Facemasks in Germany

August Torngren Wartin pointed us to this article, “Unmasked! The effect of face masks on the spread of COVID-19,” by Timo Mitze, Reinhold Kosfeld, Johannes Rode, and Klaus Wälde, and asked what I thought.

My reply: I’ve not looked at it in detail but it seems reasonable.

I’m sharing this for a few reasons. First, we report on a lot of bad things that social scientists do, so it’s good to report on good things. Second, when we hear of new research, it’s common that we just look at some combination of the title, abstract, and press release. Quick reactions are important too.

170 thoughts on “Facemasks in Germany

  1. These studies now all assume we want to “slow the spread”. At first the reason was to do it for a few weeks to avoid overrun hospitals but now that seems to be a goal of its own.

    The longer this goes on the worse the side effects are going to be.

      • I dunno about fail, but Ive yet to see a single report of any obese, aged, or diabetic volunteer or animal vaccinated then challenged. Or even cell culture looking for ADE where it is likely.

        So Id bet these vaccines arent going to be very useful for the 50%+ of the population at risk of severe covid.

        And immunity to all sorts of stuff is weakening as time goes on. We dont know how long it takes but 6 months is already getting up there. Wait long enough and its going to be like first contact between Europe and the Americas.

        • Then there is of course the even rapider centralization of economic and political power into the hands of the very politicians, banks, and corporations responsible for the problems with inequality we already had.

      • Let’s say two years from now there’s no vaccine, the virus has made its way through 30-40% of the population and 5,000 people a day world-wide are still dying…

        …are you still going to be in favor of disrupting the worldwide economy and the daily lives of billions of people just to slow it down and buy some time so that a safe and effective vaccine can somehow emerge?

        What’s the stopping rule for these emergency, time-buying, curve-flattening, maybe sorta kinda effective measures?

        • “Maybe sorta kinda effective” is wrong, at least for the more extreme measures. Compare Sweden to its neighbors, compare…well, look, you can read the news as well as I can. Shutting everything down is very effective at reducing the spread of the virus.

          To repeat something Daniel Lakeland pointed out several months ago, there is no alternative to “disrupting the worldwide economy and the daily lives of billions of people.” It turns out that when people are extremely sick they don’t or can’t work. When people have someone at home who is at death’s door, they drop everything else and take care of them. When realize that there is a substantial risk of getting a potentially fatal disease if they go to a restaurant, some of them will not go to a restaurant. To use Sweden as an example — the only country I know of in the ‘developed world’ that did not do some kind of serious shutdown — according to Business Insider https://www.businessinsider.com/coronavirus-sweden-gdp-falls-8pc-in-q2-worse-nordic-neighbors-2020-8 “Sweden’s GDP fell 8.6% during the second quarter of the year, according to its statistics body. The fall is sharper than its neighbors — Denmark registered a 7.4% fall, and Finland a 3.2% fall. Statistics suggest Norway also fared better than Sweden.”

          I think whether or not an effective vaccine is forthcoming in the near future (by which I mean the next year or so), we are going to see people and businesses adapt. We are seeing this already. Transmission probabilities are much lower outdoors, so we will see more business happen outdoors. People will have more things delivered and will go out to stores for fewer things.

          But I could be wrong. One of the things I’ve found flabbergasting is the extent to which a lot of people, including rule-makers, seem to mostly think in terms of two modes: everything is shut down, or everything is opened up. Couple that with a refusal to learn from other people’s experience, and we’ve seen a bunch of states that refused to shut down, insisting on keeping the economy going…until people started dying by the hundreds or thousands. I still can’t get over De Blasio encouraging people to go out and enjoy a night on the town, and the governors of Florida and Texas and Georgia. And of course the President of the United States. All of them seem to think that “the only thing to have is fear itself” as if the virus only affects people who are worried about it.

        • “Shutting everything down is very effective at reducing the spread of the virus.”

          Lots of talk about this but as far as I saw there was never a time when “everything” was shut down, or even most things. Most of the retail establishments that shut down are small businesses, except a few chain restaurants. Otherwise operations were and still are more or less normal. Offices have some people off site, stores have stickers on the floor to show you where to stand at the checkouts.

          So for the most part what the “shut-downs” actually prove is that it’s safe to operate most businesses with modest modifications. Bars and large entertainment venues are the main concerns for transmission.

          Schools are still a problem but they could probably be operated safely with modifications to schedules, additional staff and ventilation, but this would require a fair amount of cash.

        • Where do you live, Jim? Here in the SF Bay Area the month of March was nothing like you describe, not remotely like normal. Even now it is nothing like normal, although the restrictions are much less severe: there is still no indoor dining, retail shops allow very limited numbers of customers at a time, etc.

          But in any case I think it’s clear that the ‘shutdowns’, however incomplete they have been, have indeed been effective at reducing the Rate of spread of the virus.

          Other than that one statement, though, I agree with you: most businesses can adapt to keep operating without crippling disruptions. Exceptions are restaurants, theaters, sporting events…places that rely on having lots of people in close proximity for a long time.

        • Phil said: “Where do you live, Jim?”

          Seattle area. I commented on that below, sorry. Right now everything but entertainment venues are open here and transmission rates are falling still.

          My guess is that the problem in CA is similar to the one in WA. Here the state and county hotspots remain in low-income/agricultural areas with high concentrations of non-English speakers, living in large family groups. Much has been said about the fact that these folks have been forced to work in conditions that promote transmission, but if you’ve ever been inside an apple packing shed, you wouldn’t describe it as a small, crowded building with poor ventilation.

          This wouldn’t be the first time that cultural practices contributed to the spread of disease. Ebola as I recall can be transmitted through feces. I read that the initial outbreak of Ebola, in the 80s I guess – and the more recent ones too for all I know – were exacerbated by the cultural practice of cleaning people’s bowels before burial, and it takes some serious effort to get people to stop doing this.

        • My understanding is that, with few exceptions, pretty much everything really did shut down in Tokyo, and the number of cases fell to near zero.

          Japan stopped doing the shutting down thing, and daily cases are now twice what they were at the worst of the first wave. (Although this time around it’s mostly younger folks, so the hospitals aren’t being swamped. So far.)

          But in Japan, everyone wears masks and no one’s overweight.

          (Japan pretty much only tests people who show up at a hospital with symptoms, though.)

          But yes. By now we pretty much know what the problems are. (Kids being more effective vectors for transmission than was thought, though, is going to change things. And, from experience, it looks like colleges aren’t going to be able to reopen.)

          One thing I’d like to add here is that a stint in the ICU is something that has long-term nasty repercussions whatever the disease. Being immobilized for two weeks (they tried it on college student volunteers) is something that at least some of those volunteers fully recovered from. And people who “recover” from things that sent them to the ICU have a plethora of problems, including mental illness and reduced cognitive abilities. Being sick is bad for you.

          What this means, is that the idea that COVID is something that’s not all that bad for folks who don’t die is extremely wrong. I had a nasty flu spring 2019, and my gut still occasionally gets wonky. To the point that I have had to cancel the day’s activities. (My PCP says it was probably a bug brought by a tourist, so the flu shot didn’t work.)

          So the bottom line is that your quality of life is going to be lower, perhaps permanently, if you come down with any COVID symptoms. Given that, what do you want your government/society to be doing about it?

        • Oops.
          “at least some of those volunteers fully recovered from”
          =>
          “at least some of those volunteers NEVER fully recovered from”
          Oops.

        • So the bottom line is that your quality of life is going to be lower, perhaps permanently, if you come down with any COVID symptoms. Given that, what do you want your government/society to be doing about it?

          1) Do not create guidelines advising to use the most dangerous and expensive interventions without evidence.

          Tips for managing respiratory distress

          Keep SpO 2 > 92–95%.

          Do not delay intubation for worsening respiratory distress. Be prepared for difficult airway!

          https://www.who.int/publications-detail/clinical-care-of-severe-acute-respiratory-infections-tool-kit

          It was something that was heard and discussed without a clear source which is why we did not reference it formally. It was found in several guidelines such as the DOD, without reference.

          https://mobile.twitter.com/ThinkingCC/status/1271088533088935936

          2) Fund studies into promising safe, cheap interventions that make perfect sense for a disease characterized by elevated oxidative stress and refractory hypoxemia, ie vitamin C and HBOT. It is now august and finally the n = 20 studies are coming out and both are independently showing 50% decrease in mortality:

          https://www.researchsquare.com/article/rs-52778/latest.pdf

          https://www.uhms.org/publications/uhm-journal/uhm-journal-ahead-of-print-public/hyperbaric-oxygen-therapy-for-covid-19-patients-with-respiratory-distress-treated-cases-versus-propensity-matched-controls-2/download.html

          Instead the FBI has been raiding clinics offering IV vitamin C and the FCC is threatening people clinics that offer HBOT to not talk about the successes they are seeing.

        • I agree that the ICU is very likely to have bad repercussions.

          But saying that everybody who gets *any* COVID symptoms is going to have long-term diminished quality of life is a *much* stronger statement… especially in e.g. the young adult demographic, as ICU admission is very rare here. (The Theodore Roosevelt had, I believe, 1 ICU case out of over 1100 infections.)

          It might be a true statement! (Probably not for everybody, but a meaningful proportion of mild cases.) It’s possible… but aftereffects of ICU stays aren’t at all evidence for it.

        • Confused says:

          “But saying that everybody who gets *any* COVID symptoms is going to have long-term diminished quality of life is a *much* stronger statement… especially in e.g. the young adult demographic, as ICU admission is very rare here. (The Theodore Roosevelt had, I believe, 1 ICU case out of over 1100 infections.)”

          Well, I didn’t intend to imply “everybody who gets *any* COVID symptom” will have long-term consequences. What I _think_ is true, though, is that we will find that the number of medium term, long term, and permanent consequences (that noticeably impact quality of life) will be _at least an order of magnitude higher_ than the number of deaths. Meaning that this is something you really don’t want to catch, whatever your age.

          My reading of the Theodore Roosevelt wiki article is that of the cases treated on-board, there were two ICU cases (an early patient who died quickly and then the article, discussing a later period, mentions one current ICU case) and 7 hospitalizations, but that some number of patients were moved off the ship, and we don’t know their details. And we don’t know the details for the latter part of the run of the disease, since the Navy stopped reporting.

          Navy sailors are some of the most fit, most healthy people on the planet. Thinking that your disease will fit that profile is somewhat overly optimistic, I’d think.

        • >>Well, I didn’t intend to imply “everybody who gets *any* COVID symptom” will have long-term consequences. What I _think_ is true, though, is that we will find that the number of medium term, long term, and permanent consequences (that noticeably impact quality of life) will be _at least an order of magnitude higher_ than the number of deaths.

          I think that sounds pessimistic… I think we’d be hearing more about medium-term impacts from places where tons of people have been over COVID for months (like the Northeast US)… but it’s far more believable than “everyone”.

          An order of magnitude would still mean that the vast majority of young people who caught it wouldn’t have significant impacts to quality of life. (If death rate is say 0.

          >>My reading of the Theodore Roosevelt wiki article is that of the cases treated on-board, there were two ICU cases (an early patient who died quickly and then the article, discussing a later period, mentions one current ICU case) and 7 hospitalizations

          I think the one death never went through ICU, wasn’t that person found dead in their room? So one death and one ICU case, but not the same person.

          >>Navy sailors are some of the most fit, most healthy people on the planet. Thinking that your disease will fit that profile is somewhat overly optimistic, I’d think.

          Yeah but I was talking about risk to *young* adults, college or a bit older, say 18-29 age group. The one death on the TR was age 41. Still ‘young’ from some perspectives… but for COVID 41 is very different than 25.

          I can’t see IFR in people in their 20s being even 0.1%, no. TX (my state) has 79 reported deaths in the 20-29 age group. Even if the real number is more like 120 due to deaths not yet occurred/reported… there’s just no way that Texas only has 120,000 infections in this age group!

          (TX has 577,000 *confirmed cases*, certainly millions and millions of infections… Given that the recent surge was concentrated in this age group, especially at first…)

        • Oops, not death rate 0.

          That incomplete sentence was meant to be “(If death rate is say 0.05%, 1 in 2000, then an order of magnitude worse is 0.5%, 1 in 200”. (And I think 1/2000 is way too pessimistic for this age group or the US would have seen far more deaths in this age group than it actually has.)

        • Phil said: “Where do you live, Jim?”

          One thing I overlooked: malls and some other chain stores (clothing) were shut down, as evidenced by several bankruptcies.

          Seattle area. I agree the first week was weird. After that sub-normality was pretty much the story. Traffic was light – very abnormal – but in stores that were open things were more or less normal. Now everything is open except entertainment venues. I don’t go to bars and restaurants but I see they are open and serving, indoors and out.

          I agree the shutdowns dramatically slowed transmission. However, they did it at severe cost. And they mostly *did not* achieve the more important goal: figuring out how to reduce transmission and operate as close to normal as possible. It has taken several more months to convince people that the primary mode of transmission is aerosol, despite some pretty clear evidence early on. So that was a pretty big failure.

        • Jim,
          Now I’m really confused. You’re telling me that in Seattle everything is open and things are more or less normal, but you’re also saying the shutdowns did not achieve the goal of allowing operations to be as normal as possible. Unless you’re really quibbling — yes, things are ‘more or less normal’, but they aren’t ‘as normal as possible’ — those seem exactly contradictory.

          If your area’s shutdown, however complete or incomplete it was, has enabled your area’s life to return to something fairly close to normal without causing runaway transmission again, that’s a success. How on earth are you calling it “a pretty big failure”?

        • Phil, I think some people consider what’s going on now as still “a shutdown” and others see it as “basically normal” and “shutdown” only means literally people are mandated to stay at home.

          This is basically a fruitless discussion. Let’s just create new unambiguous words:

          1) Shutdown: actual mandated stay at home, you can be fined for leaving your house.

          2) NPI: mandated masks, social distancing, deliveries, outdoor services, etc

          3) Normal: people go around hugging each other without even thinking twice, everything is open, no one mentions covid.

          It’s very clear that we started with 1 and moved to 2, we’re nowhere near 3

        • ‘How on earth are you calling it “a pretty big failure”?’

          Because it could have been done in a much shorter time with much less damage. It’s now late August. The main change in policy since April 1 is the mask mandate. How could that not have been done on April 1? Again the evidence was there.

          It’s a massive failure. People’s life work has been destroyed by unnecessary restrictions.

          I supported the initial lockdown. It was absolutely necessary. But nothing was done during 10+ weeks of lockdown except useless modelling. It wasn’t until R0 started blowing the roof off again that leadership became so desperate for a solution that it started looking at mask use.

        • Daniel, you should add:

          2.5) NPI: mandated masks in public places; social distancing outside the household; entertainment venues closed.

        • And FYI, I’m speaking from the perspective of what I see as beneficial for broader society. For myself, I was well positioned in the stock market and the shutdown was *highly* beneficial financially.

        • Jim, Phil,

          It seems like the two of you have different points of comparison:

          Phil looks at what could have happened if we’d done really nothing… and says what we did do is a lot better than that.

          Jim looks at what the optimum is and says what we did do is a lot worse than that.

          I think both of you are right. What we did is a lot better than nothing, but a lot worse than we could have done.

          There is MUCH room for improvement, but without coordination it will not happen, and coordination is not there. So we will continue to have enormously more suffering than is needed.

        • “What we did is a lot better than nothing, but a lot worse than we could have done.”

          Excellent summary Daniel. I’m in full agreement.

          There are many areas still in need of improvement, but after six months of touting ineffective pre-conceived solutions, people are finally throwing away their pre-conceived ideas about transmission and impact and actually dealing with reality and really making progress.

        • Maybe my disagreement with jim is a matter of degree; a quantitative difference rather than a qualitative one.

          Initial shutdowns were needed. The only problem was that in too many places they didn’t happen early enough. Washington State was one of those. About six weeks ago I read an article about the Washington pandemic response in which the state epidemiologist said all of the decision makers saw it coming but realized that they had to wait until people started dying before people would be willing to accept government mandates. In retrospect they should also have focused efforts on retirement homes and nursing homes, and that probably should have been more obvious at the time, but the article did make me understand why they didn’t act sooner. It may not have been the right call but I’m no longer baffled about it.

          But I am baffled about the extremely slow efforts to increase testing capacity. I know Daniel L has been talking about this for months too. WTF. If there aren’t enough factories making cotton swabs, spend a few billion dollars and build or convert some factories. It’s amazing what you can do with a few billion dollars. Not enough PCR machines? Spend 100 billion dollars and build a factory to make more, not that it would cost that much but if it did we should have done it. But: this is not really a state-level decision. The federal response has been almost as bad as I can imagine, and the only reason I have to put the “almost” in there is that Venezuela is now rounding up sick people and imprisoning them as ‘bioterrorists’, proving it is possible to do even worse than us.

          And yes, people — including those advising governments — have been too slow to recognize aerosol transmission, and transmission by asymptomatic or very mildly symptomatic patients, and to adjust advice and requirements accordingly.

          We are gradually learning what can be done with reasonable safety, and how. I went to pick up a few things at a big-box store this afternoon and I was shocked at how many people were there. Parking lot nearly full, lots of people in the store. They were regulating the total number of people in there store at a time, presumably based on average people per square foot, but of course people were not uniformly spread throughout the store. I got a few of my items while my area was relatively empty, but suddenly it got crowded so I made a quick exit. The statistical question around this is pretty interesting. If the fraction of people who are infected is sufficiently small then what I saw today is OK. If it’s sufficiently large then it’s definitely not. My informed opinion is that what I saw today was in the gray area in between. Personally, I can pick up a new blender another time, or have one delivered, so I see no need to tolerate the gray area. I fear people are becoming too complacent (again).

          I’d give the federal government an F on coronavirus response, but from what I know about Washington State, where jim is from, I’d call it a solid B, maybe even a B+. They’ve done better than most states and a lot of countries.

        • “Maybe my disagreement with jim is a matter of degree; a quantitative difference rather than a qualitative one.”

          Hmm…yes in some respects. Surely we differ on our grades for the responding groups. You grade on a curve. :)

          I think there’s a lot more crow to be eaten by epidemiology regarding the mode of transmission. My guess is that we’re going to find aerosol transmission is a far more common form of transmission for all respiratory viruses than is even dreamed of now, and that the hand-washing mode of prevention for flu is mythology, adopted more for having something reasonable-sounding to say than from any evidence that it’s an important mode of transmission. I say that because when people finally build a physical model of transmission, it will show that R0 just can’t get much past 1.0 with “droplet” transmission, while with aerosol transmission it can climb high and fast.

        • To a degree, sure. Some disruption was unavoidable. But I really don’t think this level of disruption was unavoidable in many parts of the US. There’s enough difference in risk by age, and enough “division by age” in society (ie retired people are not going to workplaces, schools, colleges), that I don’t think 1918-19 levels of “people just dropping over dead” were ever plausible.

          >>the governors of Florida and Texas and Georgia.

          Eh, depends on what your goal is. I live in TX and have been following the situation very closely — TX was never *nearly* as close to overwhelming hospital capacity (yes, 100% of “normal capacity” in e.g. Houston, but nowhere *near* maximum surge capacity).

          Eradicating the virus is just not possible in these places — whatever the governors say, people will not follow measures well enough. So I don’t think doing more mitigation would really have done much good.

          And while the situation is serious in TX (looks like COVID will be the 3rd most common cause of death in 2020), it is not even *close* to what was seen in the Northeast.

          So — TX has made a decision to accept more risk, sure, but I really don’t think that will prove to be such a bad decision.

        • I think you need to make the distinction between *most parts of the US* vs *most people in the US*.

          CA, TX, NY, GA, FL, OH, and a few others make up a huge chunk of the people in the country, and within those states, most of the people are again within the confines of metropolitan areas.

          Sure you could in fact have not shut down rural areas and other things, but while you could have opened up say 80-90% of the square miles in the country, you’d still have had to include 70-80% of the people.

        • But I don’t think TX, including its metro areas, would have been that much worse if we’d done what Sweden did (maybe even a bit less, like left colleges open).

          We basically got the wave after reopening anyway, so the only benefit we got was the improvements in treatments. Which is *something*, sure, but I don’t think we would have seen overwhelmed hospitals no matter what.

          Same for FL, basically, and they have a much older population (median age is 42, vs ~38 US average, ~35 TX). So if *they* didn’t overwhelm their hospitals…

          The difference between Greater New York and, say, Greater Houston or Dallas-Fort Worth is gigantic. Considering all metro areas as one thing is very misleading IMO — especially since New York is really its own thing.

        • Overwhelmed hospitals is a definite very bad nonlinear problem. But cases are a problem in and of themselves in my opinion. Even “asymptomatic” cases seem to have significant potential for long term consequences, with heart injury, kidney injury, brain injury, minor strokes, and various other complications. I don’t have good data on this stuff so I can’t say, but “let them get sick” for people under say 30 is a bad idea in my opinion, yet many places people are essentially choosing that.

          Of course some people choose this on their own, and there’s something to be said for letting them make that choice, they can smoke too right? The big problem is the externalities, where people get sick who are trying not to. For example the students in the dorms who aren’t out partying it up, the parents of the middle school students… etc

        • Hmm, do you have a link or source to *asymptomatic* cases having long-term problems? The vast majority of those wouldn’t even have been diagnosed…

          I mean, I don’t know much about that side of it. But if it was *that* common shouldn’t we see huge numbers of people with those problems in parts of the US that have really high infection rates?

          I suppose I am a bit skeptical because it seems so unfalsifiable. You inherently can’t prove much of anything about the long-term effects of a new thing!

          And I’m not doubting the existence of long-term effects at all… of course there will be some. But an average flu season is maybe 30,000 deaths in the US. If we have 250,000-300,000 COVID deaths by the end of the pandemic, then the long-term effects “disability-adjusted life-years lost” would have to be significantly more than 8-10x greater than that of flu for it to change the overall picture of disease severity.

          (Which right now I would say is “far worse than seasonal flu, probably a bit worse than but very comparable to the 1957 pandemic, far less severe than the 1918 pandemic”.)

          I’ve seen one study that says the stroke risk caused by COVID is ~7x that of influenza, for example.

        • I agree w/ confused, the case for substantial long term effects hasn’t been made. Not saying it won’t be, but most of the discussion around it is NPR-ish anecdotes.

        • The case for asymptomatic transmission was based on three chinese guys back in January/February whose family members tested positive afaik.

          After that its been models (or even just expert opinions) that assume asymptomatic transmission. Eg, does anyone have more *evidence* for this happening:

          – Evidence:
          https://onlinelibrary.wiley.com/doi/full/10.1111/irv.12743
          https://ccforum.biomedcentral.com/articles/10.1186/s13054-020-2817-7
          https://link.springer.com/article/10.1007/s11427-020-1661-4

          – Models/assumptions:
          https://www.bmj.com/content/368/bmj.m1165
          https://science.sciencemag.org/content/368/6490/489
          https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0236976

        • Of course the long term effects haven’t been proven, but we have definite evidence that the mechanism is there which makes long term effects plausible enough that they can’t be ruled out. When you’ve got certain people who are otherwise asymptomatic and suddenly have strokes, which is a definitely thing, then there’s also potentially lots of other complications. We see virus in people’s brains, we see virus in people’s heart tissue, in kidney, in liver, etc.

          Will we see 2 to 5 years out people with kidney disease, or heart disease that wouldn’t have otherwise happened? We don’t know, but we can’t rule it out. Suppose it turns out that having this disease adds as much risk to your life as say smoking for 10 years, then it’ll be a public health disaster for 20-30 year olds.

          Will that turn out to be the case? maybe or maybe not, but there seems to be two ways people deal with this kind of uncertainty:

          1) The NHST style: if you can’t prove it we can rule it out as essentially 0

          2) The Bayesian style: there’s a lot of uncertainty, we must take into account all the possibilities in decision making, and there are mechanisms whereby some fairly bad long term consequences are made plausible.

          One of those mechanisms is the way people seem to actually do stuff, and the other is a good way to actually do stuff.

        • Also, I think there are at least 4 distinguishable questions/issues:

          1. What measures work at slowing/suppressing spread, and how well do they work?

          2. How much do you gain from slowing/suppressing spread if the hospitals are not going to be overwhelmed either way?

          3. Would there be large social disruptions from deaths/illness if the majority of people just went back to normal and accepted the risk? (IE – what is the “opportunity cost” of measures taken?)

          4. Is there a significant risk of long-term problems if young healthy people get mild COVID (as would the usual case in e.g. a college?)

          I’m mostly talking about 2) and 3). I don’t have the expertise for 1) and definitely not for 4).

          But on the social side of things… I just don’t think massive disruptions were inevitable no matter what we did. Some disruptions, sure, but I think there was a genuine trade-off made (IE, we could have accepted faster spread of the disease early on and had less social disruption).

          A lot of it is a question of weighing quality of life vs. quantity of life. And second-order effects (both of the disease and of social changes). And of life-years lost vs. lives lost, given the age distribution of COVID deaths. These are ethical questions, not really scientific.

        • @Daniel Lakeland: a lot of what you’re saying makes sense, but the strokes are IMO a really bad example, since flu does this too (as well as increasing heart attack risk).

          So the existence of COVID-caused strokes and heart problems doesn’t itself lead to revising an estimate that COVID is “kind of like flu but 5 or 10 times worse”, or “roughly comparable to 1957 flu pandemic maybe marginally worse” (during which we didn’t take anything *like* these measure).

          So I guess the more important question (to me anyway) is “do we have evidence that suggests long-term effects of COVID are /qualitatively different/ from long-term effects of other respiratory viruses”?

        • The case for asymptomatic transmission was based on three chinese guys back in January/February whose family members tested positive afaik.

          After that its been models (or even just expert opinions) that assume asymptomatic transmission. Eg, does anyone have more *evidence* for this happening:

          Also:

          First, it is unclear from the methods if a mixed-effects model was used to take into account the longitudinal nature of the data. If a mixed-effects model was not used, it is concerning that there was only a limited amount of data at day 0, because this may have resulted in leverage points that overly influenced the increasing angle of the average viral load trend line near the date of symptom onset. Importantly, we suspect that a large portion of individual viral load kinetic lines do not follow the general trend line and should be considered in the evaluation of the data. For example, one male subject (blue line, fig. 2, lower left1) began with a low but detectable viral load cycle threshold (CT) = 39 at day 1 and day 2 (CT = 39) before peaking at day 3 (CT = 28), which then remained elevated with viral loads of CT = 34, 34 and 35 at days 4, 5 and 9 post symptom onset, respectively. Several lines appear to have this pattern, but it is difficult to determine from the figure whether the estimated average trend fits the data well, as many profile lines were indiscernible. If individual peaks in viral load are found to mostly occur after symptom onset or after the first available viral load time point, then this may indicate that the current generalized additive model and derived average trend line do not fit the data well and that individual peak infectiousness may be more similar to previous studies that found throat swab virus titers peaking after symptom onset2.

          https://www.nature.com/articles/s41591-020-1046-6

          So yea, it is looking more and more like the main premise for general masking/lockdowns was never based on very strong evidence to begin with.

        • This is a bit like the blind men and the elephant.

          My perception of the situation in Texas is that they (you) did an effective partial shutdown, reopened after a couple of months with predictably disastrous results, and had to shut down again when cases and deaths started shooting up. Here’s the timeline: https://abc13.com/covid-19-texas-timeline-all-coronavirus-events-in-government-response-state-restrictions/6345759/

          To me, having to shut down again shows that restrictions had been relaxed too much, I.e. proved to be a bad decision. I feel like the jury is already in on that.

        • “To me, having to shut down again shows that restrictions had been relaxed too much, I.e. proved to be a bad decision.”

          Agreed.

        • We didn’t really shut down again, though. A mask mandate was added in most counties and bars were closed, but nonessential business closures and stay-at-home orders weren’t re-instituted.

        • “reopened after a couple of months with predictably disastrous results,”

          Absolutely. Why? Because nothing was done to change the situation during the lockdown. The government and the medical community sat on their butts modelling, hoping that would somehow change the outcome. And a lot of people supported just shutting down until when-the-heck-ever, because, well, they didn’t have any better idea.

          “To me, having to shut down again shows that restrictions had been relaxed too much”

          No, it shows that no one did anything to develop an effective policy. We have an effective policy now, but much of the information supporting that policy was available in April.

        • confused –

          You and I have been through this many times. The horse died and was cremated, and I just spread the ashes.

          That said, here’s another thing that I find interesting. You say:

          > But I really don’t think this level of disruption was unavoidable in many parts of the US.

          I recognize the conditionality of your statement there – but I think it isn’t enough. I mean you may actually think that, obviously, but *I think* that based on the evidence, the farthest you should be going is “I really don’t think that it’s impossible that this level of disruption was unavoidable…”

          In other words, *I think* that there are huge uncertainties here. What would the level of disruption have been if there was less in the way of NPIs, but as a result there was no stimulus, and people who stayed home from work because of concern about infection had gotten fired with no stimulus and no chance of enhanced unemployment checks, let alone any unemployment checks at all?

          I say we actually don’t have much of a clue. Counterfactuals are hard – they require a great deal of solid evidence – a standard which we can’t get close to meeting yet.

          And, we have some evidence form Sweden that absent NPI’s “disruption” in some ways are quite high nonetheless, and that in many ways the outcome of a government’s choice to implement NPIs or not is largely context specific and depends on particularly cultural components as well as, very importantly, the starting conditions that help explain why different countries (or states) make different choices. For example, the doubling rate in Sweden didn’t start our anywhere near the starting doubling rate in NY, well before NY implemented any NPIs.

          So, this to me looks like one of those questions about decision-making in the face of uncertainty, about risk decisions in the face of high damage function uncertainty. We need to be making decision in this context, IMO, not on a basis that we know anywhere near a precise calculation of the ratio of costs to benefits of different policy choices, but that we need to be deciding on the basis of (1) values and (2) the possibility of huge down side risks.

          Now you and I may disagree about where the largest down side risks might occur, and the discussion in that sense might be come recursive.

          But, I do think it’s interesting that people who are highly expert in evaluating probabilities aren’t really framing this discussion in a particularly useful way – as decision making in the face of uncertainty – but instead (IMO) focusing instead on a rather futile attempt to nail down the uncertainties at a level which could be really informative

        • Nobody wants to face uncertainty. It’s not a politically acceptable message.

          So instead the people who ought to know better just stake out a position prior to having any evidence whatsoever and tell the public with absolute (apparent) certainty that their pet intervention is necessary and to not do it will result in bodies stacked up like cordwood. As dribs and drabs of evidence appear, they will be either exaggerated or disparaged in service of sticking to the message. But at no point is there a possibility of admitting that the whole picture is still muddy and nobody knows what net effect all of this is having.

          Once their position is staked out, the actual messaging evolves. It can start from “flatten the curve”, move to “hold out until the vaccine” and from there whatever message appears to get traction when the previous ones losing their persuasive effect. But at no point will the message be, “We just don’t know what if anything will work so we’re spitballing here”.

        • Brent –

          > It can start from “flatten the curve”, move to “hold out until the vaccine” and from there whatever message appears to get traction when the previous ones losing their persuasive effect.

          So my reflexive reaction is to the political overtones of that comment of yours – with one standard argument being “Those libz started out advocating to “flatten the curve” but then moved to “crush the virus” when flattening the curve no longer satisfied their statist motivations.” (It reminds me of the “skeptics” argument that “libz started out worried about ‘global warming’ but switched to ‘climate change’ when the global warming message didn’t work.”)

          So in a sense, that argument can become another manifestation of the dynamic you describe: I think that while “flattening the curve” may have been a useful catch phrase, it was not every meant to deliver a message of “protect against hospitals being overwhelmed and then let the virus run free.” There is nothing incompatible with “flattening the curve” being one part of an overall strategy that includes “focusing on slowing the spread of the infection after the curve has been flattened.” Also, people can, and I would say should, adapt policies as we go along.

          That said, sure, the transition in focus could be a product of CYA, of people simply not wanting to say that they don’t know what will work and are instead spitballing. But just as people should be careful of shifting to different policies as a way to protect against saying they don’t really know what would work, people should also be careful about misrepresenting, or even failing to do due diligence to understand, arguments as a way to pursue an identity-oriented agenda.

          I think one key here is to avoid characterizing assigning guild by association here. We need to look at this as a collective problem of stakeholder dialog, IMO.

        • Brent,
          You’re right that politicians in the U.S. have done a bad job with the messaging. This is not surprising, and is one of the reasons the playbook for responding to public health emergencies has long called for politicians not to be the ones who brief the public. I am having an unusually hard time finding a web search that will bring up the articles I read a couple of months ago that discussed this issue.

          That said, I think a lot of the blame lies with The People. It’s not the politicians’ fault that “Nobody wants to face uncertainty. It’s not a politically acceptable message.” Or, rather, it’s not _only_ the politicians’ fault.

          You aren’t making quantitative statements so it’s hard to know how much I agree or disagree. But when you say “at no point is there a possibility of admitting that the whole picture is still muddy and nobody knows what net effect all of this is having,” I disagree…in both possible directions!

          On the one hand, I don’t think anyone is claiming to know _exactly_ what the effect of the various interventions has been. “How many people would have died if there were not a ban on indoor dining in the San Francisco Bay Area”…nobody says they have a good number for that. You’re fighting a straw man if this is what you’re railing against.

          On the other hand, if you’re not asking for a precise estimate, you would just like a general comment that nobody knows whether various combinations of interventions do very much to reduce the spread of the virus (by ‘interventions’ I mean reduced building occupancy, social distancing, mask wearing, ban on large indoor gatherings)
          then I think you’re not going to get that comment because it isn’t true. We _do_ know that these reduce the spread of the virus.

          How do we know? I think the best evidence is that many places have seen rapid growth in cases, sometimes classically exponential; have implemented some combination of these rules and gotten onto a downward curve within a couple of weeks; and more recently we have seen places relax the rules and get back on an upward curve within a couple of weeks.

          Of course, there’s also the fact that we know the mechanisms by which the virus spreads. This isn’t the Middle Ages, when we thought these things were caused by ‘bad air’ or ‘the influence of the moon.’ If you stand close to an infected person while having a conversation, you are inhaling virus particles that they exhaled, and if you do this for long enough there is a good chance you will get sick. If fewer people stand close to each other for long periods, there’s going to be less transmission. Are you disagreeing with that?

          And finally, let me turn this around: what’s YOUR plan? You aren’t happy with “kinda-sorta maybe effective measures” or however you put it earlier. You’d do nothing? Or you’d do more?

          And do you really think that even with no government measures at all, people wouldn’t be changing their behavior? What’s your plan to avoid disrupting the economy? We’re already over 200,000 COVID deaths in this country. How many would there be if people had (tried to) continue their daily lives? Obviously I don’t expect a precise answer, but how about a range? Maybe 500,000 – 2 million, with a best guess somewhere around 1 million? You tell me. And then tell me if you think that would not have caused people to radically change their behavior to the detriment of economic activity.

        • >>We’re already over 200,000 COVID deaths in this country. How many would there be if people had (tried to) continue their daily lives? Obviously I don’t expect a precise answer, but how about a range?

          Well, see, I don’t think it would be that different, because re-opening meant that places like the South and West *did* get hit. So I don’t think we gained much except time to obtain better treatment.

          That matters, sure, but I can’t see it making a 5x difference (1 million deaths vs 200,000). I can see 50%, *maybe* even 2x if you assume that most of the apparently lower IFR in the current South and West wave is due to improved treatments as opposed to differences in social contact patterns & demographics of who gets infected.

        • I agree we are probably beating the dead horse a bit, but what I was responding to seemed like a stronger claim that we know at least this level of disruption was inevitable. I can agree that “there are huge uncertainties here”.

          One comment, though:

          >>What would the level of disruption have been if there was less in the way of NPIs, but as a result there was no stimulus, and people who stayed home from work because of concern about infection had gotten fired with no stimulus and no chance of enhanced unemployment checks, let alone any unemployment checks at all?

          I’m not seeing why this would necessarily cause massive disruption. Wouldn’t most of the people who really needed the money not do that, i.e. keep going to work regardless? Especially as most of the workforce in the US is below the really high-risk ages?

          Of course this is a major *ethical* problem, but that’s a separate question from *social and economic* disruption.

        • confused,
          You seem to be trying to have it both ways: you think that in the absence of government action people “who need the money” would have kept going to work anyway, that people would have kept shopping, etc… and yet you think even a 2x increase in deaths in that scenario is an overestimate. This makes no sense to me at all.

          If you were arguing that government mandates weren’t necessary because people would have taken action voluntarily, that would make sense. That’s the Sweden plan. Sweden has suffered a lot of death and has had a major economic contraction, but they seem generally happy with the path they chose and one could argue that we could have done the same thing here. If this is what you were arguing I could imagine it. (I think there’s strong evidence that it would work even worse in the US than in Sweden, but it’s hard to know for sure).

          But to argue that in the absence of government actions most people would _not_ substantially change their behavior, but we would still have had less than 2x the number of deaths we have had…I don’t get it. We’ve seen what happens when people don’t change their behavior. In urban areas you get a doubling time of the order of 3-6 days. In non-urban areas you get doubling on the order of 10-20 days. How on earth could this correspond to less than a large multiple of the number of deaths we have seen?

        • >>How on earth could this correspond to less than a large multiple of the number of deaths we have seen?

          Largely because most of the deaths occur in the oldest age groups, who are usually retired. So I’m not sure how strong the impact of workplace closures on deaths (as opposed to infections/cases) is.

          Also, if there isn’t room for “a large multiple” of the number of infections we’ve seen before hitting herd immunity?

          I mean, what do you think the current infection prevalence in TX is? The covid19-projections.com model has it over 20%; that model assumes a rather low IFR, but TX has a fairly young population and much of this surge was driven by young adults, so…

        • confused –

          Two analysts from a right wing thinktank (via the WS Journal) (caps are not mine – I took this from a post on another blog…i don’t have paywall privileges at the WSJ).

          -snip-

          “labor demand retreated in a similar fashion across states, regardless of when a state imposed stay-at-home orders or how hard it was hit by the virus… Much of the job decline was in services industries. In many cases, it wasn’t closed schools that kept parents from going to work. It was consumers avoiding the retail and hospitality settings where people worked.

          A team of economists including Kosali I. Simon at Indiana University studied the labor market from mid-March to mid-April and found that employment fell by 1.7 percentage points for every additional 10 days that a state imposed lockdown orders. They estimate that 60% of the employment decline between January and April was caused by policy. BUT THE REMAINING 40% WAS LIKELY DRIVEN BY CONCERN ABOUT THE VIRUS.”

          “Economists at Goldman Sachs estimate that consumer spending is at 94% of its previrus level. That’s progress—but AN ECONOMY WITH A 6-POINT HOLE IN CONSUMER SPENDING IS DEVASTATED.

          “THE LOCKDOWNS ARE LARGELY OVER, and consumers are still nervous and skipping activities they view as optional or too risky. The Goldman Sachs analysis finds that the personal-care industry—for example, spas and nail salons—is still at less than half its previrus level of activity. Transportation, including airlines, is at less than one-third of where it was before the virus.”

          “Lockdown orders kept many people from dining out, but people now debate whether to stay in a hotel or visit the dentist based on their own concerns of getting sick. Even if risk tolerance is increasing, THE ECONOMY IS LIKELY TO HAVE A RECESSION-LEVEL GAP DRIVEN BY REASONABLE FEARS OF COVID.

          “States are unlikely to reimpose lockdowns no matter how widely the virus spreads, and the Trump administration won’t turn to a “stay at home” strategy ahead of the election. THE FATE OF THE ECONOMY THIS FALL AND WINTER DEPENDS ON CONTROLLING SPREAD. That will turn on prudence in personal interactions, more widespread mask wearing, and extensive testing and isolation to keep the virus under control. If we fail in these efforts, we’re going to face a lot of tragic death and disease from Covid, the recovery will slow, and THE ECONOMY COULD TIP BACK INTO RECESSION.”

          -snip-

        • That’s all true… but I don’t think we can separate “public fear” from “media response” from “government response”.

          If the WHO and CDC message from March 11 on (when it was declared a pandemic) had been “this is like 1957/1968 flu pandemic” the media response would have been different and public fear would probably have been less.

          NYC and Lombardy seem to have been outliers; from what I’ve seen in TX, I don’t think even a somewhat worse result would have been enough to inspire lots of fear from what people *were actually seeing on-the-ground*.

          So if you are talking about state/federal government measures… absolutely that’s right. If you’re talking about the whole response, including advisories, from early March on… not so sure.

        • confused –

          > but I don’t think we can separate “public fear” from “media response” from “government response”.

          You and I disagree about the inevitability of the media response. The media makes money from selling sensationalism because the public likes watching sensationalism. It is inevitable with the for profit media system we have and with the decreasing viability of a business model of straight investigative journalism.

          That is the real world. I see not much point in speculating about what might have been different if there were some completely different model for the media

        • That’s true to a degree.

          But I think the WHO and CDC response gave a lot more “credibility” to it, as did things like the Imperial College model (they ought to have known “2.2 million deaths” would be a headline… I think that model was too pessimistic for several reasons, but even it didn’t actually predict 2.2 million deaths in the US!)

          As did – ironically! – the efforts of certain politicians such as the current President to downplay it.

          If the President (presumably a different President!) had set expectations that this would be like 1957 or so, maybe 5 to 10x worse than an average flu season, rather than saying it would go away in the spring, things might have been quite different.

        • Further – I see the range for government response as pretty narrow. Every fiber in Trump’s body would be to dismiss the seriousness of the pandemic ala many of his followers in Qanon and the plandemic folks. He started out among that path until he saw that it wasn’t politically sustainable despite the sympathetic views in much of his base. Now he’s at his quick and decisive response saved 5 million moves.

          There are some baseline political realities. Even in Sweden, where the mainstream are loony lefties by American standards, and where the starting conditions left room for a lot more leeway, the political realities led to a response was significantly more extensive than your standard of comparison of a bad seasonal flu. And there was sifnifxnf political pushback even to that (although I the end it may have been sufficient politically).

          And in the US, despite that a significant segment of the public is constitutionally incapable of considering Trump fallible in any way, majorities think that our government response was insufficient even as you think it was a panic response. If Trump had anything less than a significant cult following, even our “panicked” response would be incredibly unpopular politically. Only a tiny majority of people have the knowledge base to coherently argue for a less “panicked” response, although a larger cohort can argue that we panicked because its a convenient way to hate on libz or because they like to see themselves as freedom fighters taking on tyranny.

        • >>Even in Sweden, where the mainstream are loony lefties by American standards, and where the starting conditions left room for a lot more leeway, the political realities led to a response was significantly more extensive than your standard of comparison of a bad seasonal flu.

          Well, my statement was 5-10x worse than an average seasonal flu, which means much worse than a bad seasonal flu (2017-18 was only ~2x normal), so obviously it would deserve more response than a bad seasonal flu.

          (I think 10x, ie ~300,000 deaths vs average seasonal flu of ~30,000, will be about right if we get a vaccine distributed in late winter/early spring, counting excess deaths. Reported deaths will probably be somewhat less, in the 200s).

          >>majorities think that our government response was insufficient

          Absolutely… but I’m arguing that is largely because of the late-Feb-to-mid-March response from WHO, CDC, academic groups like the Imperial College model, etc.

          I think the public perception could have been a lot different with the same facts, if health authorities had taken a somewhat different tack.

          Specifically I think the WHO focused on containment too long, rather than saying “use your flu pandemic plans”. Which the US does have, but which weren’t followed (and IIRC don’t call for anything like this). Sure, it’s not actually flu, but the effects and spread aren’t that different.

        • confused –

          > If the President (presumably a different President!) had set expectations that this would be like 1957 or so, maybe 5 to 10x worse than an average flu season, rather than saying it would go away in the spring, things might have been quite different.

          Again, seems to me that you’re operating as if these are some kind of established facts rather than an opinion that’s at a relatively extreme end of a range of uncertainties. I don’t dismiss the possibility that you’re right – but again my point is that there’s no way we can know with much of any certainty, and particularly with respect to the counter factual scenarios about how public response and incredibly complicated economics would have played out.

          This is about decision-making in the face of uncertainty, imo. It’s about risk. Trying to reverse engineer using counterfactual reasoning about such complex dynamics, with such high levels of uncertainty, involving many highly explanatory but poorly understood parameters seems to me to largely be a distraction (although I won’t deny that I engage in the same),

        • Yeah I agree there is a lot of uncertainty. I started out by arguing against what I understood to be a claim that the inevitability of social disruption at least this bad was effectively certain.

          It’s definitely about decision-making in the face of uncertainty.

          I tend to be skeptical of quick massive decisions, especially when the “upper-end” risk is pretty clear – I think that the information available in early March was already enough to make it clear this was significantly less severe than 1918 flu. (I was convinced of that by the time the WHO China mission report came out, late February I think).

          So beyond that position (which is more about values and risk acceptance than a factual thing), I’m not sure how much we fundamentally disagree.

        • Joshua and confused,
          Get a room.

          Just kidding. Actually I am finding your little debate interesting. I just want to point out that you can look at other countries to get some insight into the counterfactuals. You’ve already discussed Sweden, but you could look at Germany and Spain and so on. If the media in all of those countries covered the virus in similar ways, that bolsters the case that there was some inevitability to it here. If the media response differed but the economic response was similar in all of the countries then that bolsters the case that there was some inevitably to the economic response here.

          FWIW my feelings are much closer to Jushua’s. We saw what happened in Northern Italy, where dying patients were literally turned away from hospitals. We saw what happened in New York, where the number of severe cases was doubling every few days and tens of thousands of people died, including some famous ones and some who were not old (including health care workers). Many people, including me, decided “business as usual” would be stupid.

          Famously at this point, not everyone agreed. Then Florida and Texas and Arizona and Georgia and other places full of people who promoted business as usual started to have a lot of cases and deaths and to do partial shutdowns. We’ve seen what happens if people don’t change their behavior.

          Confused, do you disagree with that narrative?

        • When “confused” says that Covid-19 will be 10x worse than “the average flu”, doesn’t that figure already include the NPI measures that were taken? For a proper counterfactual, you need a model, and judging from past epidemics, with no interventions you get about half of the population infected, and then you squabble about the number of unreported cases and the true infection fatality rate.

          And “the average flu” is a bad thing! It’s kinda ok because it has a seasonal component and we have a partially effective vaccination that people at risk can take each year, and because large parts of the population are already somewhat immune against many strains of the flu, but actual influenza (not just your comon “influenza-like-illness” that many people experience) is a bad thing!

          The big point has always been (and well known to epidemiologists) is that it matters WHEN you institute NPIs, and with Covid-19, days mattered back in March. That’s because of the exponential growth. Getting public messaging out early, and doing a coordinated response may be the one main difference between, say, the UK and Germany, where the UK had a hospital system close to capacity, a long peak, and high excess mortality, where Germany had a shorter peak, was able to treat cases from other countries, and excess mortality being less than a bad flu season in most states (I’m qualifying this because only Berlin and Hessen participate in EuroMomo, but other states (NRW, Bayern) were hit harder.)

          It matters when and how decisively you go into the NPIs for their effectiveness.

          The German RKI says the evidence is still that droplet transmission (that prompted the physical distancing rules) is still the main mode of transmission. Aerosols have *always* been considered a possible transmission vector, see e.g. the evidence to hospital staff regarding aerosol-generating procedures such as intubating patients or having them cough, and the choir practice infections, but there is little evidence for that to occur except in indoor situations with inadequate ventilation where people sing, scream, or exert themselves. The German experts say that everyday-masks are effective because they reduce droplet transmission in situations where social distancing is compromised. (They also say that fomites seem to be a very minor vector.) There’s not a hard line between aerosols and droplets anyway, more of a continuum.

          The study done in Munich back in January/early February, published in Lancet on April 1st (but obviously pre-published long before) proved pre-symptomatic transmission, and a sister paper published in nature showed high viral loads at symptom onset, so aiming the interventions at preventing pre-symptomatic spread was based on the best evidence at the time, and I am not aware that today’s evidence has changed substantially. The fact that insummer time, with a pandemic on everyone’s mind, and symptomatic people having access to testing, many states and countries still have a hard time getting R to stay below 1 seems to show that an approach of “just isolate all the symptomatic people” will not stop the spread.

          From the information I am aware of, we need NPIs that adress asymptomatic and pre-symptomatic transmission, and the timing is critical, both for when they’re put in place and when they are relaxed. (And the latter is also clear to everyone who has looked at the Spanish Flu response).

          This is the framework that we had back in March that justified the NPIs, and in my opinion, it has been refined since then, but not overturned.

          The big problem that the US had was the slow rollout of testing compared to states that used the Chinese test or the German test recommended by the WHO, and that made it tough for many states to figure out where they were on the curve, and to hit the proper time for establishing NPIs and getting popular buy-in (with White House messaging not helping with that, either). That is what caught New York out. (The first big outbreak in NY was in Rochester, not NYC! Thinking “NYC is special” is not the whole truth!)

          * testing
          * timing of NOIs
          * build contact tracing capacity and use it to isolate potential cases
          * prevent large outbreaks

          Those are the cornerstones of a good public health response to Covid-19, in any state or country on Earth.

        • Mendel said,
          “It matters when and how decisively you go into the NPIs for their effectiveness.” and made a very convincing case for it. Thanks.

        • @Phil:
          >> Confused, do you disagree with that narrative?

          Not in terms of the broad picture of what actually happened, no.

          But there really is a pretty large difference between the bad results in southern metro areas like Phoenix and Houston and Dallas-Fort Worth, and the *extremely* bad results in NYC.

          I really do think NYC was fairly uniquely vulnerable (uniquely in the US — places like Lombardy and Madrid were also very vulnerable).

          IE, even if the response in Texas and Arizona was as bad as it plausibly could be, even if there had been no orders at all and public messaging not to change anything, I don’t think NYC-level death rates were a realistic possibility.

          @Mendel:
          >>When “confused” says that Covid-19 will be 10x worse than “the average flu”, doesn’t that figure already include the NPI measures that were taken?

          Well, sort of, but I don’t think they actually accomplished much in most of the US, because we’re going to get all the infections anyway!

          They delayed infections somewhat, so IFR will be somewhat less, but that’s about it. Maybe somewhat less total infections if we get a vaccine *really* soon, but I think mass distribution will take time even if approval is like October.

          >>For a proper counterfactual, you need a model, and judging from past epidemics, with no interventions you get about half of the population infected

          I don’t think so; flu pandemics are generally less, aren’t they? 2009-10 was 20% of the population. A vaccine was introduced late and reduced this somewhat, but the 2nd wave was already dying out before vaccination was available to the general public.

          >>And “the average flu” is a bad thing!

          Sure, it’s not as harmless as a lot of people think it is.

          But we’ve had flu pandemics much worse than seasonal flu (1957 and 1968) without much disruption to society.

        • Mendel thank you for that excellent summary and commentary.

          Confused, I agree NYC is special, with much faster transmission and thus higher R0 than probably anyplace else under business as usual. But any R0 > 1 will get you there eventually, and we’ve seen that many other metro areas will also suffer rapid transmission and R0 well over 1 if people don’t change behavior. We’ve also seen that a lot of people won’t change their behavior based on advice or requests or, well, rational thinking. Look at what happened in Wisconsin the first day they reopened bars: everyone crammed in to celebrate, with sadly predictable results.

        • “Look at what happened in Wisconsin the first day they reopened bars: everyone crammed in to celebrate, with sadly predictable results.”

          This is just a bad policy problem. Re-opening doesn’t mean everything has to be a free-for-all. Bars are open here in Seattle, but every bar that can has outdoor seating; masks are required away from tables and there are capacity limits.

        • >>But any R0 > 1 will get you there eventually,

          Well, yes… depending on what you mean by “there”.

          To the herd immunity threshold (or somewhat above it due to overshoot), yes, but lower R0 means lower herd immunity threshold too.

          To the point of overwhelming hospitals? I think not in the vast majority of the US.

        • This may explain our differences in point of view. I think you’re not understanding the math.

          Hospital capacity is much much lower than herd immunity. in that regime Reff > 1 produces cases per day that grows exponentially. eventually well before herd immunity you generate cases per day equal to your hospital capacity, doesn’t matter where in the US you are.

          You could argue that in much of the US R0 is less than 1 (like in the middle of Iowa corn fields) but that’s an argument about territory. most people in the US live in places where R0 is greater than 1. without NPI that brings Reff below 1 most people would be subject to hospital overwhelming

        • > places like Lombardy and Madrid were also very vulnerable

          There is nothing particular about Lombardy or Madrid compared to most urban areas in Europe apart from the timing and intensity of the first wave and how it was handled. If things had evolved differently we would be talking about other regions and cities.

          > To the point of overwhelming hospitals? I think not in the vast majority of the US.

          You didn’t think in May that cases, hospitalizations and deaths in Texas would see a tenfold increase such a short time, did you?

        • @Carlos Ungil

          >> There is nothing particular about Lombardy or Madrid compared to most urban areas in Europe

          Italy at least has an extremely high median age, so I think it is genuinely more vulnerable because of that. And aren’t there more multigenerational households, meaning probably more infections among the elderly?

          I agree it could have been some other region of Italy that was hit first, but I don’t think most parts of the world could have experienced death rates that high.

          Lombardy had a combination of extremely elderly population + poor medical understanding of the virus. Any place hit early on would have had the latter, but very few have the former.

          Madrid is the largest city in Spain, so yeah it may have been a matter of it being the first area to be hit hard, but it’s not *random* that it was the first to be hit hard. And Spain also has a rather high median age, though not as old as Italy.

          >>You didn’t think in May that cases, hospitalizations and deaths in Texas would see a tenfold increase such a short time, did you?

          In *early* May, immediately after reopening, I did expect a dramatic increase, though not to the point of overwhelming hospitals.

          By *late* May I expected that we would have seen the increase if it were going to happen, so by then I was thinking that seasonality + lower population density were in play and it wouldn’t happen.

          So yes, I was definitely wrong about the “time delay”!

          (Although hospitalizations didn’t increase quite *that* much. Actually 10x above early May, meaning 18000+ simultaneous hospitalizations, *would* probably have overwhelmed at least one major metro area, depending on how much emergency surge capacity could be increased.)

          @Daniel Lakeland
          >>This may explain our differences in point of view. I think you’re not understanding the math.

          It’s not understanding the math, but disagreement on the following:

          >>Hospital capacity is much much lower than herd immunity.

          I think this is not necessarily true.

          I mean, if literally 70% of the population was infected *at the same time*, yeah that would overwhelm capacity.

          But a) I think the herd immunity threshold in the vast majority of the US, including most major cities, is not nearly that high (there are no US cities like Iquitos, Peru); and b) even a 70% threshold doesn’t mean 70% infected *simultaneously*.

          A somewhat lower density means it spreads a bit slower so the % infected at one time is less for the same final % infected (and the final % infected will also be lower).

          I think some places in TX probably *have* gone to herd immunity (though reported deaths will continue to rise for a while due to lag between infection and death + a fairly long reporting lag).

          Also, the South surge saw more infections among young adults, so many fewer hospitalizations for the same number of infections. This is *very* different from what was seen in Lombardy or Madrid, or even the Northeast US, in spring.

      • Irrespective of whatever we think the optimal action should be, the silent change of strategy should be noted.

        We had the same in Finland in March – people went from “smooth the curve” to “eradicate the infection” quite fast, and I think these decision should better be conscious.

        • I agree, these decisions should be made and should be conscious. That said, I think this is the right decision.

          As Spain is learning right now, once you ‘flatten the curve’ you have to keep it flattened.

          I think one issue is that the set of collective behaviors that lead to R0 = 1.1 are not very different from the set that leads to R0 = 0.90, but these are radically different in the number of cases and number of deaths. Even if we did make a conscious decision to aim for a world in which every hospital was 90% full all the time — a perfectly flat curve, requiring R0 = 1.0000 — thus stopping the healthcare system from being overwhelmed, it would be really easy to overshoot that and find that there were 10% more patients than the system could handle, and then it looks like parts of Italy did six months ago: people who could otherwise be saved being sent home to die because there is no capacity to treat them.

          Trying to keep the number of daily new infections exactly the same is not realistic; you are going to end up either decreasing with time or increasing. But the latter is so much worse than the former, you really need to be in the regime of decreasing infections.

        • Not that I’m agreeing with your conclusions overall necessarily but I’m commenting on one specific point.

          The measure should definitely not be “decreasing infections” especially given how “infections” tends to be mis-interpreted as “positive tests”. The goal should be to decrease deaths and serious illnesses.

        • A positive test (or other positive diagnosis) is a “case”; I deliberately did not use that word. If what you’re saying is that reducing the number of positive tests should not be the goal, I agree. I never said it should be.

          I’m fine with the idea of reducing the number of deaths and serious illnesses as being the real goal, but I think it’s very hard to do that without decreasing the number of infections per day. If the number of new infections is increasing, it’s going to be hard to stop serious illnesses from increasing too.

        • I think talking about optimal Ro is futile. We (the society) cant set the Ro value at will like we set the tax rate or speed limits. We (epidemiologists) cant even estimate it well. Attempt to plan the epidemic – lets have this many cases and this many deaths – are not very sustainable.

          Lets talk a simpler question: when should we panic? Some people were panicking because hospital were going to be overburdened, and then they were panicking because number of cases is not declining, and then were panicking because change of infection is not zero. I think it is interesting and noteworthy that people move between these panic points unconsciously. This may be worth reflecting upon!

          As for me, I think the first thing is worth panicking, the second and the third — depends on the context.

        • When you say ‘panic’, is that really what you mean? I don’t think I’ve seen any panic at all.

          I’ve seen some overreaction, but it seems less common than underreaction.

        • Overreacting may be a better word, but Im still not sure what is the correct amount of reaction, so I dont know what would be an overreaction.
          By ‘panic’I mean a mental state, which may be appropriate to the circumstances.

        • Mikhail,
          Your English is excellent and I have no reason to believe it’s not your native language, but given the spelling of your first name I’m open to the possibility, so I’ll mention that “panic” doesn’t just mean fear, it means unthinking fear and behavior. Avoiding crowds, wearing masks, etc., are not signs of “panic” even if they are unnecessary. If someone is thinking about what they’re doing, they aren’t panicking.

          I agree, we don’t know the correct amount of reaction so it’s hard to know what is over- or under-reacting. Under that circumstance it’s pretty much inevitable that some people are going to overreact and some will under-react. It’s only extreme cases of either that I think we can call out as definitely non-optimal.

        • I kind of think that the possibility of “people who could otherwise be saved being sent home to die because there is no capacity to treat them” was only ever actually a possibility in a few regions, not most places.

          Italy was extremely vulnerable to that because its population is really elderly (median age ~45 vs ~38 for the US). And lots of multigenerational households, and (IIRC) not that many ICU beds per capita.

          Places like Phoenix and Houston didn’t overwhelm hospital capacity despite relatively limited measures… and at least in Houston it wasn’t even that close. (Sure, they were at “100% normal capacity” but nowhere *near* 100% surge capacity.)

        • When a thing has a potential to double in 3 days, “no where near surge capacity” means something like “never got above 10% of surge capacity” which they very much did.

          Of course, the doubling every 3 days thing stopped once people started widespread social distancing and soforth. So you might say the potential now is more like doubling every … 7 days. That still means probably “never got above 25% of surge capacity” which it did.

        • Sure, if it was doubling every 7 days. But it wasn’t.

          Peak of TX current-hospitalizations according to DSHS data was 10893 on 7/22; half that was reached on 6/29. So 24 days, not 7.

          Even at the beginning of the current peak, it wasn’t. TX was stable between 1400-1900 current hospitalizations from mid-April to early June. It broke the previous record on 6/8 with 1935; double that wasn’t reached until 6/23, 15 days later.

          TX never saw anything like NYC or Italy rapid exponential growth.

        • Texas took quite a few early steps and the growth rate was very low. Then they opened things up a lot and the rate skyrocketed (predictably) in spite of some restrictions still being in place. Then they clamped down again.

          You see, to be arguing something close to “in the three weeks that restrictions were loosened those places didn’t get overwhelmed, therefore if they had continued indefinitely with loosened restrictions they wouldn’t have been overwhelmed.”Is that in fact what you’re saying? If so, I strongly disagree.

          If you’re saying that in some areas in which few people have many interactions per day he virus can be kept under control with relatively few restrictions then I agree.

        • “Skyrocketed” is rather extreme I think. The rate was flat (R basically 1) for a couple of months, and then it went above 1. There was exponential growth for 5 or 6 weeks, but slow exponential growth, nothing like what was seen in the Northeast.

          And restrictions were loosened for way more than 3 weeks, too…

          And yeah, I really don’t see e.g. Houston getting overwhelmed to the “turning patients away” point regardless of government actions. If masks make a 40% difference in Germany (as the article in the post says) it’s going to be notably less in Houston since mask compliance is going to be worse (that’s what Texans are like).

        • Texas went from about 1000 cases per day in the beginning of June to about 10,000 cases per day in the middle of July. I suppose one could argue that the term ‘skyrocketing’ should be reserved for a factor of 10 in 3 weeks, not a factor of 10 in 6 weeks? Mmmmaybe. I think ‘skyrocketing’ is appropriate here, but I won’t quibble about it.

          You say “I really don’t see e.g. Houston getting overwhelmed to the “turning patients away” point regardless of government actions.” Maybe. I think it’s clear that without significant behavioral changes the medical system would have been overwhelmed, but maybe those changes would have happened anyway. People are certainly capable of making their own decisions and thereby changing their behavior voluntarily. Obviously you get the same public health effect whether the reason people stop hanging out in close proximity is voluntary or involuntary.

          But here in the U.S., several governors have claimed that either they weren’t going to do mandates in the first place, or that they were not going to reimplement them once their state reopened, and in all the cases I know of (I’m thinking of Texas, Florida, and Georgia) deaths eventually started to climb again and they later relented and imposed restrictions.

        • I was looking at hospitalizations, not cases, since we were talking about overwhelming hospital capacity. The growth was somewhat less dramatic there; it was basically flat around ~1700 current hospitalizations for 2 months, then grew to a peak of about 10900. So not a 10x increase in 6 weeks, more like 6.5x in 7 weeks.

          Yeah the daily case numbers increased faster, granted.

          Yeah, I wasn’t talking about literally *no* action even on the individual level, obviously people will do things individually, businesses will make changes, etc. regardless of government action.

          Although… to some degree it depends on how seriously you take the ideas about heterogeneity reducing the herd immunity threshold.

          TX reimplemented *some* restrictions, but nothing like what was in place in April! Just masks and closing bars, when in April all nonessential businesses were closed and stay-at-home orders were in place.

          If you didn’t go to bars a lot (and I didn’t), the current TX orders wouldn’t really keep you from doing most of the things you’d normally do.

          I’m being careful due to being somewhat germophobic to start with (somewhat ironic, because I’m *personally* concerned despite knowing I’m quite low-risk, but am still skeptical of a lot of the actions taken). But I could basically live my normal life now, except wearing a mask.

        • Looking at hospitalizations is good, at least until they get close to hospital capacity and that is starting to affect admissions decisions. OK, ‘rapidly increasing’ instead of ‘skyrocketing’ if you prefer.

          I am skeptical of the prospect of herd immunity, even among the more infection-prone populations, until the fraction of people who have previously been infected gets very high. Look at Spain, for example: one of the hardest-hit countries, with more fatalities per capita than the U.S. (although we may yet surpass them), but they are now experiencing a second surge of cases and fatalities. https://www.worldometers.info/coronavirus/country/spain/

          Lord knows I hope I’m wrong about this, but I think that reaching herd immunity would require infecting so many people that we would have hundreds of thousands more deaths.

        • confused, Phil –

          > Although… to some degree it depends on how seriously you take the ideas about heterogeneity reducing the herd immunity threshold.

          >> I am skeptical of the prospect of herd immunity, even among the more infection-prone populations, until the fraction of people who have previously been infected gets very high. Look at Spain, for example:…

          I think what’s important is that the to the extent that the HIT depends on heterogeneity to be lower, heterogeneity is largely dependent on mediators/moderators that vary by specifics of communities and in particular behaviors. For example, the HIT could be largely contingent on whether and to what extent NPIs are implemented. Mandating masks or staying away from bars could materially impact the HIT because those behaviors change heterogeneity.

          Like Phil’s reference to Spain, there are quite a few examples pf communities where population infection is quite high – as high as 60% or higher. Despite that modeling by respected modelers can show a HIT of between 10%-20%, I”m skeptical about the ability to predict the HIT is particularly in some kind of generic fashion.

        • Phil, herd immunity from vaccination is a real thing… herd immunity from getting the virus is just code words for “let everyone get sick”.

          The only way you get herd immunity without a vaccine is that everyone who would have gotten sick if we didn’t do anything… gets sick…

          It’s like deciding the best way to have peace in WWII is to just let the Nazis win.

        • I don’t think the Nazi comparison is at all fair, even beyond the “emotional”/rhetorical aspects of bringing up the Nazis; no country just lets an invasion happen, but plenty of disease outbreaks in history have ended naturally. Even the last few flu pandemics (2009, 1968, and I think 1957), in which vaccination existed, had infected a large percentage of the population before vaccination became available.

          And the lasting impacts of a pandemic like this are not at all comparable to a military conquest.

          As for Spain… yes Spain has (somewhat) more fatalities per capita than the US, but the IFRs are likely very different (population age structure, and Spain’s cases were mostly earlier when treatment & management of COVID patients were worse).

          So Spain might still have fewer people infected, though more deaths.

          Also, the herd immunity threshold in Spain might well be different than in the US.

          I’m not saying the US has hit herd immunity — I think it mostly hasn’t, though some high-infection-rate places may well have — but I don’t think Spain is informative one way or the other.

          But I also don’t think the threshold for the US is going to be ~70%. So I do think that there is no way the US was going to see 1 million+ COVID deaths this year. (I think the Imperial College model from March both had too high a threshold and didn’t account for decreases in IFR as the disease became better understood.)

        • @Joshua: yes, NPIs would further lower the threshold. I’m sure that’s helped.

          But wouldn’t the same principle also imply that significant differences in R0 and thus herd immunity threshold would exist between high-density places with lots of shared housing and mass transit, vs. low-density places with more single-family houses/less multigenerational households and no or practically no mass transit, etc.?

          That’s why I’m really skeptical that ~60% infection in Bergamo, Italy or ~70% in Iquitos, Peru can be extrapolated to the US as a whole. Individual towns in the US with similar characteristics, maybe (and there might well be some in e.g. far south Texas or rural Arizona), but that’s definitely not the vast majority of the US population.

        • Daniel –

          > The only way you get herd immunity without a vaccine is that everyone who would have gotten sick if we didn’t do anything… gets sick.

          That could depend on how you’re using the term. Some people might use it to mean no one else gets infected, as you describe. But it is also used to mean reaching the “herd immunity threshold” which is also a somewhat vague term – but often seems to be used to indicate a point short of everyone (who isn’t vaccinated) getting sick.

        • confused –

          > But wouldn’t the same principle also imply that significant differences in R0 and thus herd immunity threshold would exist between high-density places with lots of shared housing and mass transit, vs. low-density places with more single-family houses/less multigenerational households and no or practically no mass transit, etc.?

          I’m no expert, but from what I’ve read, read yes, that would be a factor. Highly “connected” presume staying home storks be a factor, and in particular “super-spreaders” staying home and not going to churches or bars. If course, super-spreaders won’t know they’re super-spreaders and are likely not to stay home without NPIs.

          There could be many behavioral predictors for changes in heterogeneity.

          > That’s why I’m really skeptical that ~60% infection in Bergamo, Italy or ~70% in Iquitos, Peru can be extrapolated to the US as a whole. Individual towns in the US with similar characteristics, maybe (and there might well be some in e.g. far south Texas or rural Arizona), but that’s definitely not the vast majority of the US population.

          Extrapolating seems dubious to me. But your comment seems a bit simplistic to me. For example, lots o’ god-fearin,’ church-goin’ folks in Texas. Look at the issues in Korea with church services as super-spreading events. So communities in Texas might have *higher* HITS than other communities – ndeownsonf on the stage of the pandemic (how many people have some level of immunity)…

          But it seems the real wild card is susceptibility.

          https://www.theatlantic.com/health/archive/2020/07/herd-immunity-coronavirus/614035/

        • Gonna repost without the link… Google “Atlantic; a new understanding of herd immunity”

          confused –

          > But wouldn’t the same principle also imply that significant differences in R0 and thus herd immunity threshold would exist between high-density places with lots of shared housing and mass transit, vs. low-density places with more single-family houses/less multigenerational households and no or practically no mass transit, etc.?

          I’m no expert, but from what I’ve read, read yes, that would be a factor. Highly “connected” presume staying home storks be a factor, and in particular “super-spreaders” staying home and not going to churches or bars. If course, super-spreaders won’t know they’re super-spreaders and are likely not to stay home without NPIs.

          There could be many behavioral predictors for changes in heterogeneity.

          > That’s why I’m really skeptical that ~60% infection in Bergamo, Italy or ~70% in Iquitos, Peru can be extrapolated to the US as a whole. Individual towns in the US with similar characteristics, maybe (and there might well be some in e.g. far south Texas or rural Arizona), but that’s definitely not the vast majority of the US population.

          Extrapolating seems dubious to me. But your comment seems a bit simplistic to me. For example, lots o’ god-fearin,’ church-goin’ folks in Texas. Look at the issues in Korea with church services as super-spreading events. So communities in Texas might have *higher* HITS than other communities – ndeownsonf on the stage of the pandemic (how many people have some level of immunity)…

          But it seems the real wild card is susceptibility

        • confused – (Google “Atlantic a new understanding of herd immunity)

          > But wouldn’t the same principle also imply that significant differences in R0 and thus herd immunity threshold would exist between high-density places with lots of shared housing and mass transit, vs. low-density places with more single-family houses/less multigenerational households and no or practically no mass transit, etc.?

          I’m no expert, but from what I’ve read, read yes, that would be a factor. Highly “connected” presume staying home storks be a factor, and in particular “super-spreaders” staying home and not going to churches or bars. If course, super-spreaders won’t know they’re super-spreaders and are likely not to stay home without NPIs.

          There could be many behavioral predictors for changes in heterogeneity.

          > That’s why I’m really skeptical that ~60% infection in Bergamo, Italy or ~70% in Iquitos, Peru can be extrapolated to the US as a whole. Individual towns in the US with similar characteristics, maybe (and there might well be some in e.g. far south Texas or rural Arizona), but that’s definitely not the vast majority of the US population.

          Extrapolating seems dubious to me. But your comment seems a bit simplistic to me. For example, lots o’ god-fearin,’ church-goin’ folks in Texas. Look at the issues in Korea with church services as super-spreading events. So communities in Texas might have *higher* HITS than other communities – ndeownsonf on the stage of the pandemic (how many people have some level of immunity)…

          But it seems the real wild card is susceptibility

        • confused –

          I have another response to you hung up in moderation. If it doesn’t pass though, I’ll repost breaking it into chunks to find out what triggered the filter.

          Anyway…

          > So I do think that there is no way the US was going to see 1 million+ COVID deaths this year. ( (I think the Imperial College model from March both had too high a threshold and didn’t account for decreases in IFR as the disease became better understood.)

          They also screwed up because they didn’t account for building surge capacity.

          But your comment is misleading, because you’re quoting a high end projectionbsith no allowances for NPIs and behavior changes as if it was a non-conditional “prediction.”

        • @Joshua: yeah, because of churches, I could totally see some small towns in TX where basically everyone goes to the same church hitting really high infection percentages.

          But not in *all* the small towns, as well as *all* the cities, which is what you’d need to see something like 60-70% for the *entire TX population*. Others would be much more nearly spared, because the spread can be pretty random, especially with a disease pretty strongly affected by “super-spreading” amd in groups too small for it all to average out.

          >>But your comment is misleading, because you’re quoting a high end projectionbsith no allowances for NPIs and behavior changes as if it was a non-conditional “prediction.”

          No, the one with no allowance for behavior changes was 2.2 million for the US. That obviously was impossible (there’d be some individual behavior changes, business-level ones, etc. regardless of policy … which had already happened by the time the model results were published). So I wasn’t comparing to that at all.

          1.1 million was for “minimal mitigation” without stay at home orders… more like a Sweden approach than absolutely nothing.

      • Actually, here is the first data on a covid vaccine in people over 60:

        The antibody response in the two vaccines was quite similar, and they both showed the same trends (for example, lower neutralizing antibody response in the older patients, one-third to one-half, which will be something watch for when we get efficacy data).

        https://blogs.sciencemag.org/pipeline/archives/2020/08/20/more-pfizer-biontech-data-on-their-actual-vaccine-candidate

        ADE is expected where there are “low” quantity and/or affinity antibodies towards the virus. So this is roughly consistent with that showing up in the elderly (we don’t know what “low” would be for covid). It won’t show up until they are challenged with virus though, that is why they should have done these studies in aged animals first.

        • Note that a lower level of antibody response, which might not protect against getting the disease, might still lead to a milder case if one does get it. Which is beneficial.

        • The risk factor for ADE is low quantity/affinity antibodies. That is the safety issue that came up with SARS, and has not been addressed at all for SARS2. We expect it to appear:

          1) In those with weak antibody response such as the elderly, obese, and diabetic (low quantity and/or low affinity)
          2) After antibodies wane (low quantity)
          3) When challenged with a virus with similar but nonidentical epitope (low affinity)

          Worst case scenario is combine all three. Old people get vaccinated, the antibodies wane after some time, then get exposed to SARS3 or a mutated SARS2. Now the same could occur after mild illness, but the vaccines are specifically targeting the RBD of the spike protein which is exactly the most dangerous epitope.

          To evaluate the efficacy of existing vaccines against infection with SHC014-MA15, we vaccinated aged mice with double-inactivated whole SARS-CoV (DIV). Previous work showed that DIV could neutralize and protect young mice from challenge with a homologous virus14; however, the vaccine failed to protect aged animals in which augmented immune pathology was also observed, indicating the possibility of the animals being harmed because of the vaccination15. Here we found that DIV did not provide protection from challenge with SHC014-MA15 with regards to weight loss or viral titer (Supplementary Fig. 5a,b). Consistent with a previous report with other heterologous group 2b CoVs15, serum from DIV-vaccinated, aged mice also failed to neutralize SHC014-MA15 (Supplementary Fig. 5c). Notably, DIV vaccination resulted in robust immune pathology (Supplementary Table 4) and eosinophilia (Supplementary Fig. 5d–f). Together, these results confirm that the DIV vaccine would not be protective against infection with SHC014 and could possibly augment disease in the aged vaccinated group.

          https://www.nature.com/articles/nm.3985

          We found that higher concentrations of anti-sera against SARS-CoV neutralized SARS-CoV infection, while highly diluted anti-sera significantly increased SARS-CoV infection and induced higher levels of apoptosis. Results from infectivity assays indicate that SARS-CoV ADE is primarily mediated by diluted antibodies against envelope spike proteins rather than nucleocapsid proteins. We also generated monoclonal antibodies against SARS-CoV spike proteins and observed that most of them promoted SARS-CoV infection. Combined, our results suggest that antibodies against SARS-CoV spike proteins may trigger ADE effects. The data raise new questions regarding a potential SARS-CoV vaccine, while shedding light on mechanisms involved in SARS pathogenesis.

          https://www.sciencedirect.com/science/article/pii/S0006291X14013321a

          Obesity also causes systemic inflammation and adversely impacts immune function and host defense in a way that patterns immunosenescence.32,33 Obese patients have higher rates of nosocomial infections following surgery and experience altered pharmacokinetics of antimicrobial drugs.32, 33, 34, 35, 36 Obesity emerged as an important risk factor for increased hospitalization and infection severity during both the 2009 influenza A virus H1N1 and COVID-19 pandemics.14,37, 38, 39, 40, 41, 42 The antibody response to the seasonal influenza vaccine is impaired in obese individuals, and virus shedding is prolonged during influenza illness.43 Compared to vaccinated normal-weight adults, vaccinated obese adults have twice the risk of influenza or influenza-like illness.33 Animal-based studies suggest that obesity increases the severity and duration of viral infections, increasing the potential for the evolution of pathogenic viral variants.44

          https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7205734/

      • Personally, my concern is not if the vaccines fail, but the speed at which we’re hustling them through development and the removal of legal recourse to pursue unknown long term damages.

        From the PREP act: “a covered person shall be immune from suit and liability under Federal and State law with respect to all claims for loss caused by, arising out of, relating to, or resulting from the administration to or the use by an individual of a covered countermeasure”

        • The liability problem is going to be even worse because the safety issue is not something that shows up right away. The danger is that months or years later the vaccinated person gets exposed to SARS2 or SARS3 and then they will get sicker than they otherwise would have.

  2. The actual rate of infection and the counter factual rate of infection seem to diverge about one to two days after the introduction of mask wearing. This is way to early given disease transmission kinetics (incubation period etc). Lockdowns for instance take 7 to 10 days to slow infection rates.
    The synthetic control units are in the authors heads.

    • its definitely a forking path but this is the author’s description of figure 1, panel b:

      “In panel B, we therefore start our analysis on 30 March, the day the mask campaign was announced. In this case, we observe a widening gap between Jena and the comparison group after about 10 days. Taken together, this effect appears to be plausible and is found to be robust when confronted with different sensitivity tests”

      i do feel like if the people of jena listen to their government on march 30th and start wearing everywhere, it would be surprising if they didnt change their behavior in other ways as well

  3. This is an interesting study, consistent with (anecdotal) evidence from other countries where mandatory masking was introduced quite early, such as Austria and the Czech Republic. These have all reported significant declines in reported Covid-19 infection cases following the mask mandates.

    Nevertheless there are discussion points with the report that could be raised:

    1. The “Jena Zeigt Maske” campaign obligates mask wearing in public transport, shops and workplaces but not universally, such as outdoors (the efficacy in clinical setting is not disputed). What are the exceptions, were they significant, was adherence near universal even at locations not strictly mandating them?

    2. The data reported is the number of cases, but does not give indication of additional factors that might affect the case numbers, such as testing rates. Under what conditions are tests done – symptomatic only, known contacts of infected individuals? Did the test rates (number of tests per day) differ over the period of analysis?

    3. More relevant would be the effect on hospitalisation rates, severity of infection, and death. Did rates of these also decline and, if so, by how much and how significant?

    4. There are differences between the Jena results and other regions/cities that are not explained beyond imprecise statements such as “the local population may have taken the crisis more seriously than in other regions”. Is there any justification/evidence for this assertion? If they did take it more seriously would that not also be reflected in better compliance with social distancing, hand washing and avoidance of unnecessary movement?

    5. The report summary states “Requiring face masks to be worn decreases the growth rate of COVID-19 cases by about 40% in Germany”, and that number (40%) has been widely shared in news articles around the world. But it is a point estimate and there does not seem to be any attempt to estimate confidence intervals (uncertainty), even with an awareness that other regions present with smaller effect and evidence of variability.

    6. The authors conclude “… our results suggest that requiring masks is a cost-effective, less economically harmful, and democracy-compatible containment measure for COVID-19”. A reasonable statement and sensible given the low impact (for most people) of compliance. It is unfortunate that this issue became contested on political/ideological grounds.

    It should be acknowledged that there is considerably more detail in the referenced paper by the same authors (Mitze, et al, 2020) that does provided additional context, including adjusted p-values for the effect size. It would be interesting to know of the SCM procedure as documented is as robust as claimed.

    • 1. The extent of mask requirement, public transport and shops, is what became mandatory all across Germany later. There never was a requirement to always wear a mask outdoors.

      2. I’m not aware that there are regional differences in testing availability in different German regions, aside from singular incidents where labs failed to secure necessary materials in time. Germany has always had a positive rate under 10%.

      3. I can probably run the numbers on that from the data the RKI publishes on arcgis, but given that the difference is on the order of 60 cases and Germany has an overall case fatality rate of 4%, you’re probably going to have huge error bars on the death counts that prevent drawing any conclusions.

      4. The idea is that instituting face mask requirements has knock-on effects besides the working of the mask itself; as a public health measure, we’d expect it to modify behaviour more widely, making people “take things more seriously”. As such, the question is not “do face masks work”, but “does mandating face masks work”, and the latter question encompasses all these side effects, and is really the more important question from a policy perspective.

      5. I haven’t read the original paper and don’t know whether the authors provide these intervals there.

      6. The WHO spoke out against politicising epidemic response months ago, predicting that divisiveness “helps the virus”. The major German populist party, AfD, has attempted to foment such a division, but the other parties and politicians stood together, and chancellor Merkel succeeded in getting the states to enact largely similar restrictions all over Germany, with great success.

      • > 3. I can probably run the numbers on that from the data the RKI publishes on arcgis, but given that the difference is on the order of 60 cases and Germany has an overall case fatality rate of 4%, you’re probably going to have huge error bars on the death counts that prevent drawing any conclusions.

        Im more inclined to believe the uncertain numbers based on the hospitalizations than certain number based on detected cases.

    • >>A reasonable statement and sensible given the low impact (for most people) of compliance. It is unfortunate that this issue became contested on political/ideological grounds.

      Yeah. Now I’m wondering if we’d have been better off socially/economically if many areas of the US that were less hard-hit at first, but with significant population density (eg TX AZ FL CA, or at least urban areas of those states), had done face coverings *instead of* stay-at-home / nonessential business closures in late March. Face coverings weren’t even encouraged then.

  4. Did I miss interval bounds on the simulation of this alternate reality? Extrapolating a mere point estimate seems tenuous. Also, the time span is very short. What happens to the actual line after the end of time? Does it ever come back up and reintersect with the no mask line?

  5. And to add to the complications that make it so hard to interpret what is going on, consider that “consumer confidence” about venturing out into stores, restaurants, etc. also depends on their perceptions of the prevalence of mask wearing. So, ironically, the mask-refusers who are so eager to just open everything up again are actually an obstacle to re-opening because their presence scares off customers and makes it less likely that business can successfully re-open. (And their aggressive vociferousness probably amplifies this effect beyond what mere abstention from mask wearing would do.)

    Much as we would all like to get a clearer picture of causes and effects in this epidemic, I defy anybody to come up with a direct acyclic graph that describes it in any realistic level of detail. Everything is endogenous!

    • Yeah, the “vociferousness” factor I think is underestimated. The more COVID dominates the public discourse, including due to political fights about it, the more that feeds the fear level.

      • See page 2: xayadata.com/covidstates.pdf

        One thing we can say is the excess deaths this year do seem to have started matching up with covid deaths around the week of May 9th.

        So the difference between dying with covid vs because of covid does not seem to be very large. Unless you want to suppose people who test positive get “special treatment” that increases the likelihood of death. AFAIK that (early intubation) mostly stopped by the end of May in the US, so Id assume the covid death numbers are probably largely accurate since then.

  6. Several people have mentioned the 1957-58 flu pandemic. I was in my early teens then, but don’t recall any talk of a flue epidemic. Of course, kids in their early teens aren’t often up on the news, so maybe that was why I don’t recall people taking about the flu. But I do recall getting sick once in Spring of 1957. One day, in school, I started feeling tired and like I had a fever. I was rubbing my eyes a lot — the teacher noticed and said that it might mean I needed glasses. When I got home that afternoon, I plopped face down on a bed (I’m not sure whose), and woke up several hours later feeling fine, with no sign of fever. I wonder now if it might have been the flu. I do recall that when I was a child, my body tended to respond to illness by getting a high fever (around 104° F), and I could “sleep off the fever”, with symptoms gone when I woke up.

    Does anyone else have personal recollections from 1957?

    • I wasn’t around for the one in the 50’s but I was for the next one in the 60’s. No recollection of it at all and my mother has only a vague recollection of either of those outbreaks. I do not think either one of them (at least in the sort of USA suburban middle class enclave in which I was raised) drove stores and restaurants out of business, shut down schools and universities or became the all-consuming obsession of everyone, all the time like COVID has.

      But then again, even growing up in the 60’s we tended to see mumps and measles and chickenpox and various other viral illnesses as just part of life. If my family had all come down with the flu in 1968-9 then unless someone died or suffered lasting disability it would just be remembered as one more disease on that list.

      Even if COVID turns out to be 5x or 10x as deadly as the big flu outbreaks of the 50’s and 60’s the societal disruptions is several orders of magnitude out of proportion.

      • If you don’t have a vaccination program against mups/measles/chickenpox, most of the adult population had it as a child and is immune. For the Covid-19 outbreak, we don’t think that there are large parts of the population that are immune, so with a similar “severity”, more people will be affected. SARS-CoV-2 is new to humans, and therefore devastating.
        How “harmless” infectious diseases from Europe wiped out large parts of the indigenous American population is part of the history books.

        • Mendel said, “How “harmless” infectious diseases from Europe wiped out large parts of the indigenous American population is part of the history books.”

          Very good point.

        • This is one of the reasons why I’m not worried about the prospects of COVID becoming endemic.

          In an immunologically-naive population, we’re seeing IFRs mostly below 1% (now), whereas things like measles were very devastating to an immunologically-naive population. IIRC there were Pacific islands where essentially an entire age-cohort died.

      • Well, chickenpox was “just part of life” for me growing up in the 90s! It was usual for my age-group, but people just a couple years younger got the vaccine before they got the disease.

        But yeah, the fear of COVID is I think largely because most modern ‘Western world’ people have basically never had to worry about infectious disease (except possibly HIV/AIDS which is very different).

        But there’s just no way COVID will be 5x-10x deadlier than 1957/1968 flu… current deaths are roughly comparable to that, and what percent of the population do you think is already infected? It surely must be more than 10%, and nothing ever goes to 100%… and presumably current-IFR is less than pandemic-average-IFR since treatments have improved…

        • I’m not going to bet money over the Internet, but I don’t see how it could possibly be less.

          10% of US population is 33 million people.

          CDC was estimating 25 million way back in, what, end of June? Not sure exactly. But that was based on serology results, and antibodies lag, and there’ve been a ton of infections since then.

          And we are at almost 6 million *confirmed cases*. I just can’t believe that we’ve caught 20% of all infections, given how limited testing was in the first surge and the age distribution of cases early in the second surge (ie, younger people are much less likely to be sick enough to seek care).

          I mean, it’s possible. But it seems pretty unlikely.

        • Know the old saying, “The beatings will continue until morale improves”?

          I’m thinking, “The modeling will continue until the data arrives” as the current-day version of that.

        • People stopped caring so much once they realized the t-cell immunity was even more important when it comes to coronaviruses.

        • The sad part is, even if we could guesstimate a number we’re comfortable with for nationwide population prevalence (I’d say 10% is probably within a factor of 2x one direction or the other) that’s not what we really need to know. The impact of the disease is totally different in various age groups, for starters, not to mention race or other risk factors. And the geographical clustering of infections is important, too.

          So what we need isn’t just a fairly reliable estimate of population prevalence. We need it to come from stratified random samples so we can get good estimates in subgroups. At this point, I really can’t believe that will be available at any point this calendar year. Which until recently, I would have found mind-blowing. Now nothing surprises me.

        • Oh I don’t think the US is *way* above 10%. (covid19-projections.com has about 14%). But probably above it, yes.

          But yeah. I think in early June it was 10-25% in the urban Northeast and a few other places hit hard early (New Orleans, Detroit, etc.) but really low elsewhere.

          But now I think most places with significant population density have “caught up”. Prevalence will still be very low in some places, but ones that represent a tiny fraction of the population.

        • confused,
          I understand why you think it’s over 10%, I’m just saying I think it’s more likely under than over for a variety of sampling-related reasons not worth going into here. Sadly, although we may have some good numbers in the future, I don’t think we will ever know what the percentage was on August 24, 2020 well enough to settle the bet definitively.

          At any rate we agree, I think, that the number is probably in the range 6-16%. This is why I think that, had people not made major behavioral changes, we’d be way over 2x as many deaths. I agree that infecting, say, 4 times as many people would not necessarily have killed 4 times as many as have already died: the old and vulnerable are better were already learning to protect themselves, and treatment was improving, etc. But it took some time for that learning and knowledge-propagation to happen, plus simply the availability of masks and so on. Maybe we could get away with letting a whole lot of people get infected now, without causing a major increase in deaths, but only if behavior has changed a lot since just a few weeks ago.

        • >>At any rate we agree, I think, that the number is probably in the range 6-16%.

          Well, I don’t really think it can be the lower end of that range*, but yeah otherwise that seems about right. More than 16% for the whole US seems extremely high.

          I think our real disagreement is what we expect the herd immunity threshold for the US to be.

          Although, the whole US is really not “one population”, so speaking of one HIT for all the US is probably not accurate. What I’m really thinking of is a situation where highly-populated/highly-connected parts of the US get to herd immunity… or at least to a level of immunity sufficient to suppress the disease to “endemic”/low levels, not to zero cases… Even if localized outbreaks continued in a few remote places, the overall US or regional statistics would look indistinguishable from “herd immunity”, ie it would be below epidemic threshold.

          *Given that the US is at 178,000+ reported deaths, with 900-1000 more per day, and there is lag from both infection to death and death to reporting of death, deaths resulting from *all infections occurring up to now* must be more like 200,000.

          If 6% of the population is infected, that’s 20 million people, so a 1% IFR.

          If the New York IFR was something like 1.1%, given improvements in treatments and understanding of the disease, as well as apparently a younger population infected, I think the US-average IFR must be lower.

        • Also — it’s sad that the CDC has been such a mess, but not really surprising to me. Even beyond the current political turmoil (and I think more central — meaning that electing a different President won’t fix it by itself) … it’s a federal bureaucracy that hasn’t really been *tested* in a very long time.

          Ebola never amounted to anything in the US and swine flu was generally considered almost a false-alarm pandemic. It’s been a long time since AIDS was new in the US (and US response to that was not particularly good either!)

          I think it’s comparable to NASA in the Apollo era vs. now.

      • The 1960-1970 flu killed about 100,000 Americans in a year and a half. COVID has killed more that twice that number in 1/3 of the time. The U.S. population was lower then, of course, but COVID has still killed at about 3x the rate per capita, and, importantly, this has happened in spite of measures that have reduced the spread of the virus.

        At least one person on this thread has argued that even with out most of these measures the death rate wouldn’t be all that much higher, certainly (he thinks) less than a factor of two, but I think that’s a hard proposition to defend.

        • The time itself isn’t terribly relevant, compared to the % of population infected, to judging how far along an epidemic is.

          Current COVID deaths reported aren’t directly comparable to historical estimates for past pandemics, but the deaths weren’t evenly spread over that year and a half, either (and it wasn’t really a thing in the US at the beginning; both 1957 and 1968 started in East Asia, and worldwide travel was different then, China and the US were on opposite sides of the Cold War).

        • 5 patients were given the therapy. 4 “responded well.” There’s a decent chance you’d get a result like that if you gave them tap water. I’m sure doctors all over the place are trying all kinds of things, and I hope some of them work. This might turn out to be one of them. But if it doesn’t, I wouldn’t say the initial result was a ‘fluke.’

        • Phil said,
          “5 patients were given the therapy. 4 “responded well”. There’s a decent chance you’d get a result like that if you gave them tap water.””

          Anoneuoid relied,
          “I didn’t even mention the therapy. Look at the levels of methemoglobin. How did this get missed for 8 months?”

          I agree with Anoneuoid that the crucial takeaway is, (quoting from the article),

          “In conclusion, NO, methemoglobin and oxidative stress may play a central role in the pathogenesis of critical COVID-19 disease,”

          and so it seems wise to me to start looking for markers of these conditions, as a path worth investigating more.

          In other words, it’s not the clinical trial that is most important, it’s the new approach.

        • How has it taken until August to notice that methemoglobin levels are so high in covid patients?

          For n = 25 in heathly and n = 25 in covide patients they report 2.5% and 16.4%, respectively. If you check what methemoglobinemia is you will see all the symptoms match up.

          Further, the best treatments are antioxidants, transfusions, and hyperbaric chambers:

          https://en.wikipedia.org/wiki/Methemoglobin
          https://en.wikipedia.org/wiki/Methemoglobinemia

      • Sorry, I was just speaking from experience. I’ve seen metheamoglobinaemia due to recreational use of amyl nitrate and prescribed drugs such as dapsone, but not for any other reason. It is associated with a common enzyme (G6PD) defiance. Having said that however, I did find a case report from the CDC of a man with covid and methaemoglobinaemia who had not received any culprit drugs.
        Palmer K, Dick J, French W, et al. Methemoglobinemia in Patient with G6PD Deficiency and SARS-CoV-2 Infection. Emerging Infectious Diseases. 2020;26(9):2279-2281. doi:10.3201/eid2609.202353.

        • There is also this one:

          Coronavirus disease 2019 (COVID‐19) has been associated with a range of hematologic findings and complications [1]. We have encountered three cases of significant methemoglobinemia, and five cases of relatively mild methemoglobinemia, among patients being treated for COVID‐19 in our health system during a 4 week period in April 2020. For comparison, there was only one case of mild acquired methemoglobinemia of any cause documented in our health system during the preceding year. Below we describe the three cases of significant methemoglobinemia, including their presentations, treatments, and outcomes

          […]

          The diagnosis of methemoglobinemia is seldom thought of given its rarity, and thus may remain underdiagnosed during the COVID‐19 pandemic. The typical presentation consists of abrupt symptoms of tissue hypoxia following exposure to an oxidizing substance. Notably, as this is a condition of increased heme‐oxygen avidity rather than hypoxemia, dissolved oxygen levels on blood gas may be normal in spite of clinical evidence of hypoxia and decreased readings on pulse oximetry. A high index of suspicion is thus required and diagnosis is most often made on co‐oximetry or specific blood Met‐Hb assay.

          https://onlinelibrary.wiley.com/doi/full/10.1002/ajh.25868

          Seems whenever people look for it they find high levels of methemoglobin but almost no one is looking!

  7. Replying to confused above, but it’s too deep in the comment depth to keep track…

    @Daniel Lakeland
    >>This may explain our differences in point of view. I think you’re not understanding the math.

    It’s not understanding the math, but disagreement on the following:

    >>Hospital capacity is much much lower than herd immunity.

    I think this is not necessarily true.

    I mean, if literally 70% of the population was infected *at the same time*, yeah that would overwhelm capacity.

    But a) I think the herd immunity threshold in the vast majority of the US, including most major cities, is not nearly that high (there are no US cities like Iquitos, Peru); and b) even a 70% threshold doesn’t mean 70% infected *simultaneously*.

    https://www.beckershospitalreview.com/rankings-and-ratings/states-ranked-by-hospital-beds-per-1-000-population.html

    Gives hospital beds per 1000, TX for example has 2.3 beds / 1000 people.

    Let’s just assume 5% of infections need hospital care. Then if at any time there are .0023/.05 = 4.6% of the population infected, then TX would have hospital overwhelming. If 10% of infections need hospital care, it’d be 2.3% infected… So I think somewhere in the 2-5% infected at any one time results in hospital overcapacity. Since it’s uneven, probably regions within TX experience emergencies even if the whole state doesn’t reach capacity. So let’s call it 3% for ease of calculation.

    Now, suppose we start at 0.5% infected, and R = 1.1 for an extended period. Let’s call the infectious period duration 10 days, 1.1/10 = .11 so we can approximate the infections as .005 * exp(.11t) and the derivative of this is .005*.11*exp(.11t) which we can set = .03 to solve for t… it turns out to be about 36 days.

    So at r = 1.1, 36 days after you hit half a percent infected you will have as many new hospital-needing infections per day as your entire state’s hospital capacity.

    The only reason we haven’t seen massive hospital overwhelming has NOTHING to do with herd immunity which at any case is easily an order of magnitude higher than these numbers… It’s all down to the fact that for the most part since March we’ve only really had a few days to tens of days during which R was greater than 1

    • >>So I think somewhere in the 2-5% infected at any one time results in hospital overcapacity.

      I think this depends on the age distribution of infections. If (as seems to have actually happened in June in TX, FL, and other southern states) the infections were predominantly young adults e.g. bar-goers, even 5% hospitalized might be much too high.

      USS Theodore Roosevelt showed a hospitalization rate under 1%, I believe, and this population was strongly male-biased*, while COVID seems to hit men worse.

      *https://www.cdc.gov/mmwr/volumes/69/wr/mm6923e4.htm

      >>Since it’s uneven, probably regions within TX experience emergencies even if the whole state doesn’t reach capacity.

      But hospitals move patients around, too. There was at least one case reported in the media where a patient was moved from far South Texas to the Panhandle.

      Far South Texas was much harder hit than the large cities, even Houston, though this wasn’t clear from the media reporting… I wouldn’t be surprised if those areas are essentially at herd immunity.

      Note that I **don’t** think places like Houston and Dallas have hit herd immunity, though individual hotspots within those large metro areas may have. I think the current situation is a sort of “fake herd immunity” where the measures taken were not themselves sufficient to put R below 1, but then infections moved it the rest of the way to below 1.

      So if TX totally gave up on masks and stuff tomorrow, we would see a second rise in infections, yes. But I don’t think it would be hospital-overwhelming-scale.

      >> herd immunity which at any case is easily an order of magnitude higher than these numbers…

      Are you confident that herd immunity in places like Texas would be 50% or more?

      If it’s 70% in Iquitos, Peru, that seems unlikely.

      In dense areas with lots of multifamily households in the Lower Rio Grande Valley, it’s plausible (and given the death rates and the rather low median age, 50% infection in some of those communities having already happened wouldn’t surprise me at all).

      But those don’t make up most of the TX population. The big cities like DFW/Houston are much more “suburban sprawl”/highly car-dependent/etc., very different setup.

      • > So if TX totally gave up on masks and stuff tomorrow, we would see a second rise in infections, yes. But I don’t think it would be hospital-overwhelming-scale.

        The number of covid hospitalizations in Texas reached 10k on July 10. It had doubled since June 26, it had doubled twice since June 16. Why do you think the trend changed (it peaked shy of 11k) if it was not mostly due to masks and stuff? You said elsewhere that another doubling would have put it in hospital-overwhelming territory.

        • >>Why do you think the trend changed (it peaked shy of 11k) if it was not mostly due to masks and stuff?

          Oh the masks and greater caution definitely played a role. As I said, I really don’t think TX is actually at herd immunity, and infections would rise again (at least somewhat) if we went 100% back to normal.

          Though cases/deaths are unevenly-distributed enough that some *places* in TX *could* be at herd immunity.

          But I do think immunity is a significant factor. The covid19-projections.com model shows Texas over 20% infected, and given that our R didn’t start that high compared to say NYC, that could make a big difference.

          Combined with the fact that R0 was probably not as high as in the Northeast due to the density/contact patterns of big TX cities.

          If Texas would burn out at 30% or 40%, and we’ve already gotten to 20% without overwhelming hospitals, doing so in the future would be extremely unlikely. If the HIT is really 70% in Texas (or if the current infection level is a lot lower), well, it’s a lot more plausible.

          I mean, I think I always come across in text as being more certain than I actually am. It *could* be purely the results of measures. But given that compliance with measures is not actually that great here, I really don’t think that’s the only factor.

        • Ok, I understand now that when you say things like “Houston was nowhere *near* maximum surge capacity” you mean “but it would have happened in days if not for masks and stuff”.

        • Well, no, I’m not saying that either.

          The rate of growth actually observed was not enough to overwhelm surge capacity “in days”.

          Here’s a graph of Texas Medical Center (Greater Houston area) COVID ICU patients over time:
          https://www.tmc.edu/coronavirus-updates/total-tmc-covid-19-positive-patients-in-icu-beds/

          Now, that rate of growth is going to be slower than it would have been with no precautions, sure.

          If absolutely *no one* did *anything*; no more telecommuting than happened in 2019, no more precautions in healthcare and elderly-care facilities, no masks, no social distancing; nothing, even on an individual level, then maybe hospitals would have been overwhelmed.

          But that was never plausible, and didn’t happen even in places like South Dakota or Sweden that are considered to have “stayed open”.

        • IE – I totally think masks made a difference!

          But I don’t know that they made *the difference* between overwhelming hospital capacity and not. They could have, though.

          Those are two different statements.

          It’s very hard to imagine masks not helping *some*, but I am just not sure TX was ever that close to the limit in most places (and with the ability to move patients to other hospitals, local more-rural systems reaching surge capacity doesn’t lead to turning people away).

        • > If absolutely *no one* did *anything*; no more telecommuting than happened in 2019, no more precautions in healthcare and elderly-care facilities, no masks, no social distancing; nothing, even on an individual level, then maybe hospitals would have been overwhelmed.

          Maybe? If in the real world (where people did things) ICU bed use in that chart was growing by over one hundred beds per week, I can’t imagine a conterfactual world when *no one* did *anything* and hospitals were not overwhelmed quickly. Please enlightment me.

          And in the real world the trend changed because more people did more things. Several hundreds of additional beds would have been required otherwise. If you don’t think it was the behaviour of people that made the difference, then what was it?

        • >>growing by over one hundred beds per week

          Right… and it peaked about 700 below maximum Phase 3 capacity (and the capacity chart does say there is additional capacity beyond that…), so that would be 7 weeks rather pessimistically.

          Very different from “days away”.

          >> I can’t imagine a conterfactual world when *no one* did *anything* and hospitals were not overwhelmed quickly.

          My point is that “absolutely no one did anything” is not actually a possible scenario. The range is really from “full lockdown” to “no organized actions on the government level, but many individuals, medical care facilities, elder care facilities, and some businesses respond”.

          However, in a hypothetical counterfactual where literally no action was taken by anyone, even individuals, yes, at least some of the large cities in TX may well have seen overwhelmed hospitals.

          That doesn’t necessarily mean it was a plausible real-world outcome.

          >>If you don’t think it was the behaviour of people that made the difference, then what was it?

          I think it was a *combination* of behavior and increased immunity.

          I don’t know how much each factor mattered.

          But if the covid19-projections.com model is anything close to right showing TX at >20% immunity, I can’t see that not making a difference… not herd immunity by itself, but maybe enough to turn haphazard and half-hearted behavior changes from basically ineffective (R > 1) to basically effective (R < 1).

        • confused –

          > I think it was a *combination* of behavior and increased immunity.

          See comment below. They don’t function independently of each other. Decreased infections as a result of increased immunity isn’t a fixed state. It depends on (and is a function of) behavior, and can change if behavior changes.

        • >> They don’t function independently of each other.

          Yeah, I’ll agree with that. Behavior changes affect heterogeneity, sure.

          All I am saying is that I am not convinced the difference is large enough to overwhelm hospitals in most of the US, including most large cities.

          It could be! I’m not really arguing that it isn’t; but I am arguing that we shouldn’t be saying that we prevented overwhelming hospitals by doing X and Y, as it very well might not have happened regardless.

          I wish we had good hospital use and capacity data from places like Peru and Brazil… That would probably make things a lot clearer.

        • confused –

          > It could be! I’m not really arguing that it isn’t; but I am arguing that we shouldn’t be saying that we prevented overwhelming hospitals by doing X and Y, as it very well might not have happened regardless.

          That seems awfully vague. I think it is unlikely that had people not changed their behaviors – whether mandated or not – that hospitals wouldn’t have been overwhelmed. Do we know precisely what the effect of any particular factor was? I’d say no. But given the high damage risks involved, I think it’s kind of useless to speculate as to exactly what might have happened. The judgement of the majority of people who study this issue seems to be that it wasn’t worth the gamble that hospitals wouldn’t have been overwhelmed, and I can’t imagine any real world scenario where people wouldn’t have changed their behaviors anyway so I see little point in speculating what might have happened if they hadn’t.

        • confused –

          > Though cases/deaths are unevenly-distributed enough that some *places* in TX *could* be at herd immunity.

          Look at the current rate of infection in Sweden compared to Norway, Finland, Germany, or even Italy or the Netherlands.

          Some people are attributing the drop in rate of infections in Sweden to “herd immunity” as compared to those other countries. Maybe if Sweden was uniquely “locked down” as compared to those other counties the claim would make more sense – but it isn’t.

          There are way too many variables in play to draw conclusions about “herd immunity,” IMO. It’s one thing to construct a theoretical model that says that because of heterogeneity its possible to reach a HIT as low as 10% or 20%” it’s one thing to say that as you approach a HIT, transmission rates will drop. It’s quite another to try to seat of the pants, back of the envelop reverse engineer from a relatively lower infection rate in one community relative to another to attribute causality to “herd immunity.”

          Just too many variables, like mask-wearing in play.

        • Well, I think “some places in TX” are *way* above 10-20%.

          I think TX *as a whole* is in that range or maybe a bit higher (covidprojections.com’s model has 21.7% … they assume a fairly low IFR, but TX has a low median age and at least for the first part of our surge infections were skewed toward young adults).

          Individual places that have been much harder-hit than the TX average (like the Lower Rio Grande Valley and some neighborhoods/communities harder hit than the average of the metro area they are part of) must be higher than the statewide average.

          And in some cases (like the LRGV) the difference is quite dramatic.

          I’m not saying “infections have dropped in these places therefore it must be herd immunity”, I’m saying “the % infected in these places has got to be extremely high”.

        • And now in Austin, it remains to see how much the return of students will affect the incidence rates of COVID, and also how the influx of evacuees from Hurricane Laura will affect it. (The plan had been to put evacuees in hotels, but the number of evacuees exceeded the number of vouchers, so the Convention Center is being used to house evacuees. (https://www.kxan.com/news/local/austin/hurricane-laura-evacuees-being-turned-away-at-cota-as-hotels-are-full/)

        • Yeah.

          The students issue will probably have a lot to do with how much the student population mixes with the off-campus population, and with what parts of it.

          (Well, not so much incidence, but severity and possibly detection. Infections among the college-student age group would skew mild and many might not seek care/testing, so unless testing is focused on the students, there might be a lot of infections but few confirmed cases.)

          The evacuees … I really have no idea how that will turn out.

          I think Austin has been hit less hard than the larger TX cities, so it may be relatively vulnerable (*if* one believes that the immunity levels in e.g. Houston or San Antonio are high enough to meaningfully decrease spread).

        • > “the % infected in these places has got to be extremely high”.

          Could be. But the seroprevalence surveys seem pretty unreliable to me. But I don’t think we can say that “since the infections are high, we’ve hit herd immunity.” because much depends on behavior. Certain behaviors and you *effectively* hitnit at a lower level. Certain other behaviors and its higher.

          Take Arizona. There want much spread. They opened up. Then there was a lot of spread people said it was because more yinfer and asymptomatic people were being tested. Then hospitalizations and deaths spiked. So then behaviors changed and the rate infections dropped. Because a lot of people were already infected? Maybe. But it could be that after the infections spiked people started taking it seriously. Maybe stores required masks. Maybe people stopped going to bars. Maybe an effective HIT – but if behaviors change, the HIT goes up.

        • Hmm, I’m not sure we really disagree.

          I agree that TX and AZ are not at true herd immunity, but just an “effective HIT” due to behavior.

          But that applies to the whole population of those states. If there are local communities with infection levels like 3x higher, *they* may very well have hit true herd immunity.

          >> But I don’t think we can say that “since the infections are high, we’ve hit herd immunity.” because much depends on behavior.

          Sure, but you can only go so high! Sooner or later really high infection levels will dampen spread and eventually end it even if behavior is unfavorable.

        • >>There’s no inherent reason why we HAVE to disagree.😊

          Sure. What I mean is, I think our underlying picture of what is actually happening in the US re: COVID, and why, is not really all that different, even though we seem to keep arguing about it…

        • Comment in moderation. I still can’t figure out what triggers moderation at this site.

          Anyway, my point was that basically you can’t disaggregate herd immunity and behavior because the heterogeneity affects the HIT, and behaviors affect heterogeneity.

        • > Comment in moderation. I still can’t figure out what triggers moderation at this site.

          A couple of links in a message will do it (maybe there are other factors).

    • Also, I think ICU capacity is the real limit (it’s a lot harder to increase than general capacity) and the ICU rate is *much* lower for young adults. I think that age-distribution of population infected makes a *huge* difference in practice.

      So problems would much more likely have looked like convention centers being used as hospitals (which was prepared for as a possibility, at least in Houston), not actually turning away patients (as happened in Lombardy, but I think not really anywhere in the US, even NYC… though in NYC quality of care must have dropped a lot).

      • confused said,
        “So problems would much more likely have looked like convention centers being used as hospitals (which was prepared for as a possibility, at least in Houston)”

        Which might become a reality depending on how strongly the coming Hurricane might affect the Houston area.

        • Maybe. But right now it looks like it’s not really heading for Houston, and the *other* side of the hurricane is where the heavy rain generally falls*. The NHC is predicting maybe tropical-storm-force winds and 2-4 feet of storm surge in Galveston Bay… Houston itself (as opposed to some very-low-lying places on the coast like Seabrook) probably isn’t going to be hit that hard unless the hurricane “swerves” at the last moment.

          *Hurricane Harvey really made landfall much farther southwest on the Texas coast, but the really heavy rain fell on Houston, Beaumont, etc.

  8. I’ll agree with a lot of that.

    >>The judgement of the majority of people who study this issue seems to be that it wasn’t worth the gamble that hospitals wouldn’t have been overwhelmed

    This is a true statement, but IMO the public-health and medical community tends to err on the side of caution, in many ways (e.g. pace of development/roll-out of COVID vaccines, and new drugs in general) more than I think is warranted.

    I think I can believe the data without trusting the judgment ;)

    >>I can’t imagine any real world scenario where people wouldn’t have changed their behaviors anyway so I see little point in speculating what might have happened if they hadn’t.

    I do agree with this. (But I think this also means that we can’t say with much confidence that formal government measures made any real difference…)

Leave a Reply to Phil Cancel reply

Your email address will not be published. Required fields are marked *