The IHME Report

An IHME report came out over the weekend. I was pleasantly surprised by it contents. Its conclusions and projections broadly match what I said in the last Omicron post, with only relatively minor obvious nonsense, and seemed worth their own post to break down.

Headline Takeaways

Our models for the United States suggest that infections may be currently peaking at 6 million a day, with varied peaks by state. We expect more than 50% of the US population will be infected with Omicron in the next 6 weeks. The infection-detection rate (IDR) is declining, although shortages of testing may lead to more rapid decreases in the IDR than we currently model. Nevertheless, we expect reported cases to exceed 1 million a day, peak by the third week of January, and then decline sharply.

I too expect reported cases to peak during the third week of January (to be clear, when I say we will peak on 19 January, I mean cases detected on 19 January, not infections that happen on 19 January. If counting from initial infections, I’d have said 16 January, but it’s hard to ever know what the actual lag was there.)

Once the peak arrives, a sharp decline is to be expected shortly thereafter, although I still expect cases to remain elevated (above pre-Omicron levels) for several months.

Note that ‘more than 50%’ sounds like the kind of thing one says when one does not want to make a precise prediction, but there are graphs later on.

Hospitalizations will increase to a peak that may be twice as high as last winter. These figures, however, include incidental admissions. Because the prevalence of Omicron infection is so high, many individuals hospitalized for other conditions will have asymptomatic infections. Incidental admissions may exceed 50% of total COVID-19 admissions in some states. Daily deaths will increase but remain below 2,000 a day, peaking by the end of the month. This means the peak of deaths will remain below the summer Delta surge and well below the peak last winter.

That sounds like a best case scenario. If hospital admissions peak at twice last winter, but half of them are incidental, then things won’t be substantially worse than last winter. We survived last winter. Deaths peaking below 2,000 per day means they won’t even double once, and the peak will come within a few weeks. Expecting them to peak by month’s end seems wrong given the historical lag times, but it’s only a difference of a few weeks.

Words of high wisdom follow, noting that all of this is baked in.

Our alternative policy scenarios, including more rapid scale-up of boosters to all who have been previously vaccinated, increasing mask use to 80%, and vaccinating the partially hesitant, have only a small impact on the trajectory over the next 4 months. The speed of the epidemic is so fast that policy interventions will have little impact. In previous waves, the control strategy has been to control infection and thus reduce hospitalization and death. Given that there is little prospect of controlling infection, strategies need to focus on reducing harm in the vulnerable and minimizing disruption.

It is essentially too late to do anything but mitigation. If the baseline scenario is as they describe (and roughly as I predict as well) then only those with a strong desire to avoid infection should do much. So this, very much this:

Given the massive numbers of infections in the community, testing and quarantining asymptomatic individuals may not be helpful. There appears to be no prospect for controlling transmission and considerable prospect for disruption of schools and essential services due to screening. States may need to consider revisions to their testing and quarantine strategies.

Quite right. Doing constant testing of asymptomatic schoolchildren makes no sense. My four year old son has had to have four tests in the past week alone.

There’s so much both right and wrong about this caveat:

Considerable uncertainty remains about the future course of the Omicron wave. First, the infection-detection rate may decline even more than we have estimated if testing capacity in states is overwhelmed. This would reduce the reported case rates below the 1.2 million that we have forecasted per day. Second, hospital admission screening will substantially impact the reported COVID-19 admissions. If some hospitals run out of testing capacity and do not screen all admissions, then the incidental COVID-19 admission rate may also decline. Third, a critical factor in understanding the trajectory of Omicron is the fraction of infections that are asymptomatic. Based on data from South Africa and the UK, we currently estimate this to be 80%–90%. Increases or decreases in this fraction asymptomatic have an important impact on the trajectory and severity of the Omicron wave.

The part that’s very right is that the potential lack of testing, and the uncertainty about how many infections are asymptomatic, create extreme uncertainty in measured numbers. However, when they say there’s uncertainty about the ‘course of the Omicron wave’ they are implying that all of the uncertainty is in the measurement of that wave, and any deviations from their projections reflect these changes in measurement.

Their Summary and Graphs

Here is their summary from January 3, edited for readability.

• Daily infections in the last week increased to 5.7 million per day on average compared to 4.0 million the week before.

• Daily hospital census in the last week (through January 3) increased to 99,600 per day on average compared to 76,800 the week before. [January 10: ~136k]

• Daily reported cases in the last week increased to 477,600 per day on average compared to 238,700 the week before (Figure 2.1). [January 10: ~712k]

• Reported deaths due to COVID-19 in the last week stayed level at 1,200 per day on average compared to 1,200 the week before (Figure 3.1). Estimated total deaths around 1,400 per day.

• We estimate that 57% of people in the US have been infected at least once as of January 3 (Figure 6.1).

• Effective R is greater than 1 in all states. Omicron is now the dominant variant in all states.

• The infection-detection rate in the US had declined to 22% on January 3.

Mobility last week was 7% lower than the pre-COVID-19 baseline (Figure 11.1). Mobility was lower than 30% of baseline in no locations.

A graph I found interesting, on detection rates over time, I’m surprised they think the decline in detection rates was this small recently:

A graph that should make one assume their methodology is wrong:

That number in Vermont is obviously stupid. There’s no way Covid kills people twice as often there as in all but a handful of other states, including all the states around it. I don’t believe the estimates in the other four outlier states either. It’s so weird to put a graph like this in one’s paper and not notice it doesn’t make any sense. Also worth noting that larger population states tend to have lower IFRs here (California, Texas, Florida and Ohio all <0.2%), in a way that doesn’t match anything physically happening. My guess is they are underestimating the true number of cases in New York relative to those places.

I think mobility is a good measure of extreme prevention measures, but a quite poor measure of non-extreme measures. So this is telling us that almost no one is taking extreme measures. I am surprised. If nothing else, something like 10% of people have Covid-19 right now, by their own estimates. Shouldn’t their lack of movement show up on this graph?

I simply don’t believe this graph here?

This shows no increase in test usage. If that’s true, then why are tests suddenly impossible to find? Why are systems backed up? Why does everyone I know report being told to test a lot more? If the answer is ‘we were already at our limit’ then I guess I basically disbelieve that, given we were then at it for three months and nothing changed.

Here are their vaccine effectiveness estimates, presumably unboosted. Note that they continue to believe that the vaccines were still holding up exceptionally well against Delta, and they weren’t losing that with time very much, which suggests that ‘using clinical trial data’ wasn’t serving them well here and they didn’t want to notice.

Projection Graphs

There are five scenarios here but there’s a reason I don’t talk about them before this, they all are effectively the same because it’s too late for prevention to do much.

They have us peaking at roughly 6 million infections per day, which is almost 2% of the population. When I did simple spreadsheet modeling myself I got the peak being somewhat lower and closer to 4 million per day.

One place I definitely disagree is that I do not expect this much drop-off to happen this fast after we peak. Behaviors will adjust, and I do not believe they are taking this into account. This failure matches my understanding of how they’re generating their models, so it’s easy to assume this is as simple a mistake as it appears.

I notice I’m confused how higher severity of Omicron scenario (which I do not believe is going to happen on its merits) causes a lot more hospitalizations but not many more deaths, since those extra hospitalized people should be a big source of more deaths. On top of that, if we did get to 450k hospitalizations, presumably we’d run out of resources in a lot of places.

They note other model projections for deaths.

The SIJKalpha model is clearly obvious nonsense. Their GitHub brags about how fast you can run the model rather than explaining what they’re doing. For deaths to peak this late and this high would only make sense if you think we’re mostly catching all the Omicron cases and there’s little dark matter out there? But that’s obviously quite wrong. As a sanity check I looked at their United Kingdom projections, and, well, no.

The CDC model seems like what one would expect if you had some dark matter but much less than you’d expect from the data points we’ve seen, so things don’t peak for a while longer and the true IFR is higher because Omicron is more severe. I don’t think this can match the data either, but it’s at least somewhat less insane.

The London College projection was from early December so I’m not going to be too harsh on them, but it doesn’t seem like it’s predicting the past all that well.

Delphi seems to be in the same universe as the IHME model, and the results seem plausible. Having MIT give you a sanity check seems strong.

Conclusion and Updates

There’s some overlap between my toy modeling and what IHME is doing. They’re more sophisticated in some key ways, although they seem to not be factoring in the control system properly, which is a big omission. What they are predicting here is confirmation that my expectations are reasonable, as is the Delphi/MIT projection.

I’d love to update my confidence levels a lot, but all these models are making similar predictions largely because they are making similar modeling assumptions. Thus, I can be confident that these are the correct predictions to be made given what I think are good and reasonable assumptions, and they line up with the data we see in various places, but this does not protect against errors in those assumptions.

Despite that, I do think this was worth some amount of update, and my main changes are:

  1. More confidence that the broad path of things will be what I expect.
  2. More confidence that hospitals will hold together and death rates won’t be so bad.
  3. Moderately higher expectation for the true peak of cases, closer to their 6 million per day.
  4. Moderately higher expectation about speed and magnitude of decline from peak, although I still don’t think they’re in a reasonable ballpark.
  5. More confidence in IHME and Delphi to give us reasonable projections.
This entry was posted in Uncategorized. Bookmark the permalink.

56 Responses to The IHME Report

  1. Humphrey_Appleby says:

    Anecdotal evidence time: my last PCR test (part of a university surveillance testing program) was last Thursday. I still haven’t received results. If it takes until tomorrow for results to arrive, then the test will have provided exactly zero useful information, since even if it is positive, the `isolation period’ will have ended by the time the test results arrive.

    I also got both a rapid test and a PCR test for my toddler last Wednesday, for `back to daycare’ purposes. The rapid test results took two days to arrive. The PCR tests still haven’t arrived, so well into `fully useless’ territory.

    When test results start taking five days or longer to come back, we might as well stop testing entirely…

    • Nick H says:

      Seconding that anecdote: I was exposed last week and took two PCR tests two days apart – Thursday and Saturday morning (one for compliance with the 5 day isolation + negative PCR test rule at work, one for the sake of elderly family members I saw over the weekend). Dropped them off at the state lab building. I didn’t get the Thursday PCR results back until Saturday evening, and still haven’t received my Saturday morning test results. So looking at a 50+ hour turnaround time at minimum, and that’s being generous.

    • Steve says:

      According to two testing health specialists I spoke on the phone with after my PCR test failed to come back, PCR tests are taking 96+ hours to process in Florida, which makes them effectively useless, and rapid tests are difficult to come by, whether walk-up, drive thru or at home self-tests. This is completely unacceptable.
      It’s too late to do anything now because our government at all levels critically failed us when we needed them most. These are people we elected on campaign promises to do better, and so far, they haven’t.

  2. lunashields says:

    About “10% of people being down with covid, and not showing up in mobility decreases looks wrong”. Anecdotally, a significant number of my friends had covid recently. They all treated it exactly like a regular cold – meaning continuing to go partying, visiting friends, flying to other countries even (conveniently omicron often doesn’t show in nasal swab, so you do that with a test, and you can fly). So perhaps the lack of mobility decrease is completely real.

    • Andrew Currall says:

      Also, at least half of people with covid probably don’t know they have covid, and if the estimates quoted above of ~90% of cases being asymptomatic are anywhere near true (90% seems high to me compared to other estimates I’ve heard, but hey ho), it could be much higher than that. If 90% of covid cases are asymptomatic and most people don’t test asymptomatically (which we know they don’t because daily tests are way too low for that) then the overwhelming number of cases are undetected, so can’t update their behaviour.

      I am actually deeply sceptical that the detection rate is anywhere near 50%, let alone the +50% figures being quoted in some of those graphs. I doubt even the detection rate of symptomatic cases is 50%, frankly.

      Also, single data point, but while I’ve never had a postive covid test (I’ve done a few rapid tests), I doubt I’d update my behaviour much if I did get one.number of cases are undetected, so can’t update their behaviour.

      I am actually deeply sceptical that the detection rate is anywhere near 50%, let alone the +50% figures being quoted in some of those graphs. I doubt even the detection rate of symptomatic cases is 50%, frankly.

      Also, single data point, but while I’ve never had a postive covid test (I’ve done a few rapid tests), I doubt I’d update my behaviour much if I did get one.

      • Triskele says:

        Yeah I choked on my granola this morning reading the 90% estimate. It wasn’t clear to me if that was ‘it can get up to’ because of an inevitable failure in test capacity or ‘it is’ because of real lack of symptoms.

  3. Spiralling Depression Time says:

    Zvi, I initially assumed that the number of cases after the fifth wave (January-March) will be very low, and that it would be possible to forget about the pandemic in late spring – but I’ve just learned that you “still expect cases to remain elevated (above pre-Omicron levels) for several months.” I thought I would endure the next ~8 weeks in isolation, and then return to 100% normal mode. Now it seems that in order to avoid the infection, I would need to add a couple of extra months, basically losing another summer, without knowing what will start happening in fall/winter (new variant, new omicron wave?).

    Is it continued full self-isolation vs. the high risk of infection for the next few years? :(

    • lunashields says:

      It will be endemic. Meaning if you _absolutely_ don’t want to catch it, you will need to self isolate for the rest of your life. Consider how often do you get colds/flus, and think if you can possibly avoid something like that. You will for sure get it during the course of normal life at some point, no matter the average number of cases.

      Alternatively you can self isolate until widespread availability of paxlovid, which will hopefully happen before the summer.

      • Spiralling Depression Time says:

        I don’t catch colds/flu often, but they are 5-10x less contagious, and >10x less serious/deadly. :(

        I hoped that the omicron wave could result in a massive decline in cases by ~April, making late spring and summer enjoyable, and that the fall/winter season would bring an even milder strain + better treatments, enabling us to basically forget about the pandemic. But it seems that the dystopia will continue.

    • Tom W says:

      Frankly, I would assume at this point that your meaningful choices are “isolate for an indefinite period of time into the future, between several months and the rest of your life” vs “get as vaccinated as you reasonably can, hold out till after the upcoming peak if you want to be really cautious, and then go about your life as if the risk of a bad outcome for you is close to zero”.

      Omicron will fade in time, but there’s always going to be further strains, there will always be seasonal ups and downs, the non-China world has given up on “Covid Zero” by now (not that it was ever a realistic possibility), and the rest of the world, given enough time, is bound to adjust to that reality and act accordingly. There’s no “end state” other than “people decide it’s not worth it to inconvenience themselves about the virus anymore”, and there really never was.

      • A1987dM says:

        Of course, but the question is how long “till after the upcoming peak” is…

        • Tom W says:

          I don’t know of any credible estimates that have a peak happening after mid-February, at the very latest. There’s more of a question as to how steep the downward slope will be, but a steep upward slope implies a steep downward one too.

    • TheZvi says:

      There are scenarios where by end of March you can pretty much act normally even if you care a lot about Covid, but there are also scenarios where that’s not true. Through March to me counts as several months, and my ‘baseline scenario’ has April being much better than now but not especially safe.

  4. Zvictoria Beckham says:

    Hey Zvi!

    A) Do you expect Paxlovid to be universally accessible in the EU by April or May 2022?
    B) Do you expect Paxlovid to remain ~90% effective against the future variants?
    C) If the answers to questions A & B are “yes”, then could it serve as an alternative to boosters?
    D) Are two Pfizer shots taken in 2021 likely to continue offering ~80% protection against hospitalization and death in the next few years (in case of the infection with current and future variants)?

    Thank you so much in advance for the replies!

  5. Kate says:

    Testing demand increased really fast. There were piles of rapid tests at our closest Walgreens on Dec 23 and by the 30th nothing anywhere and lines down the street at PCR sites. So if that chart ends with Dec, it’s probably not wrong.
    NYC was probably a bit ahead of the schedule elsewhere.

  6. Icebreaker says:

    Since it seems to be a critical point of data in the IHME model for estimating the true number of infections it would be very interesting to learn how they arrived at the conclusion of 80-90% asymptomatic cases.

    Although not having the math skills to prove my point I’m deeply sceptical because there’s a history in this pandemic of overestimations of asymptomatic infections (AI). In the first wave many experts believed there was a big or even huge backlog of undiscovered cases that would result in a good level of immunity in the places that were worst hit. But as we all know population studies for antibodies showed that the real percentage of infected people were far lower. The second wave one year ago sadly proved this to be true.

    This relatively recent meta-analysis ( https://jamanetwork.com/journals/jamanetworkopen/fullarticle/2787098 )concluded the number of AI before Omicron could be around 40%. But they admit there’s several limitations involved. One is the well-known fact that many who test positive with a PCR asymptomatic will experience symptoms later. Another meta study found no more than 13% asymptomatic individuals with COVID-19.
    We also have this superspreading event in Oslo, Norway that I believe has been referred to earlier in this blog. Here’s an article: https://www.eurosurveillance.org/content/10.2807/1560-7917.ES.2021.26.50.2101147 I quote:

    “In total, 111 out of 117 attendees (95%) participated in the interviews. Respondents had an average age of 39 years (SD: 9.2; median: 38; range: 26–68) and 48 (43%) of them were women. Most respondents (n = 107; 96%) were fully vaccinated. Eighty-nine percent of the respondents (n = 99) had received two doses of mRNA vaccines. None reported having received a booster dose. All respondents reported having a negative rapid antigen self-test taken at home or PCR within 1–2 days before attending the event. Eight (7%) respondents had previously had COVID-19, but none in the previous 4 months, according to information gathered through the interviews.

    Of those 81 were confirmed or probable (PCR-positive only) cases. Only one of this cases reported no symptoms at all and 91% (74 persons) reported at least three.
    Although just one incident – doesn’t it seem very unlikely that if up to 90% of infections overall were asymptomatic over 90% experienced symptoms in this setting? Especially if you consider they hade a low median age and were mostly fully vaccinated?

    • lunashields says:

      I suspect they might be using a definition of “asymptomatic” which is “low enough not to bother with testing” – couple of days of sniffles and slightly sore throat or some such

    • lunashields says:

      Interesting anecdotal report of an intentional “super spreader event”. A friend participated in a multi-day tango marathon in Lviv. This is more or less perfect setting for the spread – since you dance cheek to cheek to someone for about 10 minutes, and you change partners after that, and it goes on for hours.

      They had asked attendees to do tests every morning (and gave out free ones), at the end they had 45 people out of 450 reported positive (and I’m sure a large number also was positive, but decided not to report it due to lack of symptoms to continue dancing).

      Everyone who got it had it easy cold-like (and pretty much everyone is vaccinated and a lot of people had had covid on top of it too, this is not a crowd that’s concerned with social distancing and the likes).

      So on a surface 70-90% asymptomatic sounds fairly plausible.

  7. Bayesian Businessperson says:

    The 90% estimate for “asymptomatic” in the model is obvious nonsense. I have a few dozen employees using rapid tests *every day* and so far all our 10 positive tests have come after mild symptoms appeared. Some are very mild, but definitely real. If there were large numbers of asymptomatic cases we’d be finding some of them. If they mean “very mild boring cold symptoms” they should say that, as it’s a very different claim.

    • lunashields says:

      Majority of my friends tested negative for a few days with obvious symptoms, when swabbing the nose. Presumably for asymptomatic cases load is even less, so nose swab will definitely be negative. Unless you’re doing throat pcr swabs, how would you even know you’ve had asymptomatic cases with a rapid testing that you do?

      Don’t get me wrong, I’m NOT saying you’re doing testing the wrong way. You’re doing it correctly for the purpose of detecting contagious cases. But for the purposes of detecting all infections for “official statistics”, it’s clearly not good enough.

      • Bayesian Businessperson says:

        We switched to throat swabs a while ago. They do seem a bit more sensitive, turning positive earlier after symptom onset in an omicron-specific way. It’s possible that some people are getting asymptomatic infections that are fought off *so* quickly and with *so* little replication that there’s never enough virus anywhere to trigger a rapid test, but at some point it’s not really an infection, just antibodies killing some stuff that wandered in, right? I’m at least confident that transmission is not driven by this asymptomatic dark matter in any way, or else we’d at least sometimes get an asymptomatic positive throat swab, but but also rather curious what the minimum level of infection is that actually provides meaningful immunity in the future.

        • Andrew says:

          Yes, this is actually a key point. What exactly counts as an “infection”? It seems to me that the answer depends on precisely what you’re wanting to do with the results, and could vary widely. One obviously useful definition is sufficient viral load to trigger a meaningful immune response. Another is sufficient load to have some chance of infecting anothers. Another is just exposure to some quantity of the virus in your body, whether or not this is immediately purged by the immune system or not. They will be very different, and none of the definitions is wholly objective.
          90% asymptomatic cannot possibly be correct for the “infectious” definition. I think it’s extremely unlikely to be correct for the “meaninful immunity” one either. But it might be correct for the “exposure” definition.

  8. D. F. Linton says:

    One factor in testing reporting I’ve not seen discussed, is the fraction of at home tests that go unreported. If you home test negative, one must be especially motivated to bother to report. If you test positive, the attractiveness of reporting is driven by the expect consequences. If you have some important-to-you plans, then discarding the test as “probable false positive” might be very attractive.

    • lunashields says:

      It’s not even that. For most people there’s absolutely zero value in reporting a positive case, important plans or not. I don’t know a single person who reported their home test anywhere, why would you bother?

  9. Pax Lovid says:

    Zvi, the pandemic situation won’t get substantially better in the next few months, we won’t get Paxlovid in the next few months, there might or might not be a period of relative relief from June to late September (?), but who knows what will start happening in fall 2022. This is a virus that’s over 5x more contagious than cold/flu, and >10x deadlier/more dangerous. My family has a cumulative 3% death risk and a 12% hospitalization risk. I really don’t want to catch it. I felt really bad after each shot. The pandemic will continue straining the economy and social fabric.

    I don’t know if I’m overreacting as a relatively young and healthy person, but it makes me seriously depressed, and it feels like the dystopian vision for the next months and years justifies it. Am I really a sensitive outlier?

    • Exa says:

      You’re (probably) not going to live forever regardless – how does that 3% death risk compare to the QUALY loss of indefinite isolation? Being cautious just until the omicron peak is winding down (and then, once hospitals have spare capacity, possibly even deliberately catching it) is plausibly justified, but you shouldn’t give up more than 3% of your lifetime value to avoid a 3% risk of death.

      • lunashields says:

        I also seriously suspect that 3% number, especially now after vaccines, improved treatments and milder omicron. It is likely that it’s a very pessimistic calculations based on the original strain numbers, and the most pessimistic interpretation of every other improvement (by the vaccines, better treatments, etc).

      • Pluto says:

        That’s right, but things get complicated if 1) you’re a nerdy guy with limited social opportunities, so staying at home is not such a big loss, 2) you want to lead a very risk-averse lifestyle to maximize chances for accessing the radical life extension or mind uploading when you’re old, 3) you continuously hope that “it’s just <12 more months and we'll have a full peace of mind thanks to the mutation/new vaccine/new treatment". Sometimes all 3 factors play a role to a different degree.

    • TheZvi says:

      I mean yes it will get substantially better within a few months, in percentage terms, it simply won’t be 90%+ better than now for a while.

      If you feel like it’s a dystopian nightmare then you should probably move out, let yourself catch Covid by acting normally, recover, and then keep acting normal and move back in if you want? Or something?

      I realize living at home would otherwise be a reasonable thing to do in your mind, but if it means taking extreme precautions such that you are living in dystopia, presumably that sucks more than paying rent?

  10. Question says:

    Does someone know how long the body needs to develop antibodies after a vaccination shot? If I got the first shot on Saturday, is my risk of a severe case already significantly lower now or does it take longer for a noticeable downshift in probability?

  11. bakkot says:

    Given that deaths/day is _currently_ at 1600, and new cases/day is more than 5x what it was 21 days ago (or 3x what it was 14 days ago, for a more conservative estimate), I have a hard time seeing how we make it out of this with a peak deaths below 3000/day, much less 2000 as they predict. That seems obviously wrong. (It wasn’t obvious when they wrote their report, but it still provides a quick check on how their prediction is holding up.)

    I’m also confused by their claim that deaths will peak “by the end of the month”, given that the graph which compares model projections has their model predicting peak deaths a week or two into February (which is more consistent with their projected peak new cases).

    • Icebreaker says:

      At least two things come to mind:

      a) Many of the deaths reported now probably occured say 1-2 weeks ago. Many of those patients probably had the Delta variant which is believed having a higher IFR

      b) With Omicron a much larger part of the infected are vaccinated people (2x or 3x). Fewer of them will experience severe Covid and even fewer will die.

      This doesn’t prove that the projections are correct. Time will that in the weeks to come I guess.

      • bakkot says:

        > With Omicron a much larger part of the infected are vaccinated people (2x or 3x). Fewer of them will experience severe Covid and even fewer will die.

        I don’t think that actually ends up mattering much to the analysis, given the very large increase in total cases we’re talking about here.

        To put some numbers on it: the vaccinated were (ballpark) 5x less likely to get infected than the unvaccinated during the Delta wave, and the U.S. is about 2/3rds vaccinated, so the vaccinated were previously 2/7ths of cases; if we assume protection has decreased to just 2x less likely, that’s 5/9ths.

        But total cases have gone up 5x in the last 21 days. If we assume the unvaccinated were previously 5/7ths of the total and are now 4/9ths, the number of unvaccinated people infected has gone up ~3x.

        So taking even very conservative assumptions: supposing all the deaths reported yesterday were delta and omicron is only half as deadly as delta in the unvaccinated, so you expect the rate of deaths 21 days from per now case today to be half what it was 21 days ago, you’d still expect deaths to go up by 1.5x over the next three weeks _just from the increase in cases we’ve seen in the unvaccinated_.

        Yesterday the 7-day average number of deaths/day was 1646. If that goes up a mere 1.25x that will still blow the 2000 deaths/day prediction.

    • The article is behind a paywall, and I’m not going to grease the palms of the WSJ for access.

      Also, I’m always a bit suspicious of the “let it burn through the population” school of thought. The price for that can be very high indeed, as an article in Current Biology pointed out.

      So let’s consider who the authors are, and whether we think their qualifications or previous positions would tilt us toward believing them or not. I don’t want to be ad hominem here; I just want to see if their other writings make me want to read this article or to walk away from it.

      The authors are husband and wife, which influences us in neither direction, but which I find slightly charming. So… a small number of points for that. (Yes, I may be being slightly irrational here, but I prefer to think of it as starting out with a charitable attitude.)

      Apoorva Ramaswamy is an MD, and an assistant professor of medicine at Ohio State. So that’s a point in her favor: she has some medical chops and thus knows her way around patient care.

      On the other hand, she’s a laryngologist, specializing in swallowing disorders. So her particular area of expertise is unlikely to be relevant to public health policy around Omicron.

      The other author, Vivek Ramaswamy, is another sort entirely. He’s a lawyer (JD from Yale, so a smart lawyer), and a founder of Roivant. Roivant appears to be sort of a meta-company: its business is to create other businesses, either through in-licensing or a recent SPAC merger. The SPAC stuff makes me a little leery, as does his claim that he can solve drug development problems with technology. (As though both big and little drug companies don’t already know the relevant technology?!)

      In terms of his writings, he makes me even a little more leery, since he is fond of riding the usual conservative hobby horses. Wikipedia says: he’s the author of a book complaining about “wokeness” in the workplace, he likes to speak against shareholder capitalism (i.e., more power to management), he thinks big tech is being censored, and that critical race theory is somehow a problem.

      So for me, the article looks pretty suspicious:

      (1) It apparently advocates the “let it burn” through the population strategy, which seems to me at best doubtful.

      (2) One of the authors, while and MD and med school professor, is a specialist in an irrelevant area (though she seems otherwise worth listening to).

      (3) The other author appears to be: (a) fond of getting attention through conservative hot-button issues, (b) invovled in SPACs which at least currently raise my hackles, and (c) runs a company which appears to be less oriented toward science and more toward thinking that dumb people need firmer guidance by managers like himself.

      Based on that much, I’d look elsewhere.

      • Tom W says:

        Sincere question: what alternative(s) are there to “letting it burn through the population”?

        No one credible, as far as I can tell, seems to think that Covid-19 is going away in the next year/decade/century/whatever, what with being highly infectious, quickly mutating, difficult to identify without a dedicated test, and also that whole thing about how it infects several species of animals.

        Vaccinating as many people as feasible seems like a pretty obviously good thing, but, well, at this point we’ve seen it “burn through” some pretty vaccinated populations, and I would expect the next mutations to face even more selection pressure to evade acquired immunity (of any sort), so it also doesn’t seem like that’s a magic bullet.

        And, well to the extent that the rest of the NPIs do work (which seems questionable for many/most of them), they stop working as soon as we stop using them. Are we willing or able to keep masks on schoolchildren forever? Are rolling bans on various gathering types, in various locations, just a thing that people will live with for the indefinite future?

        So in the long term, what feasible outcomes do you see other than “Covid burns through a large percentage of the population”, and how exactly could we achieve them?

        • It’s true we don’t have much choice at this point for the Omicron wave. But that’s a result of having made bad choices in the past: low vaccination rates, low boost rates, low mask usage, low use of fluvoxamine, etc. All our policy levers are pretty much disconnected from our future light cone at this point.

          But to me there’s a big difference between (a) acknowledging that our past bad choices have baked in a bad short-term outlook, versus (b) deliberately choosing a bad policy of self-imposed helplessness again in the future. We can learn from the former; the latter is a refusal to learn.

          Having said that, you are correct to point out that there is not much we can do now. I even said more or less that yesterday (scroll/search to the bottom under “The Scenarios”).

          But I should have mentioned it here, and thank you for reminding me to do so.

        • Tom W says:

          Thanks for answering! Sorry, I could have clarified–as far as I know there’s no one who thinks we can do much about Omicron. But you say we shouldn’t “deliberately [choose] a bad policy of self-imposed helplessness again in the future”, so that’s the part I’m curious about.

          In the long-term future, stretching to the indefinite–what good policies would result in a better long-term outlook for the next few waves, what would it take to accomplish them, and how long can/should they last? Forever is a valid answer, but I feel like you ought to be explicit about it if so.

        • @TomW:

          …what good policies would result in a better long-term outlook for the next few waves, what would it take to accomplish them, and how long can/should they last?

          Interesting question. Here’s my not terribly well thought-out answer, so that we may all be thankful I will never be permitted near the levers of power. :-)

          Short term (at most a few years): Widespread vaccination, especially with variant-specific vaccines now under development (both Pfizer & Moderna have programs). Widespread production, storage, and free availability of drugs like paxlovid and some of the “fast-followers” currently in development. Probably some use of mandates to achieve those things. International negotiations to make it happen world-wide to put downward selection pressure on new variants.

          Long term: Development of pan-coronavirus vaccines (see YLE here, US Army here, and Science article here). Also continued research on anti-virals, focusing on viral targets not found in human (e.g., the 3CLpro target of paxlovid). Most of all: public education about public health policy, in which we finally teach ourselves that we can’t focus just on our own health, but also must consider the health of all those around us. “Around us” means more or less the planet, in the case of coronavirus variants breeding.

        • Tom W says:

          Thanks again! I broadly agree with much of what you’re saying here. Would be curious how you think other NPIs like masks, travel restrictions, business closures, etc fit into these policies (if they do at all) but you’re under no obligation to respond quickly (or at all) and if you end up putting it as it’s own post, wherever, I’d be happy to go read it :)

          The real subject of my curiosity is that I hear a lot of folks talking about how obviously necessary various NPIs and emergency measures and Rules and Restrictions are right now, or in the short term, and how barbaric [insert public figure here] is to oppose them, but I’ve yet to hear any kind of plausible answer for “when might they not be Necessary and Virtuous and Obvious”, and, well, as someone who would prefer they not last forever, that makes it hard for me to support them over any time frame.

        • Thanks again! I broadly agree with much of what you’re saying here. Would be curious how you think other NPIs like masks, travel restrictions, business closures, etc fit into these policies (if they do at all)…

          Honestly, I have trouble looking out that far. I’m your basic science nerd, whose political judgements are maybe not something worth anybody else relying on?

          One of the good effects of the pandemic has been that we’ve realized video calls can substitute for a lot of formerly in-person interactions, such as a lot of business travel. Lots of business people are enamored of their own judgement of character, and thought their personal presence was essential. Well… helpful, maybe, but not required. And in Japan, masks are just normal — the pandemic has made people emphasize masking a bit, but it’s no big thing to them culturally.

          On the other hand, I really want to go to my favorite Club Med, and not just by video call! (Yes, I watch YouTube videos of that particular Club Med. No, it’s not a good idea and I should know better. :-)

          So I get that there are problems.

          And I get that people worry pandemic measures might slide into permanence. I remember pre-9/11 days before the TSA occupied every airport, and before cops decided under the Patriot Act they could pat me down in the subway if they didn’t like my haircut (which they often do not; I don’t care for their haircuts either, but that’s somehow not relevant). Heck, I remember the 1970s when you could just walk in to an airport, go to the departure lounge, and watch airplanes take off & land. (People actually did that, though I never quite saw the point personally.)

          We’ve failed to repeal any of the anti-terrorism stuff from 9/11 or restore some of the privacy rights and civil rights abridged then, so the worry that we might not let go of pandemic rules is certainly not baseless.

          But unlike terrorism, which is sort of a qualitative political judgement, pandemics are more amenable to quantitative thinking by a wide variety of medical people. When they reach consensus that “it’s over”, then it’s over and we will need to apply intense political pressure to our leaders to get back out of the way.

          Sorry that’s not any more insightful, but it’s all I got right now. Maybe somebody else can think of something better.

        • Tom W says:

          That’s great, thanks again :)

      • Basil Marte says:

        I think I can reconstruct the argument just from the part that pokes out above the paywall, assuming the article straightforwardly expands the thesis. It’s that more stringent restrictions (including vaccinations) increase the selective advantage that a nastier variant has — which is true but irrelevant, because the implications usually drawn from that don’t apply.

        The problem is that standard evolutionary biology assumes a constant population size when it derives that the probability of reaching fixation increases as a function of selective advantage. For viral variants, it is reasonable to assume that policies generally move all variants’ R:s in the same direction, and thus the probability of variants with tiny population sizes (e.g. of 1) quickly going back to extinction is higher under a more stringent policy even if they theoretically have R>1, independently of how their selective advantage changes due to the policy.

        Moral of the story: trying to base policy on N:th-order arguments is generally a bad idea, because even the sign of the effect is uncertain, never mind the magnitude.

        • Moral of the story: trying to base policy on N:th-order arguments is generally a bad idea, because even the sign of the effect is uncertain, never mind the magnitude.

          Ooooooh… that’s well phrased! I like it.

        • Basil Marte says:

          Well, I quite missed their argument. It was even nuttier than I expected.

        • Sadly, yes.

          Though given the background on who the authors are, we should all have been suspicious. And it was the WSJ op-ed page.

          Kudos to you for thinking the best of them and steelmanning what their argument could have been.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s