All my various postings tracking the COVID-19 pandemic through the end of 2022.
The recent data from South Africa have highlighted an interesting problem with COVID-19 hospitalization data. Omicron in South Africa generated a strikingly large number of hospital admissions with COVID, but not for COVID. That is, patients who were admitted for some other treatment, but were found to have COVID-19 upon testing. Continue reading Post #1349: Hospitalizations for COVID versus hospitalizations with COVID
I apologize, but I have nothing to scare you with today.
The U.S. winter wave has stalled out pretty much everywhere but the Northeast. In particular, the regions that led last year’s wave (Mountain and Midwest) now look like they’ve peaked.
And we now have a South African academic doing the math showing that Omicron cases are vastly less severe than Delta cases, on average.
I’m sure you’re going to see a lot of reporting to the contrary. I address one piece of that below. But as far as I can tell, all the information now points to Omicron having vastly lower average severity per case.
Briefly, those two observations are:
1: Extremely rapid growth of Omicron cases is occurring, but this is not translating into extremely rapid growth of total cases. As I explain in excruciating detail, I think the reason boils down to this:
Net new case growth rate = Omicron growth rate – Delta disappearance rate.
2: The case hospitalization rate for Omicron, so far, is about one-fifth the rate for Delta.
Details follow. Continue reading Post #1347: Omicron, two observations on the U.K. experience so far.
This post is a short summary of where I think things stand as of 12/12/2021.
1: Omicron appears to be vastly less deadly than Delta. The first (and maybe) last graph you need to focus on is this one, showing new cases (top) and new deaths (bottom) from Omicron. For South Africa. For the entire pandemic.
The key point is that, based on South African administrative data tracking the pandemic, Omicron doesn’t kill people at anything like the rate that prior strains did. South Africa is now three weeks into their Omicron wave, and the deaths just aren’t showing up. That’s the empty circle on the lower right.
Source: WHO. Items in red are mine.
This does not appear to be some systematic issue with South African death data reporting. In prior waves, deaths rose almost contemporaneously with cases. That’s shown below, where I used the Wayback Machine to grab a snapshot of deaths in the middle of the prior South African wave.
Source: Wayback Machine. Red notes are mine.
Finally, this is completely consistent with my “two weeks’ lag” analysis from Post #1343. If I take deaths as a fraction of cases from two weeks back (to account for the time it takes to die and have that death recorded), and factor in the fraction of new cases that are Omicron, that analysis yielded an Omicron death rate that was, for all intents and purposes, zero.
The upshot is that whether you take three weeks of data and say “we should be seeing those deaths rolling in”, or calculate deaths as a fraction of new cases from two weeks ago, you come up with the same answer: The mortality rate from Omicron cases is a tiny fraction of the mortality rate from Delta.
The reason this isn’t making the news is that everybody knows that deaths lag cases. Nobody wants to be premature in making this judgment.
But they don’t lag by an infinite amount. At some point, the fact that almost no deaths are appearing means that, in reality, almost no deaths occurred.
I’ve even seen some statistically naive reporting, in that the number of “excess deaths” in South Africa doubled between October and November. That was a headline in yesterday’s New York Times. The only problems with that are a) November mostly predates the current wave and b) excess deaths during quiet periods is the small difference between very large numbers (total deaths and total predicted deaths). It’s a noisy figure at best.
I’m not going to bother with a full debunking of that because one week from now it will be moot. Give it another week, and everybody will have figured out that the deaths circled above are missing.
2: Omicron appears to generate far fewer hospitalizations. This is another situation where you see a lot of naive reporting of facts, without a lot of thought behind the reporting.
I went over the key facts in Post #1343. The key facts are that a) there are relatively few hospital admissions with Omicron, relative to what you would expect based on the U.S. hospitalization rate for Delta, and b) almost all the “Omicron” admissions in South African hospitals are not sick with Omicron. Per statistics I cited in an earlier post, in one large hospital system, 76% were admitted for something else. They only discovered the Omicron infections because they test everyone who is admitted.
This goes hand-in-hand with my third point below, the very high fraction of cases that appear to be asymptomatic. People who know they have an active COVID infection don’t go to the hospital for elective care. But they are showing up at South African hospitals now, in large numbers, precisely because they don’t know they are infected.
And so, you see a lot of breathless reporting about the huge increase in “Omicron hospitalizations”. But that is by-and-large a consequence of the fact that there are now a lot of asymptomatic Omicron infections in the South African population. It does not seem to be a consequence of a lot of people being admitted for treatment of COVID-19.
But this also means that you must fully ignore all the discussion by South African physicians regarding the lower apparent severity of hospitalized Omicron cases now. That casual impression of lower severity of the hospitalized is part-and-parcel of this same phenomenon.
So, for example, you will see headlines about the low fraction of Omicron cases that are in the ICU, the almost negligible fraction on ventilator, or the very low fraction of Omicron cases that need oxygen.
Ask yourself this: If they don’t even need oxygen, why are they in the hospital? Answer: Because they’re not in the hospital to be treated for Omicron. They’re in the hospital for something else, and the hospital picked up the asymptomatic Omicron infection.
Based on the reporting of this, I believe I understand why physicians believe that the average hospitalized COVID case is less severe under Omicron. Persons with an active Omicron infection have to be separated from the general hospital population. Best guess, all of those persons get housed together in whatever infectious disease ward the hospital has set up for its Omicron patients. (And this is also why South African hospitals are reporting counts of all persons with Omicron, rather than persons admitted for treatment of Omicron. As far as they are concerned, both populations have to be housed in the same infectious disease ward.)
Doctor now step into the infectious disease ward and notice that it’s not total bedlam, the way it was in the past. And so they say, the average person in the Omicron ward isn’t as sick as they were in the past. And they are right. But that’s because in past waves, that ward would have been filled with people being treated for Omicron. Now, it’s mostly filled with people being treated for something else, who happen to have an asymptomatic Omicron infection.
So all that eyewitness reporting from physicians in these hospitals is worthless. They are being confused by the fact that with Omicron, the typical person in the contagious disease ward is no longer there to be treated for COVID-19. And, sure enough, few of them are being intensively treated for COVID-19.
3: All the reporting says Omicron generates mild and asymptomatic cases. Nobody disagrees. That suggests that it’s true. “It’s true” is a far simpler explanation than suggesting that everybody who had had first-hand experience with this is deluded. And, best I can tell, the large majority of cases are, in fact, so mild as to generate no symptoms.
In a prior post, I had a quote from a South African reporter saying that all hospitals were reporting the same thing. Nobody was reporting a deluge of severe cases. But you can take that further, by eye, and highlight that everywhere Omicron has been found, the local authorities have said that cases are mild, so far.
In particular, keep you eye out for reports of a high fraction of cases being asymptomatic. Here’s why. In general, severity runs in a spectrum, from least to greatest symptoms. From asymptomatic cases to death. If deaths are missing and hospitalizations are low — so there are few cases at the severe end of the distribution — then it’s almost a given that the share of cases that are asymptomatic will be high.
And so, I can find headlines saying that:
India, 54 out of 55 cases so far have been asymptomatic. (December 9, 2012: Out of 55 cases, one had mild symptoms, the rest were asymptomatic).
Botswana, 16 out of 19 Omicron cases were asymptomatic (December 2, quoting the health minister for that country).
Europe, 212 cases, all mild or asymptomatic. (December 6, 2021).
USA, 42 cases, one two-day hospitalization, the rest mild or asymptomatic. (December 10, 2021).
4: Further evidence that vaccination doesn’t stop Omicron.
This isn’t really new information. It’s just an extrapolation from existing work showing low antibody response to Omicron in a test-tube.
I went through the basic (test-tube) evidence that the standard two-dose vaccine regiment does not stop Omicron. There’s now a quantitative estimate from that same group that says the two-shot vaccine should be only 23% effective against symptomatic illness from Omicron.
That compares to around 85% effectiveness against Delta.
That’s not an actual measurement from the population. That’s some sort of mathematical model, based on the observed antibody responses in vitro.
There’s still an almost-universal assumption that the two-shot vaccine regiment will largely prevent severe disease (e.g., hospitalization or death).
I was initially skeptical of the argument that Omicron results in milder cases. Even if so, I would not have guessed that they were milder enough, on average, to make any material difference.
And, early on, the hard evidence for that was lacking. But the evidence is now beginning to accumulate. And it surely looks like the average Omicron case is vastly less severe than the average Delta case.
I think this needs to be said plainly: Just because all the nuts immediately glommed onto this does not mean that it’s wrong. People who leap to a conclusion without firm evidence, and on the basis of their prejudices, may sometimes be right.
(Think of it this way: They have to be right sometimes. If they were always 100% wrong, then their opinions would convey useful information. You’d have a perfect (and therefore) useful “negative indicator”. But because these are opinions formed in ignorance, that can’t be true. Ignorant opinions cannot convey useful information. Therefore, every once in a while, purely by chance, they’ll be right.)
I’ve heard the “law of averages” expressed as “anything will happen, that can”. I think that maybe this is one of those cases.
To me, all the evidence suggests that the spectrum of illness severity of Omicron cases has shifted radically toward lower severity, compared to Delta. Near as I can tell, on a per-case basis, it’s resulting in:
If we could choose, we’d swap our current Delta caseload for the equivalent Omicron caseload in a heartbeat.
And, if I’m right about hospitalization — ten-fold fewer — even if Omicron results in somewhat more cases, we’d still be willing to swap.
My bottom line is that things are going to get better here, in the U.S., as Omicron takes over from Delta. Somewhat more cases, but vastly lower severity per case. That’s a good trade.
I think it is — purely by chance — exactly as described in my little bit of science fiction (Post #1338: Embrace Omicron). I wrote that tongue-in-cheek. But now I think it’s more-or-less correct. For us — where we already have a pretty good winter wave in progress — I believe Omicron is going to reduce morbidity and mortality and hasten the end of the pandemic.
Do I dare to use the F-word here? Why not. Flu. Compared to prior strains, I think this is going to be a lot more flu-like in its mortality and hospitalization case rates.
If so, you can be sure that all the right-wingers in the nut-o-verse will be crowing that they knew this all along. But they didn’t. It’s not that they were right. They were dead wrong, and we have the sharply increased 2021 U.S. mortality rate to show it. It’s that they are stopped clocks, the nature of the virus has changed, and even a stopped clock is right sometimes.
Call me Pollyanna. Or Pangloss. I don’t care. That’s the way the data look to me.
Embrace Omicron. It’s not as if you have any choice in the matter anyway.
Source: Calculated from William and Mary COVID-19 dashboard.
Not sure quite what to make of that.
New case rates have ticked up in Virginia as a whole, consistent with a “winter wave” of COVID (Dark line on the graph below).
Source: Calculated from NY Times Github COVID-19 data repository.
I guess that’s not a particularly worrisome level of new cases. And there’s not much that you can do anyway, as the semester is nearly over.
With this endless pandemic, I find it helpful to look back and see how things have changed for the better. In this case, it’s interesting to compare the end of the semester this year to last year.
What I find most striking is that this year, nobody is making noises about the need to quarantine returning college students. Last year, that was very much in the news (Post #988, November 2 2020.) Nor am I seeing recommendations for COVID-19 testing prior to return to home, which I believe was more-or-less the norm last year.
This year, with a much more muted winter wave of COVID, and vaccination for more-or-less everyone who wants it, I guess the risks are just that much lower. Or we’re that much more tired of dealing with it. Or some combination of the two.
That said, end-of-semester is a good time to make some ballpark estimates of risk of spreading disease. Not to be morbid, but to get some sort of a handle on the odds. Last year, everybody was offering advice about quarantining your college student. This near, nobody’s saying anything. Either way, you might want to step back and ask the simple question: Given the rate of new cases on campus, what are the odds that my kid is going to come home asymptomatic but infectious with COVID-19?
The above is an abbreviated version of the calculation given in Post #988 cited above. It isn’t rocket science. You start with new cases showing up on the William and Mary campus at a rate of 19 new cases per 6600 on-campus students for the last two weeks. From that, you take a guess as to how many students will be leaving the campus in an infectious-but-asymptomatic state. That guess will depend critically on how many days you think the average infected student remains infectious-but-asymptomatic. I have assumed a generous 8.6 day average period (calculation shown in Post #988).
Anyway, my best guess is somewhere around 1 in 600. Or so. FWIW. One in 600 will return home in an asymptomatic but infectious state. Again, not rocket science. That’s the same as saying that a bit over a week’s worth of new cases will walk off the campus undiscovered.
There’s one big problem with applying last year’s calculation to this year’s data. Last year, there was campus-wide testing, and it presumably identified asymptomatic cases. This year, there was no systematic effort to test and find asymptomatic cases. This year, we ought to add in a fudge factor for those undiagnosed, asymptomatic cases. Best guess, historically, 40 percent of COVID-19 cases have been asymptomatic. So you’d not-quite-double the estimate, to account for the missing asymptomatic cases.
On the other hand, an infected student doesn’t necessarily translate to spread of disease. Even in the pre-vaccine era, there was no guarantee that an infected household member would spread it with in the household. That “secondary attack” only occurred in about 15 percent of cases. Now, with vaccines, I’d have to guess that only a tiny fraction of those students will go on to spread COVID at home.
The upshot of that last bit is that the likelihood of my kid coming home and spreading COVID-19 in the household is much less than the one-in-600 cited above. One in several thousand, at a guess.
Even as vague as all that is, I find it helpful to ballpark the risk. Best guess, there’s something like a one-in-several-hundred chance that your kid is going to come home with COVID-19, and be infectious, and not know it. And the odds of that actually leading to an infection within your household are one in several thousand, best guess.
It’s not one-in-a-million, but it’s not one-in-a-hundred either. Scale your response (if any) accordingly. For our part — vaccinated and boostered — we’re just not even going to give it a thought. YMMV.
Students return to campus on 1/26/2022, or seven weeks from now. Based on the South African experience, and what is now predicted for Great Britain, Omicron will be the dominant strain in the U.S. long before then. It should take a month, at the outside, to go from first discovered cases to dominant strain.
My point is, this year, students are leaving the campus under Delta. They’ll be returning to the campus under Omicron.
Oddly enough, for a guy who’s clearly been worrying about this a lot, I don’t think Omicron is going to be a problem. I think it’s going to be an asset. My prediction is that it’s going to spread like crazy, but very few individuals are going to get very sick from it.
(This, despite the fact that two vaccine doses alone does not protect you from an Omicron infection. Two doses might give you some protection against a serious case (e.g., one requiring hospitalization). But with two vaccine doses only, people seem to mount essentially no antibody response to Omicron. You need three doses (two plus booster), or two doses plus prior infection, to generate significant antibody response to Omicron.)
In any case, my projection of widespread-but-mostly-harmless Omicron is based on my reading of the data out of South Africa, which you can see in my just-prior posts. The widespread part is pretty much a given. The “mostly harmless” is my reading of some tenuous data. I should be on firmer ground there in a week, so when I update these numbers a week from now, I’ll provide my best guess on how Omicron is going to play out in the U.S.
However it plays out, that’s what we’ll be facing when students return to campus next year.
The winter wave flattened out this past week. The Northeast continued to see growth in new cases, but not the Midwest.
The title of the article pretty much gives you the entire content.
First, I took data on South African COVID-19 cases and deaths from Johns Hopkins, and calculated a “two week lag” mortality rate. That is, today’s deaths, divided by new cases from two weeks ago. That’s my best approximation to the true mortality rate. Near as I can tell, the case mortality rate for Omicron in South Africa is very small. So small that I can’t distinguish it from zero given the current state of the data.
In addition, I took what could find regarding hospitalizations for Omicron in South Africa. Most of what you read about that is highly misleading, because something like three-quarters of Omicron hospitalizations in South Africa are people who were actually being treated for something else. They just happened to uncover an asymptomatic case of Omicron because everybody admitted to the hospital gets tested. Adjusting for that, and taking a couple of day’s worth of reported data, best guess, the case hospitalization rate for Omicron, in South Africa, is maybe one-tenth of the rate in the U.S., for Delta. That is once again my calculation, based on what data I could scrounge up.
And so, a bit of good news for a change. Best guess, based on the data I can get my hands on, the average Omicron case really is a lot less severe than the average Delta case.
Obviously, I’ll revise these estimates as more information becomes available.
Continue reading Post #1343: Two calculations showing much lower illness severity for Omicron
Just marking time. Trend remains at 13% per week increase in cases. Continue reading Post #1342: COVID-19 trend to 12/8/2021
Trend? ‘Bout the same as yesterday. The trend numbers I’m sketching in are “connect-the-dots” estimates: One end is 11/24/2021, and the other end is the most recent data point. Based on those two points, new cases are now rising at 13% per week. Continue reading Post #1341: COVID-19 trend and the most under-reported bit of yesterday’s news.
At this point, the data reporting issues from Thanksgiving should be in the past.
Every region of the country now shows an upturn in cases. Best guess — via connect-the-dots, below — is that new COVID-19 cases are now rising about 15 percent per week, on average, in the U.S.
Continue reading Post #1340: COVID-19 trend to 12/6/2021, 15% per week
