Quebec COVID update: Back when I projected this wave would just never generate the deaths and hospitalizations the government worried about, I made 3 scenarios for ICU cases. One assumed Case-->ICU ratios returned to historic wave ratios. One assumed they flatlined. One fell.
For January 8, the latest data, my low forecast suggested there would be 208 ICU cases. My mid forecast suggested 333 ICU cases. And the high forecast, the one the government is basically treating as "what is going to happen," was 795.
In reality, on January 8, there were 257 ICU cases: between my low and mid scenarios. So it continues to be the case that Quebec's wave is generating far fewer ICU cases per official positive case than prior waves.
Nonetheless, since Quebec "officially" has around 260-270 ICU beds, we are basically "at capacity" now. My mid forecast has always suggested we would go "over" ICU capacity.

So will we have incredibly high death rates?

Probably not.
Here's the rate at which lagged ICU cases generate official COVID deaths. As you can see, the ICU-->Death rate HAS risen somewhat, perhaps pointing to some hospital overcrowding. But even at present "spike" levels, it's 1/2 the rate as last winter.
This suggests that, overall, either the typical ICU case has become less severe, or hospital resources have expanded, or treatments have improved.
I don't think hospital resources have grown. In fact stories about nurse shortages and medical personnel quitting suggest the opposite: hospital resources may have fallen.

That implies hospitalizations occurring at milder thresholds, or improved treatment.
I actually don't think the threshold for hospitalization is likely to have fallen. If anything, fear about hitting hospital capacity and nursing shortages may have led hospitals to have gotten stricter about how they allocate ICU resources.
This implies that the *likeliest* story is improved treatments.

Note that vaccination cannot have a role here. Were talking about *among hospitalized people*. If they're vaccinated, they've already had a severe breakthrough case. Vaccine benefits failed them evidently.
Nor is it likely that this is an "Omicron is milder" story since, again, we are talking about *among people in the ICU*. Omicron may be milder on average, but for people in the ICU, it's not mild! It's by definition severe!
So my suspicion here is just that the medical system has gotten better at treating COVID patients both by finding actual medical treatments and just finding better overall courses of care.
Which is to say, it's very likely that the ICU-->Death rate, even if it could briefly spike higher, will remain generally below the levels seen last winter.
So what does this mean?

Since the Cases-->ICU rate has fallen by about 85%, and the ICU-->Deaths rate has fallen by about 50%, the Cases-->Deaths rate has probably fallen by something like 90-95%.
And indeed, on various lags, it seems like the Deaths/Lagged Cases rate has fallen by something like 85-95% in this wave vs. last winter.
This means to have a comparable amount of concern as last winter, you'd need to see somewhere between 6 and 20 TIMES as many cases.

We are seeing about 5x as many.
Since the current lockdown in Quebec is about as intense and in some ways even more intense than last winter.... this means that even if you use the MOST PESSIMISTIC assumptions about the current wave, Quebec's reaction has been excessive.
We do not need to be shutting down all businesses, imposing curfews, forcing churches to close, etc. These policies are unnecessary and do not match the actual level of risk Quebec is facing.

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Lyman Stone 石來民 🦬🦬🦬

Lyman Stone 石來民 🦬🦬🦬 Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @lymanstoneky

11 Jan
Good question!

Mongolia's demography is absolutely fascinating and it's worth taking a closer look in regional comparison!
To start with, this is not my first foray into Mongolia commentary. It's an amazing country I've tweeted about a lot because I'm lowkey obsessed with it. It's basically the Kentucky of Asia.
You can read some of my takes on the plight of Mongols in China here as well. The cultural commonality and difference between ethnic-Mongols in China and Mongolians is a fascinating divide without clear parallel anywhere in the world!
Read 36 tweets
10 Jan
This study of long COVID in Norway has a nice design: PCR tests, long follow-up, large sample, and a plausible mechanism for identifying cases: a huge pre-existing panel survey using frequent mobile-device questionnaires. medrxiv.org/content/10.110…
My only real critique is that while they do have about 80,000 people in the sample, it's Norway, so they only managed to find about 800 COVID cases. lol. problems of low attack rate!
But, among those 800, they found that no single symptom of "long COVID" occurred among more than 22% of infected people. The most common symptoms among COVID+ people were loss of taste/smell and fatigue. Loss of taste was NOT correlated with other cognitive symptoms.
Read 15 tweets
10 Jan
Statistics Canada is doing a pilot study of a mobile-device-based wellbeing-survey and they're recruiting respondents through the Very Scientific method of.... emailing everyone who's ever emailed their technical support desk.
Extremely strong selection there for "disgruntled data analysts" so that's gonna be a WILD pilot study.

I have of course signed up and downloaded the app and am PSYCHED to be a respondent in another survey.
I'm a longitudinal panelist in two surveys, I've gotten ACS AND CPS (!!), and I'm a standing panelist in multiple "big survey company" databases, and because I very reliably take the surveys, my personal traits are wildly overrepresented in studies of Americans.
Read 7 tweets
10 Jan
Luckily, this can be simply "solved" from a statistical approach by just deleting the option "Christian" from the survey.
When you have an option that is confusing respondents, there's a solution! Delete it! Force them to choose between the empirically meaningful options.
Also, the actual best way to survey religion is to ask people, "Think about the place you most often worship. What is its name?" and then to do a follow up, "Do you happen to know what religion this place of worship is associated with?" then give religion options.
Read 10 tweets
10 Jan
I like David French but this is clearly wrong.
There has been progress on racial justice, yes: but most of the progress occurred while religiosity was still very high and fairly stable. As religiosity has declined, progress on racial justice has clearly faltered, and we're now in a place were it's not even clear...
What anybody means by a phrase like "progress on racial justice." Without a shared pre-political moral landscape, statements like that are meaningless, and so unsurprisingly political discourse around them becomes irresolvable.
Read 21 tweets
6 Jan
What is the best empirical evidence that therapy actually improves mental health?

I am struggling to find anything credible.
This meta-analysis of 147 studies seems to suggest p-hacking is very common, publication bias is huge, and even with that the typical effect of therapy vs. care-as-usual is clinically insignificant. pubmed.ncbi.nlm.nih.gov/21770842/
This more recent one specifically on CBT for adult depression suggests that CBT has been wildly overrated by creative research practices. journals.sagepub.com/doi/abs/10.117…
Read 32 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(