Finally found it!

pure.ed.ac.uk/ws/portalfiles…

Top line results:

* 35% as severe as Delta (which was in turn ~200x that of Wuhan-Hu-1) - in contrast with 95% as severe in the Imperial College report.

* 10x greater risk of reinfection than with Delta

* Boosters give 57%...
... protection against symptomatic infection.

The nuance:

* Age *is* controlled for. This is important, as 49% of Omicron cases were age 20-39.

* Vaccination status is controlled for, but past infection is not.

* Small sample size (15 hospitalized patients)
* They *do* appear to control for time bias with a Cox proportional hazards regression model.

* "Only individuals reporting symptoms at the time of test were included in this study" - Not sure why they're doing this, sounds like a good way to introduce bias.
* The comorbidities controls look decent, and they do further break them down by age as well, although only into two very broad groups (16-49 and 50+), so they sort of Simpsons Paradox it. But the main control for the risk associated with age itself is based on a spline (good).
* Sensitivity analyses: if the study is limited to only people with at least 7 days followup since diagnosis, the results stay about the same (33% the risk of Delta). If limited to 20-59yos, the risk increases significantly to 44%, but with overlapping confidence intervals.
* Limitations include using S-gene negative tests as a proxy for Omicron and S-gene positive for Delta. Not a big limitation, though. They also can't verify that all hospitalizations post-infection were due to COVID, but account for it by proxy.
* Another limitation is that their model assumes the same delay between infection and hospitalization. If this changes, then it alters the associated risk of Omicron fairly significantly.

Overall:

It looks like a good study, dealing with the challenge of limited data...
... over very limited timeframes. I'd say it even looks better controlled than the Imperial College study (which was in turn better controlled than the Danish report), though they're both important datapoints. If this 1/3rd risk holds then Omicron is about 70% the risk of...
the original Wuhan-Hu-1 strain, before controlling for vaccination and past infection status.

The main risk with Omicron, as explicitly noted in the document, is the high case growth rate and high percentage of the population that is vulnerable, ...
... leading to the risk of swamping the medical system even if the strain is less severe than Delta.

Not discussed or investigated: Delta is still circulating widely, and it's not clear what protection Omicron infection gives against Delta (given that the inverse is "little")

• • •

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

Keep Current with Nafnlaus

Nafnlaus 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 @enn_nafnlaus

1 Dec
More detail on the preliminary Omicron data reported by Israel the other day:

jpost.com/health-and-wel…

Beyond the mentioned:
* +30% transmission
* boosted=90% protection v. infection (was 95%)

It's:
* boosted=93% protection vs. severe (same)
* 2,4x more severe strain than Delta
The 2,4x more severe ("those not inoculated have a 2.4 times greater chance of developing serious symptoms") runs contrary to the narrative being spun in some African nations that it's a "mild strain" - statements often made right before insisting on lifting the travel bans.
Statements that - I should add - were just repeated by an anonymous WHO official, along with claims that vaccine efficacy isn't reduced, and also demanding the lifting of travel bans:

news.sky.com/story/covid-19…

This is all so preliminary, though, so again, massive grain of salt.
Read 6 tweets
30 Nov
A reminder when viewing exponential growth charts: a wave with faster case growth, plotted against waves with slower growth, will *inherently* seem to take longer for hospitalizations, ICU admissions and deaths to "catch up". Simulated scenario with varying doubling times below.
In the above graphs, the slower waves were seeded with more initial cases, so that the faster growing waves past the slower ones on day 27. But hospitalizations don't pass until day 34, ICU admissions until day 38, and deaths on Day 41.
This is an inherent result in the delay between cases and more severe outcomes. If case growth reaches a given height corresponding to an earlier wave, but at a faster growth rate, it hasn't had as much time for cases to become more severe - thus they plot lower vs. cases.
Read 5 tweets
28 Nov
Since comments were shut down on this thread... a reply thread.

1) The existence of a finite number of things that professionals have been wrong about in their field does not in any way imply the likelihood that a random person is likely to be more correct than professionals.
E.g., if climatologists get some specific detail of climatology wrong in regards to *rapidly evolving news*, that doesn't mean that one should listen to John Doe over the IPCC.

2) When one posits "theoretical harms" of COVID restrictions up against the actual measured harms...
.... the real, demonstrable harms win. I used excess death figures to avoid the "died with COVID vs. died of COVID" red herring. And to be clear: yes, these excess deaths are mainly due to respiratory disease, & to a lesser extent cardiac - the way COVID kills people.
Read 18 tweets
26 Nov
To anyone repeating the "B.1.1.529 is due to a lack of vaccine sharing" line: South Africa is not vaccine limited, and hasn't been for months. They've been vaccinating children 12+ for a month. Anyone there can get vaccinated just by showing up. The problem is hesitancy.
A friend of mine from SA messaged me the other night as I was going to sleep, to complain about the mess over there. He has multiple comorbidities, but almost nobody around him is vaccinated. Lots of cases of companies even banning vaccinated people because of "spike shedding".
His company is making everyone (incl. software developers) work at the office. Not for work-related reasons - so they can do extracurricular "bonding activities". A coworker has an adopted child with HIV (immunocompromised & vulnerable) who sought an exemption. Exemption denied.
Read 6 tweets
14 Nov
I was today years old when I learned that all six of the programmers behind ENIAC, the first digital computer, were women (whose work went almost entirely unrecognized until the 1980s).

en.wikipedia.org/wiki/ENIAC

Their first program was nuclear simulations for a hydrogen bomb.
The six - Jean Jennings, Marlyn Wescoff, Ruth Lichterman, Betty Snyder, Frances Bilas, and Kay McNulty - were "human computers" who got their positions due to wartime labour shortages. They kept their jobs after the war because their experience was too difficult to replace.
As generative programming languages did not yet exist, "programming" ENIAC meant switches and rewiring cables, which required a full understanding of its blueprints. Photos of them next to ENIAC referred to them as "refrigerator ladies" & they had to serve as hostesses for guests
Read 14 tweets
8 Nov
@spmanipulator @andjrison @GidMK Colour vision is weird, and a lot more complicated than one might at first think.

For a greenhouse project, we had the idea of using red-blue grow LEDs (mostly red, since it's super-efficient) for primary grow lighting but mixing in various small amounts of (inefficient...
@spmanipulator @andjrison @GidMK yellow) HPS with the idea of shifting the appearance of the light over the course of the day for our guests, from reddish sunrises to yellow dawn to white daylight. You know, we'd have red, yellow, and blue, and just blend through them in the right ratios - easy, right? Except...
@spmanipulator @andjrison @GidMK it totally didn't work in testing. If there was almost any blue at all (we couldn't fully shut off the channel), it looked purple/pink, unless the (inefficient yellow) HPS was cranked *way* up, wherein you'd get yellow light. And only with *tons* of blue and little of the...
Read 11 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!

:(