In many data systems, a “COVID hospitalization” is a hospitalization of a person who has tested positive for COVID. Period. This includes:
- people who are being treated for COVID
- people who are being treated for something else entirely:
- for whom the finding of a COVID infection is completely irrelevant or
- for whom COVID might have exacerbated the underlying condition that got them hospitalized.
My guess is that the final category involves relatively few people. So, crudely, this breaks down into two groups:
1: People admitted for COVID, who wouldn’t be in the hospital if COVID hadn’t been around, and who represent a true additional burden on the hospital system caused by the pandemic. And then,
2: People admitted with-but-not-for-COVID, who would be in the hospital in any case, and who don’t represent any additional burden on the hospital system caused by COVID. For the simple reason that they are sick with something else.
Any additional cost burden from that second group is truly incidental. They have to be kept isolated, and they are capable of infecting hospital staff with COVID. So the cost of their hospital stay may be higher than it would normally have been. But they would have been hospitalized COVID or not.
This is now greatly affecting U.S. COVID hospitalization counts.
In prior waves of COVID, this distinction didn’t matter much. The overwhelming majority of patients admitted with COVID were being admitted for treatment of COVID.
In the Omicron wave, that’s no longer true.
This was first noted in South Africa, where one large hospital system there found that 76% of their patients “with COVID” were in the hospital for treatment of something else. This was particularly true of their pediatric patients in South Africa, most of whom had an asymptomatic COVID infection that was only found after they were admitted to the hospital for the treatment of some completely unrelated condition.
Now I’m seeing reports of the same thing happening in U.S. hospitals (emphasis mine):
Roughly two-thirds of patients who have tested positive at hospitals run by the L.A. County Department of Health Services were admitted for something other than the coronavirus, according to Health Services Director Dr. Christina Ghaly. This time around, many “may have not known they were COVID-positive ... but they’re in the hospital for something else,” Ghaly said.
Source: LA Times, 1/4/2022.
Dr. Fritz François, chief of hospital operations at NYU Langone Health in New York City, said about 65% of patients admitted to that system with COVID-19 recently were primarily hospitalized for something else and were incidentally found to have the virus.
Source: AP News, 1/5/2022
Compared with earlier waves, a smaller percentage need intensive care. At Hackensack Meridian Health's hospitals, 140 of 1,100 COVID patients on Tuesday, or 13%, were in ICUs, and 75 were on ventilators. That's down from about 20% requiring ICU care at the pandemic's start. Some patients are admitted for other illnesses, only to discover after admission testing that they have COVID. Over the last month, that portion has increased to “45% who were admitted for something else but were found to have COVID, too,” Bennett said.
Source: NorthJersey.com, 1/5/2022
Gov. Ron DeSantis said 50% of COVID hospitalizations at Orlando Health are not being treated for the virus. “We have found that a great majority of our hospital in-patients who have been diagnosed with COVID-19, were admitted for medical conditions not related to the virus,” Orlando Health spokeswoman Alayna Curry said.
Source: ClickOrlando, 1/4/2022
I am pretty sure I could find more if I looked.
Why does this matter?
Pediatric admission trend. First, I will point out that numerous sources — including Dr. Fauci — suggest that this is particularly important issue for pediatric admissions. I.e., that the number of children admitted for COVID in prior waves is far less than the number that has been counted (per this research).
And so, currently, pediatric admissions are increasing as a fraction of all COVID admissions. Does that mean that Omicron differentially attacks children? Or does this just mean that these “incident” cases matter more in this group, where the number of admissions for COVID-19 is quite small? Nobody can give you an answer to that.
True comparisons of illness severity. One of the main ways everyone compares Omicron to Delta is by the likelihood that it will put you in the hospital, if you catch it. But if a larger share of Omicron hospitalizations are persons for whom COVID-19 is an incidental finding, it makes Omicron appear to have higher severity than it actually does.
True estimates of hospital stress. If half of the Omicron cases in the hospital are people who would have been hospitalized anyway, regardless of Omicron, then the raw hospitalization count overstates the true degree of hospital stress because caused by Omicron.
Finally, it’s not even clear what hospitals are reporting to the Federal government. For the stock of patients in the hospital, it’s clear that the definition includes anyone with a positive COVID-19 test. For “new admissions”, the guidance says to count patients ” … who had confirmed COVID-19 at the time of admission … ” . If the hospital gives every incoming patient a rapid test for COVID-19, then, aside from the high false-negative rate of those tests, presumably, they are going to count every patient with detectable COVID at time of admission. (I went through all of this in Post #1394).
My contention is that everybody talks about that new admissions number, but it’s not crystal clear what it actually represents. Plausibly, it does include most of the “incidental” COVID-19 cases, if they are found by antigen testing during the hospital admission process.
A big problem with the math.
Here’s my main problem with this. There’s no plausible set of circumstances that I can come up with that would generate that many incidental COVID-19 cases. I.e., that many people who were hospitalized for something else, and just happened to have COVID discovered once admitted to the hospital.
It’s easy enough to understand how these incidental COVID-19 cases get into the hospital. In the U.S., in 2018, per the AHRQ HCUP hospital database, 55% of hospital admissions occurred via the emergency room. (And so were presumed to be emergency admissions). Those patients are going to have to be admitted no matter what. Another chunk would be “urgent” admissions that cannot be postponed indefinitely. Those people would likely be admitted regardless of COVID-19 status. (And somewhere around 20% are “elective” or “scheduled” admission where, presumably, anybody testing positive for COVID would not be admitted.)
And so, presumably, these “incidental” COVID cases are just randomly distributed across all emergency (and urgent) hospitalizations, based on the prevalence of COVID in the community. For emergency and urgent care, those persons end up hospitalized. And they would carry any existing COVID infection to the hospital with them.
In other words, if 1 percent of the people in your community are walking around with an active COVID-19 infection, then, to a first approximation, about 1 percent of those showing up to be admitted to the hospital for any random condition will have a COVID-19 infection.
The problem is that a) yeah, somewhere around 1 percent of the population probably is walking around with an active COVID infection right now, but b) 1 percent of ongoing hospital admissions wouldn’t come close to explaining the high number of cases admitted to the hospital with COVID as an incidental finding.
Here’s the math.
There’s nothing tricky here.
- First, the U.S. has about 30 hospitalizations per 100K population per day (in a normal year).
- Second, based on the number of daily new COVID-19 cases, I estimate that just over 1% of the U.S. population is currently walking around with COVID. (I’ve gone through my factor-of-six rule of thumb in prior posts).
- Multiply those two together, and you get an estimated 0.4 “incidental” COVID-19 admissions per 100K per day.
- But right now, we’re getting 5.5 COVID admissions per 100K per day.
- So … those incidental admission would only amount to 7% of the total. Nothing close to the 50% to 66% being reported anecdotally.
This has been puzzling me for weeks, ever since the South African finding came out. And it boils down to “something most be wrong with the assumptions behind that calculation”. But I have yet to figure out what.
One possibility is that there are vastly more asymptomatic Omicron cases than anyone has estimated. There is a tiny amount of evidence to suggest that this might be plausible. (I.e., this one research study). But you would need so many cases that I don’t think you could account for this with asymptomatic infections alone.
That is, to bring that 7% figure up to 50%, you’d need to have seven times as many infections present in the population. I think that, at that rate, the entire population would have been infected with Omicron within a couple of weeks. It just requires too many people to have an active infection.
A second possibility is that persons who need hospitalization for some other condition have a vastly higher rate of COVID-19 infections. But recall that most of these individuals are entering the hospital with asymptomatic COVID. So it’s not like the finding that certain comorbodities increase risk of hospitalization for COVID-19. Here, you’d have to believe that comorbidities likely to get you hospitalized also vastly increase your likelihood of asymptomatic COVID infection. I can’t quite get my mind around that.
A third possibility is that maybe tests are turning up a lot of false positives on old COVID-19 infections. So that, as these persons enter the hospital for some other reason, and get tested for COVID, we’re picking up not just those currently infected, but a significant fraction of those who have been infected sometime in (say) the past 90 days.
But, while such false positives do occur with PCR tests, I don’t think anybody says that they are common. (The possibility of false-positives in that timeframe is why you are not supposed to get routinely re-tested for COVID within 90 days of a positive test.) And nobody, so far, suggests that such false positives are more common with Omicron than with prior variants.
None of these explanations seems plausible to me, given the number of persons that would have to be involved.
Let me just spin the numbers the other way, and you can see what I’m talking about, in terms of not being able to get the math to work out.
On any given day, per 100,000 population, these days the U.S. would see
- about 30 hospitalizations unrelated to COVID-19
- about 6 COVID-19 hospitalizations.
If half of those COVID hospitalizations are incidental ones (where the person is admitted for something else, but has an asymptomatic case of COVID), that means you need to see 3 “incidental” COVID admissions coming out of the 30 individuals who were admitted for other causes.
In other words, if half the COVID admissions are “incidental”, that requires that 10 percent of all regular hospital admissions have an asymptomatic COVID infection upon entering the hospital. (Or manage to trigger a positive on a COVID-19 test for some other reason.)
Which, to a first approximation, requires that, at any given time, 10 percent of the population would have to have an asymptomatic COVID infection?
Which I think is not even remotely plausible.
And that’s where I give up. I cannot gin up any plausible scenario under which half the current set of COVID-19 admissions are “incidental” COVID cases. That would require something like 10% of all regular hospital admissions have asymptomatic COVID. Or somehow, certain diseases and conditions attract asymptomatic COVID infections. Or Omicron results in a huge number of false positives, based on having had a prior (not active) infection.
None of those seems plausible. But without something along those lines, the numbers just don’t make sense. At current rates, if half of COVID-19 admissions are “incidental”, that equates to 10% of all admissions for treatment of anything but COVID walking in the door with an active asymptomatic COVID infection.