Covid 8/4/22: Rebound

I’ll be at Future Forum over the weekend. If you’re here, come and say hello. I don’t plan on scheduling a lot of meetings because I hate scheduling, but talking to people is the plan. I have a few people that I know I want to see, but mostly I’m going to be improvising, and whatever happens happens.

Because of the trip I wrote most of this earlier, and all I did Thursday was fill in the numbers section and check for anything big from overnight – some stuff likely got shifted to next week as a result.

It was probably a good week for a trip, as Covid news continues to quiet, to the point where the majority of this week’s post is about non-Covid things. So far I am mostly fine with this transforming gradually into a weekly roundup. Indeed do many things come to pass.

Executive Summary

  1. Covid continues on its expected trajectory.
  2. Monkeypox continues on its expected trajectory.
  3. Nuclear reactor design approved, hope springs eternal.

Let’s run the numbers.

The Numbers

Predictions

Prediction from last week: 850k cases (+5%) and 2,850 deaths (-3%).

Results: 730k cases (-10%) and 3,054 deaths (+3%).

Prediction for next week: 750k cases (+3%) and 3,000 deaths (-2%).

Adjustments were the reverse of what I expected, but everything involved was relatively small. Everything points to a stable situation, with reporting errors and gaps likely to be most or all of the story.

Deaths

Cases

Included 10k presumed cases in South Carolina, since they did not report, along with standard Florida adjustment.

The drop in the West isn’t mostly due to one particular state although some (~20k) of it is Nevada being obviously too high last week. I am continuing not to make adjustments that aren’t super obvious, especially retroactively. Better not to put one’s finger on the scale even when it’s probably in the correct direction.

Physical World Modeling

President Biden took Paxlovid early, and got a rebound case of Covid for his trouble. He then continues to report feeling completely fine during the rebound. If you test positive for a Covid rebound, but wouldn’t have noticed without the test, should you care? In terms of preventing further spread you care a bit, but if nothing changes for you then nothing changes for you. The extent to which Paxlovid is being postponed to minimize rebounds reflects the degree to which what people actually want is felt social permission to not insolate and they are not so afraid of the disease as such.

Get well soon, Joe.

You might dislike the bit but you have to admire the commitment.

Joe tested positive again on Wednesday, but continues to feel fine (except for a bit of a cough) and has done a ‘light workout.’ I would not be working out if I was him, even if he feels fine.

John Mandrola’s takeaway from Biden’s treatment is about worrying about the differences between clinical trials and practical conditions, and emphasizes the role of Omicron in the effectiveness of Paxlovid, then looks at similar examples with other conditions. The general point is well taken. I am guessing the rebound issue is more about getting people Paxlovid faster than about Omicron being different, but in practice it matters not.

Another study takes untreated Covid patients who did not get Paxlovid, and finds viral rebound and symptom rebound in 12% and 27% of participants respectively, 10% of the 27% being after all symptoms had gone away. This suggests a strong possibility that all this ‘Paxlovid rebound’ worry is more ‘this is what happens when you are looking for rebounds’ than it is ‘there are a lot of rebounds caused by this.’

Trevor Bedford notes that the seasonality effects of Covid are increasingly clear, and that it seems plausible we end up with a seasonal waves indefinitely. That’s seemed plausible for a while, and Trevor is playing it safe (as in ‘not making a prediction.’)

Bob Watcher notes two key things here, both that this case might discourage use of Paxlovid for the wrong reasons and we should try to minimize this, and also that we still haven’t done a trial of 7-day versus 10-day course and we should do that. Alas, that sounds like the type of information it is illegal to figure out, so we likely will never know.

Can anyone verify whether or not this source works especially for Paxlovid? Would be good to have a simple answer to tell everyone to ensure they get access.

This week’s ‘Long Covid is devastating the labor market’ slide deck is out, claims 9% decline in labor force participation rate a year after Covid. I once again suggest doing the math on the population-level statistics.

Meanwhile, the mask debates continue as overly cautious people debate other overly cautious people the way medieval Christians used to debate the finer points of doctrine while sharing quite a lot of baseline assumptions. I find it interesting that Wachter has a mask he thinks is best, a fitted N95, but often wears a KN95 instead because the fitted N95 is too uncomfortable.

Here’s his bottom line.

In case it’s not clear, I think this is not a reasonable reaction.

In Other News

Take this story for what it is: I saved a link to a Tweet talking about how the CDC suppressed information. When I went back to look in detail, the account at the top of the thread had been suspended by Twitter. One screenshot still remained of various posts labeled ‘misinformation’ that the CDC wanted suppressed. By ‘misinformation’ they seem to mean ‘give an impression the CDC thinks will be harmful and misleading’ rather than ‘says that which is not.’

Washington D.C. has a vaccine requirement for students 12 and up, which would currently mean that 40% of black students get sent home.

“Our goal is that no child should miss a single day of school,” Asad Bandealy, the chief of the D.C. Department of Health’s Health Care Access Bureau, said at a news conference this week at Mary’s Center, a community health clinic where children can be vaccinated. “And that means we need to get started now.”

Yeah, that is not exactly a realistic goal.

Then again, maybe it is?

D.C. has a long-standing reputation of failing to enforce its immunization requirements in schools.

We need to get started on that today.

It can always be worse.

Elsewhere in the country, the New Orleans public school system in February added the coronavirus vaccine to its list of required immunizations for children 5 years and older.

Meanwhile, just 31 percent of children nationwide between the ages of 5 and 11 have been fully vaccinated.

Multiple friends of mine, including Brian Kibler here, call for TwitchCon San Diego to impose a mask requirement, as San Diego ComicCon did, with the other saying ‘if you attend without a mask I won’t see you for a year because I don’t trust your decision making.’ In a world of cosplay I presume showing up in a P100 is always an option. I notice that the requirement to mask makes me a lot less excited about New York ComicCon, but also I likely will need to go for work and this is a pretty damn good way to expose a lot of people, so both ends of this make reasonable points. There being a mix of cons with different rules doesn’t seem so bad.

Claim from UNICEF via BBC that in South Asia maternal and child deaths due to healthcare disruptions (239k) caused by Covid exceeded deaths caused directly by Covid (186k). Also claims 3.5 million additional unwanted pregnancies due to poor contraception access. Regardless of the details this is a good reminder that the reactions to Covid and the resulting secondary effects were the primary story, while the actual effects of Covid were in a relative sense a side show, and still are.

New York Times continues to apply the label ‘anti-vaccine’ to those who oppose mandating Covid vaccinations for children, or otherwise oppose any public health intervention. This is a problem.

A foolproof two-point plan to not get Covid: Have everyone wear a mask, in 2004.

A study estimates college mandates saved ~7,300 lives, reducing total national Covid deaths in fall 2021 by 5% (PDF). They take a ‘interesting’ approach.

The identification assumption underlying our approach is that areas without college vaccine mandates provide an accurate counterfactual for what COVID-19 outcomes would have been absent the college mandates.

Of primary concern is that counties with colleges that decided to implement vaccine mandates have other protective policies in place or have populations that were otherwise more likely to be vaccinated or that more stringently followed public health advice. There are strong geographic and political components to vaccination rates, public health measures, and compliance with these measures (Murthy et al. 2021; Yuan et al. 2021; Gollwitzer et al. 2020; Ye 2021; Clinton et al. 2021; Adolph et al. 2021; Fraser, Juliano, and Nichols 2021). Hence, we include not only county and calendar week fixed effects in our models but also region-byweek fixed effects, interactions between week fixed effects and the Democratic vote share in the 2016 presidential election, and interactions between week fixed effects and the baseline county vaccination rate (measured during the last week of June 2021, before any college in our data began fall classes). Furthermore, we present evidence that our results are (1) robust to controlling for differences in colleges’ masking and testing policies, and (2) unlikely to be confounded by other policy or behavioral changes that alter community-level vaccination rates or mobility, as neither of these outcomes differentially change in counties with vaccine mandates after college semesters begin.

Some results reliably were strangely absent anyway, but deaths are still gold standard:

Hospital-based measures, such as the number of ICU patients and hospitalizations, contain error because they are based on hospital locations rather than patients’ counties of residence. Our findings for these measures are generally statistically indistinguishable from zero, though ICU estimates are consistently negative. Deaths are measured by county of residence and likely contain less noise. We document that vaccine mandates led to 5.4 fewer deaths per 100,000 in the county, an estimate that is also significant at the 5% level.

My response: No. You can’t do that. You can’t pretend that colleges that implement mandates are otherwise similar to colleges that don’t, and you definitely can’t pretend that the surrounding counties are similar, when it comes to these implications. This is all quite absurd.

That does not mean the result is not useful. It is a sort of an upper bound estimate of the impact of all the differences between these two types of areas. As a rough guideline here, college towns have 18.2 million people in the United States, there are about 19.4 million college students at a given time, and about half of colleges had a mandate. Their estimate of the value of the intervention is $9.7mm to $27.4mm per 100,000 residents, or $97 to $274 per resident (average ~$185), which seems so much more intuitive to me. So, is a generally cautious approach to Covid over a year versus a less cautious one, broadly construed, worth more or less than that range?

If you do think this is purely the mandate and you fully accept the result, and presume that a mandate results in ~30% of college kids getting vaccinated (given 47% of residents aged 18-24 were fully vaccinated in general, that seems like a reasonable ballpark, since college kids are more likely to get vaccinated anyway) that puts the value of the externalities from an additional vaccination at something like $550 from deaths, let’s (very roughly) say $1,000 since getting sick without dying is also no picnic.

Monkeypox

Say hello to your new White House Monkeypox Coordinator Robert Fenton and Dr. Demetre Daskalakis as his deputy. They have over four decades of experience in emergency response and public health, including COVID response and HIV prevention. So we can expect that level of quality public health response here.

Still this is good. Better to have someone in charge than no one. Dr. Demetre has the endorsement of NYC mayor Eric Adams, who going by the replies is not having the best time on Twitter.

Airborne transmission of Monkeypox is essentially not a thing.

Letting people test other people, not as much of a thing.

If someone has consistently good luck, it aint luck. If we keep making the same catastrophic error, perhaps it is not an error given the right set of preferences.

Then again, this is the first reply, file under ‘understanding of inevitable exponential growth continues to elude most humans’:

Also one thing this definitely is not about is worrying we would be wasting money on making too many tests. For example, I continue to be confused why states can’t get along when it comes to stuff like this.

I do get why an ‘emergency’ doesn’t change the answer, since the characterization changes nothing and public health authorities have already decided and made clear over and over and over again that following arbitrary regulatory procedures trumps human life.

Systematic review finds the secondary attack rate for Monkeypox among household members to be 8%. Previous link reports that rate to be under 1%, which is crazy and I presume won’t be sustained, but still. The 8% is not zero but it is also not very high, and points strongly towards the disease not surviving in the absence of sexual promiscuity.

Thread of advice on what to do about your pets if you have Monkeypox, basically taking extreme precautions to avoid infecting pets. It’s a combination of extreme caution and calling for health-obsessed choices not being willing to make what a health expert would see as actual hard choices. One could compare this to the official advice to take much lower precautions when attending conferences and orgies, or Celine’s own advice in another thread where pets aren’t involved, especially if your pet is isolated from other animals.

The big worry is that Monkeypox will infect animals sufficiently to become endemic, but this is mostly going to depend on how widespread it is rather than getting a few people to take these kinds of precautions. The chance of getting this level of caution in general is zero and also it would wipe out Monkeypox rather quickly if we did get it.

This California report came out. Three quarters of cases have been between ages 25 and 44, of the 554 cases with known sexual orientation 92% were gay and 6% bisexual. There were 230 ‘unknown’ orientations so in theory the unknowns could be different but I can only think of one way to interpret this graph:

My perspective is still rather straightforward.

  1. If you are neither infected nor having contact with high risk groups (and the highest risk group would be people who are already positive and won’t isolate) then the risk levels involved are minimal.
  2. If you are contacting high risk groups, or especially if you are a member, take precautions, consider scaling back, and also probably get vaccinated.
  3. If you are having contact with someone with known Monkeypox, stop doing that.
  4. If you have Monkeypox and are having contact with others, stop doing that.

Yes, my bold four point Monkeypox plan mostly involves getting people who have symptoms of Monkeypox away from those not known to have Monkeypox until no one involved still has Monkeypox, and having those exposed get vaccinated. That’s how I roll.

As everyone says Monkeypox is the next big public health failure, it’s worth noticing that this is zero percent unexpected as measured by prediction markets. There was never a salient alternative path. None of the monkeypox markets have changed much, which seems right as there hasn’t been impactful news. Metaculus market on number of Monkeypox cases this month in California. New York market is similar. Deaths this year holding at around 500. Prediction for MSM share of cases holding steady at 94%.

If you’re not sure what to do next, you can always ask the audience.

What’s odd is that wasn’t even the question.

A different question continues to be the social justice implications of Monkeypox, or of talking about Monkeypox, as embodied by sweaty guy facing multiple buttons involving ‘help people in potentially stigmatized community who will otherwise get sick and sometimes die perhaps change their behavior to make that less likely’ and ‘avoid talking in way that might stigmatize that community.’

David Mack chose the first approach. Some of the internet disapproves.

This problem leads to a lot of confusion.

Image

Versus:

Image

I would say less ‘somehow don’t look’ as ‘are highly motivated not to notice.’

As in, file it all under This Is Not a Coincidence Because Nothing Is Ever a Coincidence.

Image

Not Covid

New nuclear reactor design approved, woo-hoo! It’s a small modular reactor first bound for Idaho.

Alex Jones knows how to put on a show.

There is video and it is glorious.

This piece on Jane Street seems unusually accurate.

What is School? as an ongoing series.

There is at least some value in teaching kids to hula hoop, helps with dexterity and physical coordination, teaches them how to get skills. It’s far from the craziest thing to do and might be better than many physical education classes and definitely is better than some ‘other’ things we teach them, and it seems fine that they later stop hula hooping. It does not mean I would judge its success purely via international hula hoop score comparisons, or that we should compel our kids by force to learn this hula hooping.

It’s still better to use a metric that has some value and isn’t that easy to fake than to use something easy to fake or that has zero value. Test scores in math and science fall into that category, and it isn’t obvious how to easily do better.

Meanwhile, Robin Hanson notes that no teachers charge students based on their level of improvement on tests.

Washington Post covers movement against alcohol, and towards a variety of non-alcoholic drinks as substitutes. I am in favor of not drinking, but I am also in favor of not drinking non-alcoholic drinks designed to mimic alcoholic drinks. This is a good way to drink a lot of calories and money for not much gain. I have a highly limited caloric budget, so this applies especially to me, but I think it’s true of most other people as well. Avoid drinking calories, water is great. When you do drink a few calories, keep it simple, nothing requiring explanation – milk, juice, tea, coffee.

What’s weirdest of all is seeing quotes like this in a country where three out of ten of us don’t drink at all, and a majority drinks quite rarely, so not-drinking seems rather normalized, and I’ve never had an issue politely declining.

Others see a drinking culture in America that’s slow to change. “I want not-drinking to be as normalized as drinking is, and we’re really not there yet,” Julia Bainbridge told me.

If anything, drinkers face a big tax, as they not only pay taxes as such but also provide lots of extra profits to restaurants. Here’s to you, my friends.

Tyler Cowen’s book Talent mentions that many successful people seem to drink diet coke, perhaps as ‘something to do.’ I don’t know what to make of that, but I don’t see how drinking water isn’t also something to do in the same way.

Otherwise, the desire to mimic drinking means you’re optimizing in all the wrong places. A good principle is that if you have to say what something isn’t, but is still trying to mimic, that’s something you do not want to consume. We don’t consume ‘non-alcoholic milk’ because that would only raise further questions. Similarly, I’m fine consuming vegan or gluten-free or sugar-free foods and what not, many are excellent, but if it prominently says it is vegan or gluten-free or sugar-free I assume it is going to be terrible and am rarely surprised.

Also, we must notice statements like this one:

As I delved deeper into the neo-moderation movement, something kept nagging at me: I’m all for personal choice, but are people really equipped with enough knowledge, information and support to do this by themselves?

The ‘this’ in question is not drinking alcohol, or at least drinking less. I mean, people, doing a thing, by themselves, good heavens. Or rather, not doing a rather generally destructive thing, all by themselves, without ‘enough information’ and ‘enough support’? Something clearly must be done.

Montana would suggest you drink milk instead, as long as it hasn’t been around 12 days, in which case they will insist it be thrown out despite there being no physical justification. Yes, it’s usually either taxes or fraud, but sometimes it is also regulatory capture.

Things that are going to happen one way or another but perhaps miss the point…

This entry was posted in Uncategorized. Bookmark the permalink.

9 Responses to Covid 8/4/22: Rebound

  1. Anonymous-backtick says:

    “Thread of advice on what to do about your pets if you have Monkeypox, basically taking extreme precautions to avoid infecting pets.”

    The jokes write themselves.

  2. sniffnoy says:

    I was at a party not too long ago where there was a guy who liked to make (alcoholic) drinks and was making drinks for people. I declined to take any on the basis of, I don’t drink. He was like, oh, I can make non-alcoholic drinks too!, and mixed me something up. It was bitter as hell! A non-alcoholic drink designed to mimic the taste of alcohol… @_@

  3. Anonymous-backtick says:

    “Otherwise, the desire to mimic drinking means you’re optimizing in all the wrong places. A good principle is that if you have to say what something isn’t, but is still trying to mimic, that’s something you do not want to consume. We don’t consume ‘non-alcoholic milk’ because that would only raise further questions. Similarly, I’m fine consuming vegan or gluten-free or sugar-free foods and what not, many are excellent, but if it prominently says it is vegan or gluten-free or sugar-free I assume it is going to be terrible and am rarely surprised.”

    Agree in most cases, but decaffeinated black tea (whether hot or iced) is excellent.

    • TheZvi says:

      Right. I consider this is a strange case where tea happens to have caffeine but has other advantages, so you can not filter usefully. Importantly you are not mimicking. It’s not my cup of… Ya know. But hey.

  4. captainjc says:

    I didn’t see anything on findcovidcare.com about getting Paxlovid? I did recently have luck with Plush Care and know others who have as well. They take your word about test results. They would have been $30 copay with my insurance if I hadn’t been too impatient to deal with their system, which didn’t work to book an appointment with out of state insurance. I just paid the $130 instead of waiting on hold.

    While I am at it, another anecdote about rebound. I am experiencing it now. Aside from myself, I know one other person (out of 2 people who I know who have taken it recently) who had it. Amazingly, Covid symptoms mostly went away hours after taking it. Tastes terrible as everyone says, mild stomach upset. 3 days after stopping, I started testing negative. Flew home 5 days after taking it (after 9 day extension of trip). At 7 days, got a bit of a runny nose and tested positive again. 9 days from last dose today and not feeling much worse, but isolating and really hoping I didn’t infect family came home to. In retrospect, I think I should have tested every day since returning home. If I lived with anyone immunocompromised I would not return home or stop isolating until at least a week after last does of testing negative (they say rebound is 2-8 days, so maybe longer).

    If they haven’t figure out dosing by then, the next time I get Covid, I will seriously consider getting two courses and taking it for at least a few days longer.

  5. hombre says:

    “I saved a link to a Tweet talking about how the CDC suppressed information. When I went back to look in detail, the account at the top of the thread had been suspended by Twitter.”

    That tweet: https://web.archive.org/web/20220728234553/https://twitter.com/MdBreathe/status/1552701968409313281

  6. Another study takes untreated Covid patients who did not get Paxlovid, and finds viral rebound and symptom rebound in 12% and 27% of participants respectively, 10% of the 27% being after all symptoms had gone away. This suggests a strong possibility that all this ‘Paxlovid rebound’ worry is more ‘this is what happens when you are looking for rebounds’ than it is ‘there are a lot of rebounds caused by this.’

    I dug into this study quite a bit, looking at the rebound rates for COVID-19 absent paxlovid. There are a couple oddities, but all the statistics that I checked indicated that their results were pretty reasonable. In particular, the probability of high viral rebound (enough to be a spreader) and high symptom score rebound (enough to be sick), happening together in the same patient, was pretty low.

    They used a lot of different criteria, but choosing a reasonably representative one I got upon reanalysis that the probability of meaningful rebound for COVID-19 without paxlovid was 1.15% (CL: 0.20% – 4.55%).

    So, not frequent, but not rare either.

    I also looked at another study on the rebound rate in paxlovid-treated patients. For them, the probability of rebound after paxlovid was 0.83% (CL: 0.26% – 2.26%).

    A test of proportion shows this is not a statistically significant difference, as you can tell from the way their confidence intervals overlap.

    Conclusion: It’s not paxlovid rebound; paxlovid has little to do with it. It’s COVID-19 rebound.

    And, in one of those notes of irony, after writing that, I tested positive with a COVID-19 rebound 5 days after finishing paxlovid. Because of course I did.

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 )

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