I think the evidence shows that the current Chinese COVID surge is going to fade. Although the national numbers are still increasing, that's due to COVID spreading to many different cities. Within each city, the case numbers plateau or drop a week or two into lockdowns.
These are the daily new case numbers for Beijing (upper line is asymptomatic cases, lower line is symptomatic).
Compare this to cities that haven't yet entered full lockdown (Shanghai) and still have acceleration in case numbers.
I would only expect COVID breakout if the daily increase continues a month into lockdown in a major city. Do you think that is physically possible? I genuinely don't know.
For me, the bottomline of this masking study is that if you wear a respirator only for a relatively small amount of time in a hospital setting, you might as well go maskless, because you'll just get infected when you're not wearing a respirator (because non-respirator masks don't work well at preventing covid due to poor face seals, inferior filter media, etcetera).
This week China experienced their largest protests since 1989, and according to many sources the most dangerous as well. China has painted itself into a corner. It cannot abandon Zero Covid, yet the costs of Zero Covid rise with each passing day and the policy is also slowly failing. What happens next? See the section below for further discussion.
Beyond that, there were a few more studies on a variety of topics, none of which represents any kind of major update. Cases did not drop off for the Thanksgiving holiday, which is odd and creates some uncertainty in the short term.
Executive Summary
Let’s run the numbers.
The Numbers
Predictions
Predictions from Last Week: 223k cases (-12%) and 1,850 deaths (-15%)
Results: 275k cases (+9%) and 1,779 deaths (-18%).
Predictions for Next Week: 290k cases (+5%) and 2,050 deaths (+15%).
Cases were up across the board, with no anomalies, despite deaths showing the holiday drop. I checked, and last year’s Thanksgiving did have the expected drop off on cases. A possible theory is ‘last year people took precautions and this year they didn’t.’ I have learned to be skeptical of such theories of dramatic impact of singular events shifting with the times. Mostly we have to wait and see next week to resolve the uncertainty. I do not see that much downside risk if the worst happens, I expect normality to continue even then, but it could still be rather not fun.
We are potentially close (Science magazine) to a universal flu vaccine via mRNA, as opposed to our current every-year only semi-effective one. This would potentially save tens of thousands of lives per year in America (290k-650k globally) and also save 10% + of people a few days of feeling terrible each year. We’re talking months of not-being-sick per person, even if we entirely discount what one might call Long Flu. There’s also promise for cancer, for HIV, for tuberculosis and for malaria. Some would say we might want to hurry.
Trying to be accurate is hard. Both Health Nerd and Zeynep this week were noting that one thing making it harder is the general attitude (in particular on Twitter, but there’s nothing here that is unique to Twitter) of turning on such folks the moment they start dismantling the wrong studies, in particular studies that are scary versus non-scary. Which means they get it from both sides.
This should not be weird. What is weird is what Nerd is attempting to do, caring about accuracy rather than taking sides. I briefly checked out his blog, seems like a good source of debunking if one is in the market for debunking.
The easier to implement and more common tactic, to this day, is to decide if Covid Scary or Covid Not Scary, and judge everything accordingly. Observe.
Angela Rasmussen analyzesthe new preprint case report referenced above that claims to have identified a bunch of outdoor transmission. Whole incident is super weird, starting with the original infection that, as Mazer notes, they claim came from an airplane a day after the source of the infection was on the plane. Which, as Angela says, isn’t impossible or anything, but it’s highly unlikely, and the logistics of the later infections also do not make much sense.
I continue to think the same way about outdoor transmission. It is definitely not impossible. It is still highly unlikely. Anyone who uses this to go Covid Scary Even Outdoors wrote their conclusion at the bottom of the page a long time ago.
It also points to the Chinese obsession with surfaces. It makes sense that a Chinese report tracking infections would be buying an infection from a day-old surface inside an airplane, suggesting the regime is not acting crazy when it totally acts crazy.
Zeynep reports things like people saving her admitted and corrected early mistakes and then quoting them out of context, or in many cases doctoring her writing, to paint her as sometimes a Covid denier or minimizer, other times a maximalist, to make her look bad in context.
My quick reading of the study, before seeing reactions, says that it was a good design that was somewhat underpowered. It seems to show that medical masks are likely to be modestly worse than N95s, but not dramatically worse, everything within margin of error. The reading above shares this view.
I do not get all the fuss about what conclusions the study intended to draw based on different answers. You can say you are checking for ‘non-inferiority’ all you like, your evidence is still your evidence and I am still going to update based on what it tells me. In this case, yes we can safely conclude that in practice medical masks won’t double a healthcare worker’s overall Covid risk versus an N95. We can learn considerably more here. We also can’t conclude that the decision does not matter.
Masks are not something our society was able to consider in context while weighing costs and benefits. Everything was absolute, on both sides, featuring a perfect top response.
Costs of masking? Misinformation, say Very Serious People. What will they make of this UK Government report?
They will, of course, do their best to ignore it, as they always do. Such folks do everything they can to suppress any evidence of masking harms or even any challenge to their presumption that such harms are impossible.
I wonder what the equilibrium is here. If work is remote, the pool of potential applicants is much bigger, since no one has to move. A 5-to-1 ratio of applications is high. It isn’t obviously wrong, especially if ‘applying’ is dirt cheap and even cheaper on remote jobs. For an in-person job they might ask you to (gasp) go somewhere and talk in person with a human during the application process. You might have to put on pants, maybe even a suit. Can’t have that.
We need to see the 2022 numbers, especially the second half of it, to get an idea of how sticky the changes will be from Covid. Either way this is a real long term trend and it is definitely not good for us. On a collective level I have no idea what to do about it.
Twitter is no longer enforcing its ‘Covid misinformation’ policies (Reuters). I am fine with this. The alternative involved censoring a variety of true claims, including ‘masks work.’ I do think it is important to worry about what information is given to people who do not follow the account in question, since Twitter by default (reminder: use Tweetdeck!) randomly shows you stiff you are not following.
How is the media covering this refusal to decide what is true and false? The usual. Calling it dangerous. Implying that because someone bought an $8 blue check that suddenly Musk is at fault for everything they say and that everyone will believe everything they say because they have a blue check. That the flaws in deciding what is true and false, resulting in the censoring of true things, only proves the need to devote more effort to deciding what is true and false so the right things can be censored.
The first candidate for proximate cause of the new protests seems to have been a fire in an apartment building, in which some combination of the fire department response and people’s ability to leave was believed to be impaired by the Covid restrictions. Which makes a lot of sense as a trigger.
There are of course those who are blaming the spam on Musk, simply because he fired so many of the people in charge of stopping spam, thus making him blameworthy for any failures whether or not they would have happened anyway. I have no idea if more manpower would be able to solve the problem here, or the degree to which the technical answer is ‘work smart’ versus ‘work hard’ versus ‘have enough manpower.’
Lockdown procedures seem to be more… negotiable? Spottily enforced? Than previously observed. This report is from a Beijing apartment complex that had a positive case and managed to stay unsealed for a while, and where it was not clear that official rules were being enforced at all. This seems like a pretty serious breakdown of the parts of their procedures that were doing the important work.
As another example, Christian Petersen-Clausen, one of my regular China sources, reports waking up to be told to report to outside quarantine for having been in a sports bar, despite already being in a locked down apartment building. After clarifying that it was not him, and playing up his medical conditions over several phone calls, he was able to bargain them into a four day lockdown at home for him and his wife. Sounds lovely. Due to his not having left his apartment due to the building having previously been closed, the rest of the building can for now remain open.
China has no good solutions. If they loosen up they face a wave of infections and deaths that could be a large black mark on the regime in place of the credit it continues to claim, and render the last few years in a very different light. If they tighten back up, they face escalating protests and economic damage. If they do neither, they likely get the infection wave anyway.
What about the infection situation? Are you winning, son?
Or here is a log scale.
Time range was chosen to reflect China’s previous peak in cases. This rise was faster and is already bigger than the previous one, yet the previous one suddenly reversed, so the game is not yet over.
Here is a reminder for scale, China is now reporting ~20% of the American case rate.
Also for scale, the biggest USA peak was at about 2,385, another 20x current rates and 6x the top of that graph. There are still a lot of potential doublings left to go if things are not contained.
The log scale graph to me tells the story of things slowly getting worse. There are continuous flare-ups being contained, but underneath it a clear trend line that has been there since early 2021. Either China will need to be even harsher and find a way to sustain that, or the situation will slip through its fingers sooner or later.
Later, however, could still be a year or two (or even three) away if Xi feels sufficiently trapped. While I would not be surprised if the dam broke here, I am guessing that scenario remains a modest underdog for now.
We identified positive likelihood ratios for 250 clinical findings and negative likelihood ratios for 241 findings. These findings were associated with 67 distinct diagnoses. For our analysis we considered a diagnostic test to be useful if it had at least a moderate effect on pre-test probabilities. Authors have often equated this with a positive LRs ≥5 or negative LRs ≤0.2. On average, individual clinical findings were not powerful diagnostic tests. The median LR+ was 3.4 and the median LR- was 0.59.
Positive findings were more informative than negative ones. 28% of positive findings had likelihood ratios greater than or equal to 5 while only 12% of negative findings had likelihood ratios less than or equal to 0.2. The mean positive and negative likelihood ratios were 6.4 and 0.57, respectively. Figure 1 shows the likelihood ratios of positive and negative findings on a logarithmic scale. Figure 2 shows the findings categorized by likelihood ratio: 1 to <2, ≥2 to <5, ≥5 to <10, and ≥10 for positive likelihood ratios and 1 to >0.5, ≤0.5 to >0.2, ≤0.2 to >0.1, and ≤0.1 for negative likelihood ratios. Considering the most impactful diagnostic findings, there were 31(12.4%) positive findings with likelihood ratios greater than or equal to 10 while only 13 (5.4%) negative findings had likelihood ratios less than or equal to 0.1. These most diagnostic findings are listed in Table 1.
Discussion
In this study we found that only 21% of findings were associated with likelihood ratios that cause at least moderate shifts in the probability of a diagnosis (≥5 or ≤0.2).
There is clearly a lot of room for improvement in selecting the most informative tests to run and information to gather.
According to this parents’ report, child medical care in Quebec seems to be in a very bad place. No painkillers or fever relief at the pharmacy so you have to improvise it from adult stuff, and more importantly if you actually need to see a doctor, good luck with that, at minimum you are looking at ~20+ hours in a room full of sick kids. Oh, and a quest is required to find antibiotics.
There at least used to be a potential equilibrium for some activities that equates to Bill Clinton’s old ‘safe, legal and rare.’ If you want X badly enough, you can get it, quietly. In exchange, we all agree to feel vaguely bad about the situation when it happens, to low-key cover up that it happens so often, and not to non-quietly advocate for not feeling vaguely bad about the general class of such situations.
In some cases this was very much not the right choice, in either direction. In others, it served us well.
I worry that we are losing this civilizational ability. That so long as a given X is legal, X is going to get hyped by those who commercially or socially benefit from doing so, no matter the costs to others. Thus we continue the ramp of only having things that are either compulsory or forbidden.
