The genesis of this is a simple question: At what point in this pandemic does COVID pose no more risk than flu does, at the peak of a typical U.S. flu season?
This turned out to be yet another of my TLDR posts, so let me just give the spoiler right here: At 30 new Omicron cases / 100K / day. And at 16 per day, it poses no more risk than the risk that flu poses for the average of the entire flu season (instead of the peak week). Those are the rates at which, by my calculation, the average American faces no more risk of hospitalization or death from COVID-19 than from normal seasonal flu.
To be clear, my definition of “risk” involves both the likelihood of catching the disease, and then the typical severity of illness once you’ve caught it. It’s the risk of just walking about minding your own business one day, and ending up in the hospital a week later with COVID or with the flu.
When you think of it that way, there’s always some point — once daily new COVID-19 infections have dropped to a sufficiently low rate — at which the risk of serious harm from COVID is less than the risk of serious harm from flu at peak flu season. The point of this is to calculate that rate, to serve as one benchmark for a return to normalcy. Once new cases get to that level, rationally, if you are a risk-neutral individual, you should only take about as much precaution against COVID as you do against flu.
This is, in a nutshell, how we’re going to get by in the shadow of “endemic COVID”. We hope. The disease itself isn’t going to become harmless. It’s just going to become rare enough to pose little day-to-day threat. We hope.
Think of Ebola, brain-eating amoebas, the plague, and so on. Those are dread diseases, for sure. But no rational resident of U.S. spends a lot of time worrying about them, or taking precautions against them, because the odds of contracting them on any given day, in the U.S., are more or less nil. They are serious diseases, but they are not a significant threat in the U.S.
That’s where we now hope Omicron is headed. And this posts answers the question “how will we know when we’ve gotten there?”.
This analysis is a rough cut, for sure. I’d like to do one more iteration, because the only people I care about, in this analysis, are those who are fully vaccinated and boostered.The people who won’t bother to do that pretty clearly aren’t worried about their risks. So why should I be?
That vaccinated-only comparison turns out to be difficult to do, for a variety of reasons. I’ve been trying, without success, to write this up that full calculation, for about the past week. I may never have the data to do that well. So let me just grind out the calculation for the average of all persons — vaccinated and not — and be done with it. I’m quite confident that the results I show here are conservative, in the sense that they overstate the COVID-versus-flu risk for fully-vaccinated individuals. Think of these results as a lower bound on the levels of COVID that should cause only minimal concern to a fully-vaccinated individual.
Background and recap
In my first post in this series, I made the argument that we lucked out with Omicron. Evolutionary pressures move successive variants of COVID-19 toward greater infectiousness and toward “immune escape”. But the virulence of each new dominant variant is really a matter of chance. Delta was much more virulent, Omicron is much less virulent, compared to the original (Wuhan) strain. And, just by chance, the variant that has come out on top (for the time being) generates a lower average severity of illness compared to prior variants.
In my second post in this series, I tried and failed to get my mind around what “endemic Omicron” will actually mean. We’ve got another month or two of very high new case rates, in the current Omicron wave. And after that, the consensus of scientific opinion seems to be that we’ll probably have COVID-19 circulating in the population, at some low level, permanently.
But Omicron isn’t very much like any of our other endemic infectious diseases. In particular, that combination of being extremely infectious and airborne, and yet lacking a permanent vaccine or permanent immunity of any sort, just seems to put this in a different league, to me. I can’t help but think that serious outbreaks will be part of the landscape, to a far greater extent than we see with (e.g.) seasonal flu.
That said, from what I read, the model for “endemic Omicron” that most people seem to be settling on is flu. In the long run, so the story goes, COVID-19 will be just another disease like influenza. If you’re worried about it, get vaccinated, stay out of crowds when there’s an outbreak going on, maybe mask up if you are vulnerable. Wash your hands. But otherwise, just get on with your life.
In this post, I want to do as thorough a job as I can, to compare the risk of harm from Omicron and typical flu at the peak of flu season. My focus is in estimating the risk of serious illness for someone who is fully vaccinated and boostered against COVID-19.
It’s not as easy to do this as you might think, because where data are available, they will refer to averages for the entire population. I’m going to have to “back solve” to break those down for the vaccinated and unvaccinated populations.
But the basic point is simple: If we can get to the point where your risk of illness from Omicron is essentially the same as your risk from flu, then … if you don’t take extraordinary steps during flu season, you really shouldn’t feel compelled to take extraordinary steps against Omicron season.
Outline of the analysis.
So here’s the exact question. Let’s say you’re the average American, out and about on a day during typical U.S. peak flu season. You incur a (small) risk of getting flu, and an even smaller risk of getting a case that’s so severe that you’ll end up either hospitalized or dead. And you typically think nothing of it.
How rare would new Omicron case have to be, in order for Omicron to pose the same level of risk, compared to a peak week of seasonal flu?
And I want to know that for the fully-vaccinated population, in each case. Vaccinated against flu, and vaccinated and boostered against Omicron.
There is going to be a lot of imprecision in this analysis. Rather than bore you with the methodology first, I’ve decided to present the numbers first. And bore you with methods last, if I have the time to write that up. Just rest assured that a lot of quantitatively important issues are being swept under the rug at this point.
I’m boiling this down to risk of hospitalization and death, fully acknowledging that there are aspects of illness beyond those two simple measures. The problems are that a) nobody tracks the numbers on those other aspects of COVID-19 illness, and b) there’s no flu equivalent to things like long COVID, or loss of sense of taste and smell.
In short, the only possible head-to-head comparison between COVID-19 and flu is in terms of risk of hospitalization and death. Because those are the two things that are routinely tracked, and that are common to both illnesses.
The plan of attack is as follows:
- Get the raw (average) case hospitalization rate and case mortality rate data put down in black-and-white.
- Do the crude flu-versus-Omicron comparison based on those raw numbers.
- As and if possible, adjust the numbers to account for differences in severity between the vaccinated and un-vaccinated populations.
Given the vast uncertainties involved, at the end of this, I’ll just wave my hands about methodological uncertainties. I mean, at this point, we don’t even really know whether Omicron is what we’ll be living with in the long run. So there’s no justification in splitting hairs on the accuracy of the analysis.
Step 1: The basic numbers on case hospitalization and case mortality rate
One step at a time.
1.1 Flu case hospitalization and mortality rates.
Let me start from the CDC’s estimates of the illness burden of flu, on this CDC web page. Here, I’ve just ignored the statistical uncertainty (the 95% confidence intervals) and taken the median of values for the past ten US flu seasons. I’m also ignoring some potentially significant differences between the definition of “a case” as used here, and as used in the Omicron analysis below.
Source: Calculated from CDC illness burden of flu web page.
In a typical year, calculating from the CDC data, in the U.S., 1.4% of persons with a symptomatic case of the flu end up in the hospital, and 0.13% die. So those are the benchmarks for something we can routinely live with.
1.2 Omicron case hospitalization and mortality rates.
Source: Calculated from CDC COVID data tracker files access 1-19-2022.
As I have noted in numerous prior posts, there’s essentially no lag between new cases as reported, and new hospitalizations as reported. Therefore we can calculate a true case hospitalization rate right on out to the last available day of data.
Mortality is not so straightforward. In theory, you only know the ultimate mortality rate months after-the-fact, because you have to wait to see whether or not people diagnosed on a given day ended up dying.
In practice, I’ve had good success with a much cruder approach, which is to divide current deaths by new cases from two weeks earlier. This accounts for the roughly-two-week median time from diagnosis to death for COVID-19 decedents. (Whether or not that has remained at roughly two weeks, with Omicron, is not known at this time).
From the look of the graph, the case hospitalization rate has stabilized, and is currently about 2.6 percent. The case mortality rate appears to be continuing to fall, and I’m not sure what to make of that. The final value in the series is 0.34%, so I’m sticking with that.
1.3 Final estimates, flu versus Omicron case rates, raw data (no adjustments for impact of vaccination).
And so we get the table above, showing un-adjusted values. In round numbers, a symptomatic Omicron infection is about twice as likely to land you in the hospital, and about three times as likely to kill you, compared to a symptomatic flu infection.
I want to note that this is vastly better than the way things looked earlier in the pandemic. The very first wave of the pandemic had a case mortality rate around 8 percent, both here, and in China, and in the U.K. Later, the Delta wave — which occurred after mass vaccination of the population — had a case mortality rate and case hospitalization rate were more than three times what’s shown above. The point being that the current situation really is different from what has gone on before.
In short, with Omicron, and with a lot of vaccination and prior infection, we’ve finally gotten a COVID-19 variant whose danger is somewhere in the ballpark of that of flu. That’s a first for the pandemic.
Riskier, for sure. Twice the hospitalization rate, three times the mortality rate. But no longer orders-of-magnitude riskier.
Step 2: The crude calculation, before considering the separate impact of vaccines.
This part is not rocket science. If COVID’s case rates for hospitalization and death are 2-times and 3-times that of flu, then COVID presents the same risk as flu when your odds of getting infected are one-half (hospitalization) to one-third (death) as high as they are for flu.
2.1 Just how bad is a typical peak week of flu in the U.S.? How bad is the average week, year-round?
As noted in the prior section, the U.S. sees about 30 million symptomatic flu cases in a typical year, according to the U.S. CDC. To allocate that to individual weeks, and find the peak rate, I used the 2017-2018 flu season as a model, and assumed that symptomatic flu cases occurred in proportion to the CDC’s estimate of visits for influenza-like-illness at their network of sentinel (reporting) providers. All of the underlying data come for the CDC web pages on the disease burden of flu.
When I do that, I come up with about 2.1M new U.S. symptomatic flu cases in a typical peak flu week.
Translating that to the language we use for Omicron, that works out to 100,000 x ((2.1M / 330M) / 7) = 91 new flu cases per 100K population per day at the peak of flu season.
If I crudely define “flu season” as the 18 weeks were new case rates remain at or about 500,000 per week, then the average daily risk of flu, during flu season works out to an average of 100,000 x ((20.6M / 330M) / (18 x 7)) = 49 cases per 100K population per day.
2.2 The crude calculation
In round numbers, Omicron has twice the case hospitalization rate and three times the case mortality rate. Taking the greater of those two, the “case rate risk” of Omicron is three times that of flu, averaged over all persons. So we’d need to see one-third the number of new Omicron cases for the overall risks to be equal.
If we take the peak of flu season as our benchmark for risk, the mortality risk from Omicron infection would be no higher than that of flu when Omicron gets down to 30 new cases / 100K / day. If we take the entire flu season as the benchmark, then the risk from Omicron matches the all-season risk from flu when Omicron gets down to about 16 cases per 100K per day.
Conclusion for now.
I think this has been a good start, and I’m going to stop right there, for the time being.
The clear bottom line is that, at present, Omicron poses a vastly higher risk of hospitalization and death, for the average American, relative to seasonal flu. Call it ten-fold higher risk.
In part, that’s because it’s two to three times as virulent, per case. But mostly, that because there’s just so much of it in circulation.
I’m starting to home in on a level of Omicron at which, if I am rational and risk-neutral, I should be no more worried about Omicron than I am about season flu. Based on this first cut of the calculation, that ought to occur at either 30 new Omicron cases / 100K / day (if peak flu week is your benchmark), or 16 new Omicron cases / 100K / day (if all of flu season is your benchmark).
While that seems like it will be an eternity from now, if cases continue begin to fall at 30 percent per week (typical internationally, slower than what we see in individual states right now, per last post), it would only take about six weeks for case rates to get down to that upper benchmark level.
Mid-March? Will Omicron in the U.S. reach that first benchmark by mid-March? If so, and if I’m rational, and risk neutral, and the next analysis (including vaccination) does not materially alter this, then that’s the point at which I resume all prior activities. (Maybe masked, because why not).
Because, to put it plainly, it would never even occur to me to avoid going to the movies during flu season. Just would not cross my mind. And that’s now my benchmark for que sera sera. It’s the point at which I’m willing to throw in the towel and say I’ll just live with endemic COVID.
The final question is whether these benchmarks are materially incorrect for the fully-vaccinated? My suspicion is that they are, but they err on the side of caution. That is, if I can work up a comparison of a fully-flu-vaccinated and fully-COVID-vaccinated-and-boosted individual, I’ll find that the COVID vaccine is way more effective than the typical flu vaccine.
Putting that another way, all these years that I have faithfully gotten the flu vaccine, I had no idea what poor performance flu vaccines offer. Worth doing, for sure. But nowhere near as effective as COVID-19 vaccine plus booster.
If I can find the data, and work out the math, I’ll do that final phase of the calculation. But what I have found to date is that looking into how the U.S. monitors and treats flu is an exercise in turning over rocks and seeing what crawls out from beneath. Every time I look into the details, I wish that I hadn’t. So I may or may not get around to my ultimate goal, which is to find these benchmark rates for completely vaccinated individuals.
