ourselves to years of intermittent lockdown controls as local outbreaks come and go.

This paper is a similar (if slightly more mathematically detailed) approach, and is more recent: https://www.pnas.org/content/early/2020/09/21/2008087117. It comes to the opposite conclusion. What they find is that while it's technically possible to achieve herd immunity this way, it's logistically unfeasible. It needs monitoring, compliance, and reactiveness that we demonstrably can't (or won't) implement - if we could, we wouldn't be in this mess.

Besides which, neither this paper nor that supports any idea that these three are "leading experts". As far as I can see they're vocal and have a history of being proved wrong by events.

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#155

Earlier quoted context omitted.

Here's a published epidemiological model that supports Kulldorff, Gupta, and Bhattacharya position. Though the author is not one of them. https://journals.plos.org/plosone/article/peerReview?id=10.1... One of the authors' twitter thread with the paper's summary: https://twitter.com/WesPegden/status/1288140129677332482

That paper doesn't consider reinfection risk or non-fatal outcomes. There are many problems with the GBD, but the simplest is that we don't know who the high-risk groups are . Yes, we know age and certain categories of pre-existing condition make for higher risk of death. But we also know that perfectly healthy young people end up with strokes, heart damage, and lung damage, and we're not really sure why. We don't kn…

> but the simplest is that we don't know who the high-risk groups are

We absolutely do. We have such a wealth of data and the signal is very strong.

> That paper doesn't consider reinfection risk or non-fatal outcomes.

That's because reinfection is extremely rare and risk for non-fatal outcomes is typical of other influenza like illnesses. An interesting note is that many / most people have some sort of cross-protection through T-cell immunity (likely from other coronaviruses).

> We don't even know if herd immunity is actually possible

Yes we do. Pretty much every disease tails off. The only debate right now is where this threshold is at for various jurisdictions. It is likely as low as 20%. The 60% number quoted early in the pandemic was assuming homogenous population with equal susceptibility and perfect mixing.

> This paper is a similar (if slightly more mathematically detailed) approach, and is more recent: https://www.pnas.org/content/early/2020/09/21/2008087117. It comes to the opposite conclusion. What they find is that while it's technically possible to achieve herd immunity this way, it's logistically unfeasible.

All models are wrong but some are useful. If this model cannot explain real data from cities and countries (eg: stockholm, UK locales) then it is relatively useless.

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#156
post #116
post #80

Earlier quoted context omitted.

This also assumes that acquired immunity from an infection is complete and permanent. Immunity to other coronaviruses decays over a year or two -- meaning that before the five years are up in your scenario, there would be a significant cohort of prior victims open to reinfection. See, e.g., https://www.nature.com/articles/s41591-020-1083-1

So, maybe that means by slowing the infection, what we're really doing is ensuring a neverending slow wave of infections ... forever? Maybe what normally happens with these kinds of viruses is they infect everyone then die out? And if you slow that, they never die out?

No, it means that the expectation that "herd immunity" will be naturally reached, without an effective vaccine, is a pipe dream. Before vaccines, no population ever reached "natural herd immunity" for polio, which has similar R values; you need effective vaccines, effectively delivered, to get that result.

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#157

Earlier quoted context omitted.

That’s part of the thesis of the article. Some populations mix more than others. In the mixing populations the threshold will be higher. In populations with limited mixing, lower. 20% of NYC can have the virus and the general level for herd immunity can still be what is postulated. NYC is incredibly dense compared to the rest of the USA and as such will naturally have more mixing and a higher threshold than say Topek…

The USA as a whole is already at ~10-15% and the rate of infections increased since late summer, and we are not even in flu season

Just because the US average is X% doesn't mean that the every community in the whole country is X%. There's clearly heterogeneity in the data.

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#158

Earlier quoted context omitted.

Maybe I'm missing something but it seems like you're just independently inventing the concept of a vaccine.

Or a Pox Party: https://en.wikipedia.org/wiki/Pox_party Not that outrageous.

Jail inmates in California did that in an attempt to get released.

https://www.cnn.com/2020/05/11/us/california-inmates-coronav...

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#159

Earlier quoted context omitted.

That paper doesn't consider reinfection risk or non-fatal outcomes. There are many problems with the GBD, but the simplest is that we don't know who the high-risk groups are . Yes, we know age and certain categories of pre-existing condition make for higher risk of death. But we also know that perfectly healthy young people end up with strokes, heart damage, and lung damage, and we're not really sure why. We don't kn…

> but the simplest is that we don't know who the high-risk groups are We absolutely do. We have such a wealth of data and the signal is very strong. > That paper doesn't consider reinfection risk or non-fatal outcomes. That's because reinfection is extremely rare and risk for non-fatal outcomes is typical of other influenza like illnesses. An interesting note is that many / most people have some sort of cross-protect…

> We absolutely do. We have such a wealth of data and the signal is very strong.

We know who is likely to die. We do not know who is at risk of a life-long debilitating illness.

> That's because reinfection is extremely rare

We don't know this. What we know is that reinfection with a different strain is rarely detected, and that's a long way from the same thing.

> risk for non-fatal outcomes is typical of other influenza like illnesses

This is false.

> An interesting note is that many / most people have some sort of cross-protection through T-cell immunity (likely from other coronaviruses).

At best this is optimistic. We know some (less than half) have a T-cell response. We don't know yet if that response is beneficial, harmful, or has no effect at all. It would be premature to start any sort of public health intervention founded on this assumption.

> Yes we do. Pretty much every disease tails off.

This strongly depends on the reinfection rate. Which we don't know.

> The only debate right now is where this threshold is at for various jurisdictions. It is likely as low as 20%.

This is false. To get anywhere near 20% you need to know the effect of the T-cell response, or have some other mechanism for discounting a large portion of the population.

> All models are wrong but some are useful. If this model cannot explain real data from cities and countries (eg: stockholm, UK locales) then it is relatively useless.

Have you read either of them? Both models in this thread are predictive models of situations that haven't happened yet. Both use real data (from the US and the UK). Neither can describe reality, so do we throw them both out? That leaves the GBD lot with no epidemiological support at all, which would make my point rather concisely.

We simply don't have enough information to know whether the GBD proposal is safe or, even if it was, whether it could be implemented, and it's all the more suspicious because its three proponents have been making very similar arguments against general lockdown since at least April, when we knew dramatically less. They do not seem to have changed their stances based on new information, which moves the GBD out of science and into politics. Only they're leaning on their academic credentials to lend it airs of legitimacy it can't back up, which makes it complete, utter bullshit that nobody should pay any attention to. It's preying on desperation and optimism to deepen social division and reinforce political hysteria at the worst possible time. No credible health authority is paying any attention to it, nor should they. Please don't bring that sort of content to HN.

Re: Updating Herd Immunity Models for the US: Implications for the Covid-19 Response

#160

Earlier quoted context omitted.

Nope, you're not even close to calculating the survival rate correctly. The population of ~43 million is the entire population, not the set of people who were infected.

Schucks, you’re right, I didn’t read the fine print on the CDC site. Still it’s not as far off as that sort of error would seem to indicate. According to this Sep 2020 article on reason.com (1) they quote CDC numbers of a 99.98 survival rate (0.02% IFR) for 20-to-49-year-olds. Or a 99.997% survival rate (0.003 IFR) percent among people 19 or less. 1: https://www.google.com/amp/s/reason.com/2020/09/29/the-lates...

Glad you took another look!
Post reply on HN