I don’t know about you but as an ecologist, I am not an expert in disease dynamics nor part of the inner community rapidly exchanging ideas and data. But as an ecologist I have a better handle on notions of population growth, species interactions, individual encounter rates, etc than the average population (and probably the average scientist) and I have felt in a frustrating vacuum of information.
To address this, we’re trying something new here at Dynamic Ecology – an open thread, the main purpose of which is to have a place for the community to have a conversation. Our comments sections have long been the most interesting part of the blog, so now we’re creating a direct path to comments without your having to read 1000s of words of bloviation from me!
First, a few thoughts to give some common terminology/framing to the questions. I think ecologists all know about the power of exponential growth (although this is new and still poorly grasped to most of the world). R0 is the discrete growth rate with no immunity (naive population) and no efforts at social distancing. Best estimates I have seen for Covid 19 is about R0=2.5 which is a good bit higher than flu (and a good bit lower than measles). It seems to be becoming clearer that R0 is as high as it is because people can be infectious before they show symptoms (or even if they never show symptoms like children). Once immunities start to build up or quarantine/social distancing measures start to be put in place a lower growth rate Re (effective growth rate) is observed. So as far as I can tell there are three strategies.
- Squeeze it – extreme social distancing to reduce Re<1. This seems to be what China as well as Japan and South Korea are doing (probably not coincidentally all Asian countries that got hit most by SARS and MERS).
- Let it burn – do nothing to lower Re=2.5. Sadly many (all?) countries started down this road – with exponential growth the speed of reaction required seems to be faster than governments can handle.
- Stretch it – social distancing to get Re~1.2 (nb 1.2 is an example, not a carefully calculated number, just a wild guess proxy as it is about what influenza does) so that the case load does not exceed hospital capacity. This is what everybody is talking about as “flattening the curve”.
With the stretch it and let it burn strategies the number of people who get sick and then have immunity rises to about 1-1/R0 or about 60% of the population (assuming getting sick once confers immunity – assumed right now but a few counter examples are out there). Then the effective growth rate Re drops below 1 and “herd immunity kicks in”. Individuals can still get sick but it can’t become a self-sustaining epidemic. The primary difference between let it burn and stretch it is the rate at which people get sick which is inversely correlated with how long the epidemic lasts.
I’ve posed several questions below to get this started. I’m not an expert. So the answers to some of these may be obvious in which case, I’d love to know the answer. But I have not seen the answers to any of these despite voracious reading. If they’re not so obvious I expect we could all learn from discussing them.
If you want to respond to a question stay in the same thread (even if the nesting stops at 3 levels). If you want to pose a new question, start a new thread. This is NOT a place for politics, so anything stronger than “many governments have been incompetent at X” (e.g. naming specific individuals, blaming one party or another, or getting distracted off science) will be deleted.
Question #1 – How long will the let it burn or stretch it processes take to reach herd immunity? Is it 2 months? 6 months? I’m pretty sure it is not 2 weeks like many people and governments are acting. Why are we not talking about this number? What is it (pick your guesses at parameters)?
> 6 Months, < 1 year. I think the 2 weeks number is confusion between the duration of the pandemic and the duration of an individual isolation period to be sure you're not infected. More than 6 months because it's taken near 4 months to rein in case numbers in China with a squash it strategy, I think the other two must take longer. Stretch it gets R_e down to seasonal flu levels, and that is over each year, more or less.
The “stretch it” scenario could take a really long time to get to an immunity level of 60% without overwhelming the health care system. Take the US, with roughly 300 million people. If the limit on health care capacity is ventilators, and 5% of cases require a ventilator, that results in:
3*10^8*0.6*.05 = 9 million people needing ventilators.
If there are ~100K free ventilators (which I think overestimates the current capacity), and typical ventilator use duration is 2-weeks:
It would take 180 weeks for the stretch it out process to work. Obviously all of these number have a lot of uncertainty. But even if the numbers are much more favorable, the “stretch it” scenario would still take a very long time.
This is an excellent point, but gives us humans more agency than we have. To avoid a lot of deaths we need to keep peak cases below the number of ventilators available, and that means we need stretch it to last for > 3 years (with your numbers). But that’s not really up to us, even with social distancing the virus will do what it will do. How many humans die along the way is up to us, or rather, was up to us because of policy choices about how many ventilators we need to have.
There is a really nice, and short, discussion of the challenges of forecasting COVID-19 from a stats prof here https://robjhyndman.com/hyndsight/forecasting-covid19/
There is also a web-based SIR model here https://alhill.shinyapps.io/COVID19seir/
where you can choose different parameter values and see the effect.
Question #2 – What is the long term goal of the “squash it” strategy. With SARS and MERS they got the infection rates low enough they could revert to a contain (track each case and isolate all contacts) until they killed out the disease. But that seems much less likely to me with Covid-19. For one thing the infectiousness before symptoms makes the track and isolate very difficult. For another this virus is all over the globe, so what do China or Japan do while the virus is still running around? Hard closure of borders (something that has worked rarely in history)? My understanding is that although the case load is down in China it is not eliminated – so isn’t China opening itself up to a new spread once the social distancing is lifted?
South Korea and Singapore are two other examples of squash it working in the short term. I think this is an excellent question; seems like long term we have to figure out how to live with COVID-19.
Question #3 – What would a clever let it burn strategy look like? There are rumors that Britain is trying to do this. Since most people have mild symptoms (milder than the flu) keep R0 high and get herd immunity built up, but keep the vulnerable population out of the 60% that are infected. A very calculated strategy in this direction would keep schools and colleges open (since youth rarely – but not never – suffer serious symptoms) but put hard lockdowns on nursing homes and retirement communities and tell people with immuno-compromised or respiratory-compromised to isolate in their homes. Is this diabolical or crazy like a fox? Is it ethical? Could it even work? What else would be needed?
Keeping schools – especially universities – open would be madness under current conditions. There are very few places in the world other than universities where large numbers of older (>60 yrs), highly vulnerable adults routinely inhabit the same room and engage in face-to-face contact with young adults (ca. 20 yrs) who can survive COVID-19 and act as carriers. Keeping universities and colleges would endanger senior faculty. I do not think we want to burn our elder faculty to the ground (as a card-carrying member of that group, I feel especially strongly about this point!). Perhaps post-docs angling for a job might feel differently. ;^)
Same argument applies to K-12 schools. And I know this argument was the primary reason given by Harvard (one of the earlier universities to shut down).
Yes, but my impression comparing K-12 and universities is that the older folks make up a substantially larger fraction of the faculty at universities. Hence my use of “especially”.
Here are a few entry points into discussion of this. Trying to keep it to the most science-focused, technically-informed discussants of which I’m aware. But I’m sure I’m missing some good people.
Adam Kucharski, who is involved in the UK modeling efforts that are informing the UK government’s decision-making, is an essential read here. https://twitter.com/AdamJKucharski. For instance, see these threads (the last of which argues that the UK government’s plans have been widely misinterpreted, or perhaps poorly communicated):
And here are some fairly informed-sounding critiques of the British strategy:
One bit of further context: the British have a lot of the best epidemiological modelers in the world. They have long experience with developing epidemiological models in real time to evaluate alternative policy options, going back to the foot & mouth outbreak in 2001. I know a couple of epidemiological modelers with British training, and have read papers from many more, and they’re all *very* good. So even if you don’t think much of Boris Johnson or his senior advisers, the modelers who are feeding them information and advice are intellectually honest, hardworking, care deeply about protecting the public, and are as good at their jobs as anybody in the world. That obviously doesn’t mean they’re infallible, of course! They could be getting it very wrong. After all, there are other very good people–including in the UK–who think they *are* getting it very wrong. But I doubt that they’re *obviously* wrong. They’re not incompetent. And they’re not Dr. Strangeloves–they don’t have some evil level of unshakeable overconfidence in their own smarts.
Now at least one of the British teams working on this seems to have come to different conclusions than some previous British analyses:
The short version seems to be that they’ve now decided the UK medical system would be overwhelmed before herd immunity was achieved.
“even if you don’t think much of Boris Johnson or his senior advisers, the modelers who are feeding them information and advice are intellectually honest, hardworking, care deeply about protecting the public, and are as good at their jobs as anybody in the world.”
Undoubtedly true Jeremy, but a lot depends on whether the government is willing to implement that advice. And its track record so far is not inspiring: for years it ignored expert advice on the effects of badger culling on spread of bovine TB and continued to kill badgers. It’s only just reversed that decision.
Another issue with closing Universities is that you are likely to send large numbers of people in low risk (outcome wise) groups back into the community, and interacting more frequently with at risk groups. Many universities are campuses which can be nicely contained.
Universities aren’t that well contained; many students at my institution live off campus. I’m less worried about closing universities because students are adults that can follow social distancing rules if they choose to do so. We’ve cancelled classes for this week, next week is spring break, and then will do online classes for the rest of the semester. I’m pretty sure this is a good response that will contribute to stretch it.
K-12 is a bigger problem, especially for lower income families that may struggle if parents can’t get to work. Might make more sense to keep them going to school, and work at detection and isolation. Big might there. Identify teachers and children at high risk and isolate them, that would be key.
Was referring more specifically to socially contained than physically (and that is from my UK perspective), but also most students will living on campus (in the UK). Which does highlight why different countries will have different strategies: We all have different social/economic/healthcare/political systems in place, so strategies which are optimal in one country are not guaranteed to be optimal in the next.
Question #4 – What is the science on travelling or not? I get banning travel from say Iran into the US when there were no cases in the US (or even today given that Iran has a higher infection rate). But is there a scientific justification (medical benefit) to cutting out travel between say the US and Europe which have roughly equal levels of infection? (although with such limited testing it is hard to know exactly). Obviously if people are on crowded planes and airports that’s not good. But what if flights were 1/3 full so you could sit 2 meters apart and airports were empty and you could avoid touching surfaces in airports? That’s a lot of ifs but might be conceivable in a travel-averse world. Is there then any reason to eliminate travel?
If air travel from abroad is coupled with the idiotic Trump policy over the past 36 hrs off greatly slowed screening -> massive crowd contact and resulting balmy days for contagion in the customs line, then banning air travel is a very good idea. Otherwise, yes, increasing social distance and/or using effective masks would allow travel between countries/states of roughly equal infection rates without elevating infection rates in either.
I think airplanes use 50% recycled air and 50% fresh air, which is mixed/filtered through a HEPA filter. HEPA filters are not fine enough to capture viruses. Any airplane engineers, please correct me if I am wrong.
And no matter how empty the airport is, everyone is usually squeezed through the same security sections. Australia’s policy right now is mandatory 14-day quarantine for all people coming in on an international flight, regardless of where.
I think a case can be made for a wide travel ban. International travellers can be categorized into two broad groups: leisure and business travellers. Those travelling for leisure can be easily persuaded to stay at home but business travellers may not be for the obvious reason. With a wide travel ban, businesses could suspend their overseas business travels without losing the business opportunity or risk the health of their employees because all the parties are aware of the constraints. This is perhaps more effective than asking a business traveller to take a moral initiative. But the economic fallout is another story.
Early on there has to be, to stop establishment. But once the epidemic is raging, the added input from immigrants is probably so small it’s not a major problem. Obviously this will depend on the rates of movement and changes in the epidemic in the areas people are coming from.
I actually worked on this issue during my PhD. OK, so it was mildew on barley, and not human viruses, but that means I can claim more expertise than most people. 🙂
Chinazzi et al. 2020 (DOI: 10.1126/science.aba9757 ) modeled the effects of travel restrictions in combination with transmission reductions (social distancing) and found even 90% reductions in travel didn’t make much of a difference. I think assuming it is everywhere already is probably the most reasonable. So if you’re high risk, don’t travel. Otherwise it probably doesn’t matter anymore.
I don’t understand the travel bans either. At this time, most countries have relatively high numbers of cases, and those that have lower rates are also short on test kits available, which means that they likely have a lot more people actually infected than they are reporting.
So if the main impetus to restrict flying is to limit the airplane and airport ‘incubator effect’, we should restrict *all* flights. Both domestic and international.
Question #5 – what is the science about small groups? Obviously banning large gatherings is key. And on a certain level more is better. But is allowing groups of 20 people who live in the same city crazy? What about 10? or 5? Is there some trade-off. For example if having no gatherings >20 people drops Re down 0.9 and having no gatherings>3 people drops Re down to 0.88 is that worth the disruption? What is the science of how these smaller groups affects Re? My observation is that the US has gone from no limits to trying to eliminate anything much more than a 1-1 meeting (group size of 2 outside of family you live with). What is rational and reasonable here? What science is there on small groups and impact on Re?
It depends on how frequently group members meet with other groups. If everyone meets with everyone one-on-one, there is no point to banning large groups … the metapopulation becomes the population even though only two people are ever face-to-face.
Yes – I guess ultimately this is a question about the nature of social networks. Which has been fairly well studied. I wonder if those studied have ever been applied to epidemiology?
See Christakis, Nicholas A, and James H Fowler. 2009. Social Network Visualization in Epidemiology. Norsk epidemiologi = Norwegian journal of epidemiology 19: 5-16.
I think the variability within groups is the reason for banning large gatherings. For instance, a gathering of 20 people for a recreational soccer game would shift Re upward compared with the gathering of the same size for a departmental seminar. The latter group is more predictable, and even with that, there are still some uncertainties.
I’ve seen no good information about what the distribution of virulence looks like. Anyone know? This would seem to be pretty important.
Maybe I should have been more clear. I’m aware of the effects of age, but it seems more complicated than that. Some people are apparently non-symptomatic, while others are more vulnerable.
Not a fully satisfying answer, but this is the best I’ve seen so far: https://www.vox.com/2020/3/12/21173783/coronavirus-death-age-covid-19-elderly-seniors (it cites several pieces of the primary literature I have seen). Age matters (e.g. increased odds of cytokine storms) but so do the risk factors like heart disease and diabetes regardless of age.
I don’t know if anybody knows yet what causes people to be asymptomatic (even at the level of why young people are like this more often)
“do nothing to lower Re=2.5. Sadly Italy and Iran started down this road.”
This is just not true (and honestly I find it a bit offensive). Let’s talk about Italy.
– with the exception of a few tourists, there where 0 reported cases in Italy before Feb 20. On Feb 23 there were <200 cases mostly concentrated in a few towns in Lombardy and Veneto. Those towns were put under complete lockdown on Feb 23rd (no one was allowed to enter or leave).
– On Feb 24rd there were 221 confirmed cases. Schools were closed in all northern Italy starting Feb 23rd. In Lombardy and Veneto pubs and bars were allowed to open only from 6am to 6pm. Movie theaters were closed in Lombardy and Veneto. With 200 cases.
– In the region where I live (Friuli Venezia Giulia) there were 0 reported cases (1st case was reported on Feb 29). Schools and universities were closed on Feb 24th. Also, seminars and conferences were suspended from Feb 24th.
Clearly, these measures were not enough. But, retrospectively, Italy acted *much earlier* than many countries (France, Spain, UK, Germany and let's not even talk about the US).
Saying "do nothing like Italy" is a dangerous message. Nobody knows how much is enough to reduce the spread of epidemics. Saying that the current situation in Italy is due to inaction gives the false impression that "doing something" might be enough. The important point is exactly the opposite: closing schools, closing restaurants, creating red zones when you have 200 cases (and, please note that most of EU countries and US have many more cases and are not doing even that) is not enough. The only (surely short-term) solution to avoid having a collapse of health-care is lockdown. The other measures have some effect ( https://cmmid.github.io/topics/covid19/current-patterns-transmission/global-time-varying-transmission.html ), but the effect is too small and too slow.
So the point is exactly the opposite. Italy acted much more quickly than France, Spain, Germany, UK, US, etc. And it was not enough.
There is another point, perhaps less important, that hurts me a bit in the comparison between Italy and Iran. Iran governement is hiding the numbers, the Italian government is updating everyday at 6pm the data, which are available on github https://github.com/pcm-dpc/COVID-19 There is also a discussion about testing and transparency in other countries, e.g. in the US, that should be addressed
@Jacopo – I take your point. I apologize. I certainly agree Iran has gone down a different worse path than Italy. I think much of the offensiveness came from the name I put on the approach of “let it burn” which I know nobody in Italy (or any country) intentionally did. And I’m not sure I really meant anything more than Italy hit the exponential growth phase faster than other European and North American countries which is just a matter of bad luck (which gave other countries a chance to observe and alter their behavior, which as you point out most did not). For sure, my country, the US has stayed on the exponential growth phase until far too late.
I have removed the language. And I appreciate your facts. They are indeed frightening about how fast a response was needed. And you won’t get disagreement from me that the US lack of testing has surely depressed our numbers.
Brian, thanks a lot! I re-read my comment and I apologize if I was too harsh on your post.
My comment came out of worry for all the friends I have in the US and in other EU countries. Nobody knew how of much a measure (e.g. closing schools and restaurants) decreases R0. What the “Italian experience” shows is that closing schools, asking people to wash their hands, closing restaurants (measures that some good governors in the US are taking *now*) is not enough. Doubling the number of ICU beds, is not enough (you can’t fight an exponential with linear measures). My feeling is that all these countries will eventually go under lockdown as Italy (assuming that lockdown is enough). But waiting a few days means doubling the number of cases and put healthcare under huge stress. If you are trying strategy 1 you should commit to it and do it as fast and effectively you can.
It is just a bit frustrating to see that other countries are “wasting” this “experience”.
I really do not see how strategy 3 can work. Fine-tuning R0 seems much harder than pushing it below 1.
Your comments were entirely fair. I appreciated them.
Yes we are all pretty much guessing in the dark. It has been 100 years (4 generations) since we had a pandemic like this (Spanish Flu). It is frustrating to see how slow countries are to learn the lessons of those first hit.
I think you’re right that we cannot calibrate our response just do everything we can whether that is R0=1.2 or R0=0.9. The US, after taking no action has basically slammed into high gear really in the last 4 days (yes certainly too late).
My concern though is if we are all being realistic about how long the extreme social distancing measures can be sustained and if that is even close to how long they would need to be sustained to beat the virus.
Not a question, but an invitation — I’m testing out my population modeller forecasting skillz at drewtyre.rbind.io . Stop by and tell me how badly I’m doing!
General remark: there are two big difficulties with forecasting exponential growth:
-small errors early on multiply into big errors down the line
-lots of different models that have very different dynamics down the line can fit exponential growth about equally well early on
Indeed! Which is why I’m working on adding more models to the mix, as well as countries. I think of exponential growth as a null model which is as informative when it fails (and how) as when it is correct.
Just a thought: if you’re just playing around with data, you might want to keep it private? Or set it up so that only people you specify can see the page? You don’t want a half-baked forecast to go viral.
With the “stretch it” strategy, doesn’t the herd immunity only go up to ~ 1 – 1 / R_e, not 1 – 1 / R_0? I guess then you stop stretching and and get a second epidemic that brings it up to 1 – 1 / R_0. But for this second epidemic to be not nearly as bad as the original “let it burn”, you need to relax the stretching very slowly.
That sounds right to me. It basically means the stretch it out and the squash it have some common features of not reaching full herd immunity and risking a second phase. My hypothetical stretch R0=1.2 would get immunity only up to about 17% – which still leaves a lot of people likely to get infected before it goes up to the natural “non-social-distancing” required herd immunity of around 60%.
Of course one of the goals of the stretch it out is to get us to the place where we have a vaccine which will achieve robust herd immunity in a completely different fashion.
Yeah, “stretch” and “squash” seem similar enough to me in good effects, while being being so different in number of cases, that I think “squash” basically dominates “stretch” as a strategy. The only way it wouldn’t be true is if those last social distancing measures needed to push R_e down from 1.2 to 0.9 were incredibly more onerous than those needed to push it from 2.5 down to 1.2, but I think that the Asian countries show that this isn’t the case.
Basically, there’s a phase transition in the disease at R_e = 1, and there’s no reason to expect there to be a similar huge change in the costs of social distancing at the same point, so it never makes sense to leave R_e hanging around just above 1.
Not sure how useful this is to share preprints from reputable groups that I happened to stumble across–I’m clearly not the best person to be filtering the massive preprint literature on corona virus projections. But caveat emptor, here’s this:
Preprint: comparative analysis of effectiveness of various transmission control measures. I’m relying on the names of some of the co-authors to think it worth sharing. It’s from folks who are working with Ottar Bjornstad and Brian Grenfell.
Another good Twitter follow for pointers to news, data, and modeling: Ottar Bjornstad: https://twitter.com/BjornstadOttar. I suggest Ottar because I know him and know his work; he’s very good.
It seems to me that the large scale differential-equation analytical-type modeling (i.e. R0, Re etc) would be far less useful than stochastic modeling for this disease. Of course, those stochastic parameters could vary widely, and would be hard to get a handle on. Consider what happened just recently, the fiasco at several airports where hundreds of persons, some probably infected, were all bunched together for hours.
At best, one might be able to model many bulk effects analytically, but you’d have a significant extended contibution from discrete idiotic incidents such as those airport customs/health screening fiascos, or from non-compliant persons — and the latter effect is no small issue, many younger people have no real sense of danger (doubly-so when the danger to them is relatively-small), and many right-wingers — of any age — seem to reject the precautionary principle.
As biological analogy, I think you are talking about an invasive species’ seed dispersal in tornado country. Yes, you can probably model dispersal reasonably well most of the time, in good weather, but when a storm comes along, all bets are off.
Question: is there a way that those of us now sitting at home can remotely assist the epidemiologists/virologists/companies developing treatments as they attempt to rapidly ramp up their work? Many of us are not trained in the specifics of these fields but have some familiarity with them, and skills that could perhaps be helpful.
It’s entirely possible that the optimal contribution from random academics they do not know is zero. But if not, I’m sure many of us would gladly pitch in.
I haven’t looked into this yet, but there seems to be a way to donate your computer’s background CPU and GPU to researchers and clinicians. Could be worth a shot!
Thanks! Will look into it.
Here’s a crazy thought.
What about a 4th strategy: the “covid-19 party”? Its even more extreme than the “let it burn” strategy, by increasing infection rates in healthy people in order to rapidly boost herd immunity and protect the vulnerable ones. So basically, get everyone who has strong immunity together into a massive covid-19 party and get them sick. We do this for chickenpox. So why not? We’ll all be mildly sick for a week, then herd immunity will be very high, and the vulnerable people are less likely.
The caveat here are:
(i) I am assuming that the fraction of healthy people getting seriously sick is almost nil, so that barely anyone ends up at the hospital (otherwise we would still be overwhelming our health systems), and
(ii) I am assuming that the virus isn’t evolving fast enough to outpace herd immunity.
This is definitely extreme and I am just having fun brainstorming scenarios, but if the two assumptions above are correct, couldn’t it work?
Pretty sure that’s a very bad idea, for three reasons:
1.kids and other young people can infect other people, even if asymptomatic themselves
2. following on from #1, those who get infected can in turn infect others, and so on.
3. sometimes even young people do experience severe symptoms
#1 and #2 together imply that infection parties would increase the short-term growth rate of infected cases, increasing the odds that the healthcare system gets overwhelmed in the short term. Which in turn leads to more deaths. And some non-zero number of those deaths would be kids.
Have you all seen the Imperial College Covid-19 Response Team’s recent modelling analysis: https://t.co/ZejfSQcO0Y?amp=1
I’ve seen a lot of flatten the curve conceptual diagrams, but this was the first analysis I’ve come across with realistic scales for the x- and y-axes. The results are pretty disturbing. Basically, a “stretch it” or “mitigation” strategy that involves case isolation, home quarantine, and social distancing of >70s, implemented over a period of 3 months (but would only start after things get much worse in April), would flatten the curve by 2/3rds and cut the predicted mortalities in half. However, even this strategy would overwhelm the health care capacity by almost an order of magnitude and result in an estimated 1.1 million deaths in the US. The authors come to the conclusion that such a strategy is unacceptable, and instead a strategy of suppression (like China has implemented) should be used instead. The catch is that this strategy would have to be maintained for ~18 months to allow time for a vaccine to be developed. Such a long “lockdown” would obviously have profound economic implications.
From their analysis all three options looks pretty bad at the moment:
No action: ~ 2.2 millions death in the US alone
Mitigation: ~1.1 million deaths (in the US), fairly severe economic/social disruption for 3 months
Suppression: much fewer deaths, very severe economic/social disruption for ~18 months
My hope is that suppression now buys time to find new solutions to this problem. For example existing treatments for other diseases may turn out to be effective treatments for covid-19 and could be rolled out on a faster time scale.
Just linked to that up above; thanks for summarizing.
This is what I have been looking to find. Their 3 cases are very parallel to my “let it burn” “stretch it” and “suppress it”. They are specific in that “stretch it” would involve isolation of identified cases, quarantine of people living with cases, and social distancing of at risk. And “suppress it” requires adding social distancing of all ages/health risks (the school closures, bar/restaurant closures, etc)
But they confirmed my basic suspicion/worry that “suppress it” cannot be relaxed (until a vaccine comes along) without repeats of the epidemic (although they raise the possibility of episodic imposing and releasing from social distancing from all).
It remains to be seen whether even an autocratic government and fairly socially oriented society like China can maintain broad social distancing for 12-18 months. I find it hard to imagine this being tolerated in many other countries. Not to mention I cannot imagine what havoc that would have on economies.
Not very good news.
The more I read about the development of the UK government’s strategy (e.g.: https://unherd.com/2020/03/the-scientific-case-against-herd-immunity/), the more confused I get about what data and evidence it was based on, and why it apparently changed. Were the epidemiological modelers making dubious assumptions, or was the government relying on dubious assumptions about people’s short-term tolerance for social distancing, or what?
The prime minister of Netherlands (where I just started a TT job 2 weeks ago), gave a speech on Monday, where he actually listed these 3 different strategies of do nothing, mitigate, and suppression. He said that under the guidance of the Dutch public health institute, RIVM, that we are going to focus on mitigation. The idea is to control the rate of the epidemic and gradually establish herd immunity. For now schools, universities, restaurants, and bars are closed, but they will adaptively adjust social distancing measures as new data and science comes in. It was encouraging to hear the emphasis on science, but I will note that lags and measurement error generally make controlling dynamical systems quite difficult. It is also interesting how different but well respected public health research groups are coming to very different conclusions about what needs to be done. The Dutch PM, on the guidance from RIVM ruled out a suppression strategy on the basis that we could shutter people indoors for months just to have it rapidly return as soon as we relax these measures. Clearly the ICL group came to a different conclusion, but in the discussion of the paper the they allude to the fact that they may have changed their minds as new data on ICU demand came in from Italy. I am not sure what strategy the CDC is advocating for.
“I am not sure what strategy the CDC is advocating for.”
As a US citizen, I’m old enough to remember when the CDC was the world’s gold standard. Sigh.
Very interesting news about the Dutch policy.
So what’s the best critique of the Imperial College study? Here’s Nate Silver arguing that their assumptions about what interventions are possible, and how effective they are, *might* be pessimistic: https://twitter.com/NateSilver538/status/1240156067323052033
And here’s a critique from Nassim Taleb:
But I have yet to see good critiques from actual epidemiologists, as opposed to media figures with some quantitative expertise. Anyone found any?
And here’s a critique from another person associated with Taleb. Argues that the Imperial College paper gets it wrong because it ignores contact tracing:
And here’s some commentary on the Imperial College London report from virologist Trevor Bedford:
Includes detailed commentary on how scaled-up testing and contact tracing could work, and why that might be the way out of the catch-22 described in the Imperial College London report.
Pingback: Ecologists in a time of covid-19 | Jeff Ollerton's Biodiversity Blog
I’m a little disappointed that all of the discussion so far has been about the epidemiology of the disease with no one considering its origins and how we can reduce the risk of these zoonoses emerging in the future. I appreciate that this is how Brian framed the questions to begin with, and that’s fine. But as ecologists we need to look at the longer-term implications of such diseases. It’s clear that many (most? all?) of these diseases are coming from human contact with wild animals killed for food or other uses. That in turn is due to people entering previously inaccessible areas more frequently because of logging, road building, etc., and is being facilitated by poor or corrupt enforcement of laws protecting wildlife and ecosystems.
Rather than just speculating as to how the virus will spread and what its effects on human populations might be, I think that ecologists should be promoting the message that these epidemics will keep occurring until we strengthen and enforce existing laws. This can ‘t stated too often. Here’s a good starting point for anyone interested in learning more:
Maybe its a timing thing?
Everybody just wants to figure out how we’re going to survive the virus now? Then next we turn how to avoid repeating it of which leaving wild animals alone might be a good part?
Yeah, maybe. My worry is that once the pandemic is over we’ll forget about how it was caused in the first place. Then we just go on as before and next time it will be an emergent disease with even greater impact….
Anyone interested in the Chinese data check: https://ncov2020.org/
Pingback: Dinâmica de populações, crescimento exponencial e coronavírus – Mais Um Blog de Ecologia e Estatística
If you are still on this thread, I have a lingering question concerning the end of measures, though I am not sure if the answer is hidden somewhere in the answers. How do we know when we can reduce the measures now imposed on travel bans and closing of society? This virus will likely survive in a lot of pockets over the world, but somewhere along the road we will have to open up the society again. We can of course wait until there is vaccination available, but it may be that the situation improves within single countries, or measures just have to be reduced for economic reasons. We will not know how many persons that have had the virus. So, are there any models handling the behavior of epidemics in the other end that can be used for this purpose?
Yes – this is basically my question as well. Its one thing to say “do everything we can to stop the spread” but if that means lockdown for 18 months (as the paper cited above suggests is at least possible) that may not be possible. I’m not seeing a clear description of the long-term plan.
I have a feeling that after reaching a certain number of cases, it won’t make sense to continue the lockdown anymore and governments may just open up the economy. Until then, governments may try their damnest to control the spread with these restrictions. What that number is could be specific to each country, depending on the population and depending how the society functions.
Not sure if this is THE answer, but I feel this is plausible.
Anyone seen results of economic models of the different strategies? You can’t deal with the health aspect without affecting the economy and vice versa. Are epidemiologists and economists working together to carry out reality checks?
I agree this is important. And I have not seen the whole package put together (health+economics). Which I find very worrisome.
Sorry, but I haven’t trusted economists for years, and I’d advise a high level of wariness regarding collaboration with most of them. It’s not just for Malthusian reasons that economics is referred to as the “dismal science”. It may take a substantial effort to separate any wheat from the mounds of chaff. And many economists know “the cost of everything, and the value of nothing”. Additionally, factors such as speculation, over-exuberance, demands for maximum efficiency and return-on-investment, and the almost complete subjugation of humanist values to financial ones, all encouraged by many “free-market” economists, have led us into at least three major crises in the past two decades: the dot-com bust, the 2008 financial collapse, and the current unpreparedness and capacity shortfalls of healthcare systems in multiple countries. (In my listing of crises, I haven’t even included the large adverse role of market economic “values” in the biodiversity crisis, the climate change crisis, and the higher-education funding crisis. Monumental as these problems are, their extended time frame for both development and effect means that “crisis” may not be the most apt term for any of them).
As far as economic reality checks are concerned, here’s one, about the predictive abilities of economists: if they are so good at it, why aren’t more of them highly-charitable benefactors and/or extremely wealthy, with large fortunes accumulated primarily from their own investments? (Even if not at all interested in a fortune for any personal extravagance, surely they must be aware of the beneficial uses that accumulated wealth might be put to).
Fred – you may or may not be right about economists (not really the place for it here). But surely the economy (whether people have jobs, can keep a roof over their head, basic supply chains of food & medicine functioning) is needing consideration!
The combined effects of control strategies on public health and the economy are fraught. Ultimately, we’re talking about a tradeoff between lives and income. You could ask which strategy would yield, say, the greatest GDP or GDP/capita, in terms of benefits at a national or international scale, but how do we value a person’s life against $$$? We see what Trump is proposing now … which seems to value life very little and income a great deal. My own sentiments are that there will be no economic recovery until public health is restored, and no real recovery if we sacrifice tens of millions of human lives worldwide in order to minimize short-term economic disruptions.
“But surely the economy (whether people have jobs, can keep a roof over their head, basic supply chains of food & medicine functioning) is needing consideration!”
— Brian McGill
Agreed. And I’m not adverse to all economic analyses. For example, I greatly appreciated the Stuart Pimm et al. “Can We Defy Nature’s End?” 2001 paper with supplementary information that attempted to quantify the cost of preserving a certain level of biodiversity.
Economic PREDICTIONS, though, add additional layers of uncertainty onto the uncertainty of present estimates, and the result is often much more imprecise than desired — yet acted upon by politicians with no appreciation of error bands. Perhaps even worse, both economic predictions and their interpretation can be strongly biased by ideology, as well as just general “wishful thinking”. No surprise there, since even objective “science” can be so influenced. But with economics, I’d say it’s more often and to a greater extent.
All of the above is with regard to “honest” misprediction or misinterpretation. Haven’t even touched on the issues from the standpoint of today’s (March 25) Jeremy Fox column here on “scientific fraud vs. financial fraud”, which makes things even more complex, but again, I believe affects economics much more than science.
Nor does any of this consider the points well-made by commenter Thomas J Givnish, just above, regarding the “valuation” of human life, and public health restoration as precondition for economic restoration.
Twitter thread suggesting that the initial, subsequently discarded UK strategy was too reliant on previous research and plans tailored to other diseases, and was insufficiently responsive early on to new information coming in from other countries:
I would imagine that postmortems on this are going to go on for a long while…
If I didn’t mention it upthread, Carl Bergstrom is another very good Twitter follow for the science of the COVID-19 epidemic. Here for instance is a good informed discussion of why different epidemiological models tend to sort roughly into “best case” and “worst case” models, without much of a middle ground:
Pingback: Open discussion thread: field-based courses in the time of coronavirus | Dynamic Ecology
I don’t know if this thread has died but I want to bring up one issue that is much discussed here in Sweden. One reason why Swedish policies are less restrictive than many other countries is the worries about the second or third wave of the epidemic. It is clear that we will not be able to eradicate the disease and vaccination seem to distant. Therefore it is likely that the disease will come back when restrictions are reduced, and more strongly so when restrictions were very hard in the first place. I read somewhere that a Harvard study suggested that one should aim for 20-40% infection during the first wave in order to reduce these risks. I have not been able to locate this study but it would be interesting to hear epidemiologists view on the more long term strategies and not only on how to reach very low infection during the first wave.
Smart commentary on the new Reuters article about the UK government’s early use of scientific advice around the coronavirus pandemic: https://twitter.com/jana_bacevic/status/1247815492850585601
Marc Lipsitch makes an important point:
“A lot of the confusion, in general, is premised on the misunderstanding that if you control the epidemic once, then you’re done,” Lipsitch said. “There’s no reason to think that.”
In my opinion, this is very relevant to discussions about fall teaching plans.