jsnell
1930
It could be below zero if vaccinated people e.g. behaved differently (this is not a double blind trial, after all). In particular, people with less time since their 2nd shot are probably younger. But also the error bars are wide and intersect zero.
The X axis is a bit confusing. It starts off counting weeks from 2nd dose, then switches to counting time from booster once that is given.
In the fine document:
I’m pretty strongly convinced by the UK data that we can rule out founder effects - plenty of replication data on Omicron out-competing Delta. Time to get your booster and duck and cover (though the results from Sette Lab and work by Redd et al are good news.)
abrandt
1934
Booster gotten, hopefully not ducking and covering until after Christmas doesn’t bite me.
Also anecdotally, my Moderna booster hit me way more than my two doses of Pfizer did. Worst headache in recent memory(this from a migraine sufferer), much longer lived arm pain, and now a couple days later I’ve had a lymph node swell up in the arm pit of the arm I got my shot in.
From the fine nature article below:
If the virus evolved in this way, it might become less severe, but that outcome is far from certain. “There’s this assumption that something more transmissible becomes less virulent. I don’t think that’s the position we should take,” says Balloux. Variants including Alpha, Beta and Delta have been linked to heightened rates of hospitalization and death — potentially because they grow to such high levels in people’s airways. The assertion that viruses evolve to become milder “is a bit of a myth”, says Rambaut. “The reality is far more complex.”
https://www.nature.com/articles/d41586-021-03619-8
[It’s a really good article!]
Also, from the UK data…I’m gonna wait on someone more qualified than I to comment, but this feels weird to be buried at the back. Granted, it’s only 8 sequences so maybe it’s not something that will rise in frequency and be important.
Omicron VOC-21NOV-01 (B.1.1.529) has been separated into 2 clades: BA.1 and BA.2. These
clades share the majority of the mutations initially identified in B.1.1.529. Figure 9 shows the
mutations across spike present in the 2 lineages (BA.1 = 793 sequences, BA.2 = 8 sequences).
Sequence data that showed signs of contamination or were low quality were removed from this
analysis (n=13). Figure 10 shows a maximum likelihood phylogeny for the 2 clades and
highlights the distance between the 2 groups, despite the number of shared mutations.
jsnell
1937
I’m obviously not more qualified, but will comment anyway :-P BA.2 doesn’t show up with the SGTF signal. So it was in the news a bunch a few days ago, with various “stealth version of Omicron” clickbait headlines. That’s probably why it’s being addressed in the report, I haven’t seen any suggestion that it’s otherwise interesting.
jsnell
1938
The data from Denmark is being released daily, and now has useful comparisons to data on non-Omicron variants in the last 3 weeks: https://files.ssi.dk/covid19/omikron/statusrapport/rapport-omikronvarianten-13122021-i30w
In particular, for hospitalizations Omicron is actually more likely to lead to (non-incidental) hospitalizations over that time period:
This is kind of shocking, because my intuition would have been that the Omicron number should absolutely be lower than the Delta one. The Omicron cases are far younger, more likely to be vaccinated, and skew more recent and should thus have not yet had time to be hospitalized. No reason to panic since the numbers are small, but this at least should reset the expectations to “we don’t know at all”.
Well, what if cases are less severe for Omicron? You might expect that due to the fact that testing isn’t random that you’re missing cases for people with mild/no symptoms for any variant. It could be that there are many more cases of Omicron than you’re ascertaining, and thus your denominator (not been hospitalized) is under represented.
Also, from the look of it the Omicron wave has already passed through Denmark, so no biggie. [I’m not being serious.]
(So indeed, even if – and the jury remains out on this one – only a low percentage of those with some existing immunity from previous infection or vaccination experience severe covid, we’re heading into the territory of “Even small percentages of very large numbers are still large numbers” territory.)
I’m sure there’s actual science and modeling behind it, but without context, that estimated growth graph is pretty funny, in an ??? → profit kind of way.
Yeah I have some feelings about the way the authors drew that line.
My approximation:
jsnell
1943
That’s an interesting idea. I’m a bit resistant to it just since we have the evidence from the Norwegian super-spreader event of basically 100% of infections in young vaccinated people being symptomatic. That makes the option with no symptoms seems implausible. Cloud the symptoms being different (e.g. no loss of sense of smell) cause a sufficient sampling bias? I guess it’s possible.
Have you considered Tweeting this? You could have 100k followers in no time!
Yes, the line was indeed too conservative. There’s three more days of data past the 6th already, and the growth since then was faster than the dark red line. And the prediction for the 13th seems to be about 5k cases, the reality is 7.8k. They’re only barely withing the 95% confidence interval.
Anyway, what’s really killing me is having the x-axis checkmarks at 3.5 day intervals…
Just wait until that line completes the curve and goes back the other way.
I get what you’re saying, but it feels like the sequence of events and timing for how the entire testing journey or whatever would be the precise term for that Norwegian event is very much anomalous to what we’re already seeing as far as identifying and confirming Omicron cases at large, now. I’m sure there’s still some of the very focused contact tracing going on, but what I keep seeing is labs that are flagging SGTF results in “regular” testing (which to me implies that there may be some selection for symptomatic cases) for potential sequencing to confirm Omicron.
Does someone have that same plot with the Y axis logged? I would assume this is a curve fitting exercise where the growth is exponential, and the rate of the exponential increase is fitted from the first set of data points, and then projected. It doesn’t look like there’s a big increase because in absolute numbers as even an exponential increase in a small number is still a pretty small number for a few doubling cycles.
A different Y axis for this plot would really help with this, though I bet it would make the delta rate of increase look pretty funky (are they using an exponential there? I think they are using a linear fit. That’s an interesting choice.) Also, it is reasonable to assume that the infection rates for delta should decrease as it becomes less prevalent in the population, so seeing it continue to project up means that these are two naive extrapolations of the data.
[edit after searching down the twitter thread.]
Oh look, someone else posted the log Y axis. I don’t even know why you’d present the non-logged Y.
jsnell
1947
Coming back to this, now that an expert had said something:
So BA.2 indeed seems unlikely to be interesting, but there’s another variant that’s already 10% of Omicron globally that has a mutation that’s likely to lead to increased immune evasion. I had seen the latter mentioned, but had no idea it was such a large proportion. One thing worth noting is that on the nextstrain map the R346K variant seems substantially more common outside of ZA.
jsnell
1948
Analysis from South Africa’s largest health insurer. The whole thread is worth reading, but this is the important Tweet:
Vaccine efficacy against hospitalization down from 93% to 70%. That’s much better than I was expecting, especially since the it is also dragged down by immunity waning over time rather than just Omicron. I’d count this as good news. The efficacy against infection is in line with the UK numbers, which gives this some additional credence.
(Though the way the maths works out, a 93% → 70% reduction in vaccine efficacy still means 4x as many vaccinated people being hospitalized as with Delta, and over a shorter period of time. So it feels like the concerns around health care systems being overloaded are not off the table even with those numbers.)
A new study worth checking out once the pre-print hits. As suspected, that third dose of vaccine providing a stronger NAbs response is pretty important versus Omicron:
And then as this thread notes, this isn’t the whole picture – just antibody neutralization. Not included here is the cellular immunity response.
