Post #1556: COVID-19 trend, cases up, can’t blame it on July 4th

Posted on July 14, 2022

 

The US now stands at 39 new COVID-19 cases per 100K population per day, up about 20% in the past seven days.  I can no longer blame the uptick on July 4th data reporting.  So this appears to push the U.S. up above the level of daily new cases that has prevailed since late May of this year.

Data source for this and other graphs of new case counts:  Calculated from The New York Times. (2021). Coronavirus (Covid-19) Data in the United States. Retrieved 7/14/2022, from https://github.com/nytimes/covid-19-data.”  The NY Times U.S. tracking page may be found at https://www.nytimes.com/interactive/2020/us/coronavirus-us-cases.html

I spent yesterday’s post trying to explain away an unexpectedly high case count as probably just an artifact of July 4th data reporting.  But at this point, all the noise from July 4th should be beyond the seven-day-moving-average window.  The upshot is that the new case count did come down — a bit — once we got past all of that July 4th disturbance.  But only a bit. 

So that, plus the fact that this uptick is widespread (not just a few regions or states), strongly suggests that it’s real.  The little circled area above is most easily interpreted as the impact of July 4th data reporting issues superimposed on an upward trend.

At this point, I think I have absolutely no other data source that I can use to validate this.

Back in the day, cases and hospitalizations were closely synchronized.  Reported cases and reported hospitalizations rose and fell virtually simultaneously (if not to the same extent).  You could see that (e.g.) a decline in case counts during the holidays was a reporting artifact, because hospitalizations would not dip.

But cases and hospitalizations went their separate ways a month and a half ago.  We’re now closing in on 6000 daily new COVID-19 hospitalizations.  With almost no increase in daily new cases.

Source:  CDC COVID data tracker.

Everything else — self-reported COVID symptoms, office visits for COVID-like symptoms — is either no longer available or weeks out-of-date.Looks like Carnegie-Mellon has stopped gathering information on persons with self-reported COVID-19 symptoms.  Other measures (e.g., office visits with COVID-like symptoms) are weeks to months out-of-date.

All I can say is, it looks like we’re facing a bit of an increases in officially-reported daily new cases.  After maybe seven weeks of no change. 

In the past, there was also a reliable link between new COVID-19 cases, hospitalizations, and deaths.  Hospitalizations would rise in tandem with cases, and deaths would follow two weeks later.

That all broke down on or about May 21 of this year.

Noted above, for the last seven weeks, new cases have been flat, but hospitalizations have been rising steadily.  If anything, their rising faster now than they were a month ago.  We entered this period at just over 3000 COVID-19 hospitalizations per day.  We’re now closing in on 6000.

And yet, despite rising hospitalizations, daily COVID-19 deaths remains right at 300, same level as was occurring seven weeks ago. 

Speculation

I can only guess as to what might be going on.  You have to start by saying that the COVID-19 characteristics of the U.S. population are reasonably constant at this point.  There’s no huge push for new immunization.  There’s no huge ongoing wave.  There’s a steady stream of daily new infections, and there’s the slow loss of immunity as time passes since last vaccine, booster, or episode of infection.

Really, the only thing that’s changed much, over this time period, is that BA.4 and BA.5 have taken over as the dominant strains.  Could that be causing this?

1:  Maybe its real.

On the one hand, the hospitalizations number may be real, and reflect the cumulative loss of immunity among those with high risk of hospitalization, or maybe some different behavior of BA.4 and BA.5 compared to prior strains.  So that, either way, you really are more likely to end up hospitalized for COVID-19 now, than you were seven weeks ago.

But loss of immunity seems an unlikely cause.  The CDC still says that vaccine-related protection against hospitalization remains strong.  Even months after a booster.  Even with the Omicron variant.  You can see that in the vaccine section of the COVID data tracker website.  And any loss of immunity should have been occurring at a fairly steady rate, not just in the past seven weeks.

It could also be a real increase if BA.4. and BA.5 are more likely to result in hospitalization than the prior strains.  But I don’t think the international experience supports that.  And surely, if that were known, it would make irresistible click-bait.  So we’d have seen headlines screaming about it.

And in either case, it’s hard to believe that — for whatever reason — far more people are being sickened enough to require hospitalization.  Yet no more of those are dying while hospitalized.  That makes no sense at all.

2:  Maybe it’s not

On the other hand, maybe the hospitalization number is an artifact of methods.

One way that could happen is with an increase in late (i.e., post-recovery) false-positive COVID tests.  As I noted on a long-ago post (Post #1351), the hospitalization number is persons hospitalized with COVID-19 (with principal or secondary diagnosis of COVID-19), not persons hospitalized for COVID-19 (with principal diagnosis of COVID-19).  It’s persons who were hospitalized, who also tested positive for COVID-19 at that time.

Unfortunately, nobody tracks the information to separate those two — cases with principal diagnosis of COVID, from cases with some secondary diagnosis of COVID.  Years from now, sure, data will be available to allow you to check principal versus secondary diagnosis.  But in real time, nobody is doing that.

I didn’t make much of that for-versus-with COVID-19 issue before, because the numbers didn’t support it.  The number of actual active-but-asymptomatic COVID-19 cases was small enough, as a fraction of the population, that you wouldn’t expect to see them boost the hospitalization numbers much.

An interesting point is that everybody admitted to a hospital these days gets tested for COVID-19.  In effect, that population is screened for COVID — tested whether there’s any reason to suspect it or not.  And so, another way to get the same effect is not with true active-but-asymptomatic cases, but with an increase in the late false positive rates for COVID-19 tests.

We know that in some cases, individuals test positive for COVID-19 months after they have recovered.  Further analysis shows that the tests are picking up fragments of dead virus, not live virus, from those persons.  And, finally, that these leftover fragments were more likely to occur if a person had a high viral load in the first case.  The greater the amount of virus in a person originally, apparently, the longer it takes to clear all of those fragment out, so the longer it takes before you stop triggering positives on COVID-19 tests.

Is it possible that post-infection false positives are more common with the new variants (BA.4 and BA.5), than they were with prior variants?  That might plausibly be able to generate enough false-positive results, out of the entire population being hospitalized, to drive up the “with COVID 19” number.  Unfortunately, nobody has even suggested this, let alone put it to the test, or publish data showing average viral loads for BA.4 and BA.5 relative to earlier strains.

As was true in the past, it’s hard to make the numbers work out.  There just aren’t enough COVID-19 cases in the population.  Between May 21 and today, the U.S. has accumulated more than 5 million additional COVID-19 cases, per the official statistics.  Maybe the true count is three times that, or 15 million.  In very round numbers, you’d estimate that somewhere around 5% of the U.S. population had a COVID-19 infection in the past seven weeks.  Of which maybe half was the new strains.  So, something like 2.5% of the entire U.S. population has been infected with BA.4 or BA.5, over the past seven weeks.

What would happen to the COVID-19 hospitalization count if all of those people generated a false positive?  The U.S. has about 100,000 hospital admissions on the average day.  Then 2.5% of that is 2500 hospitalizations.  Which is just about exactly equal to the observed increase in COVID-19 hospitalizations over the past seven weeks.

In other words, to account for this rise in COVID-19 admissions from false positives alone, you’d have to posit that every BA.4 or BA.5 infection resulted in a post-infection false positive.  And that’s just not credible.  So, as was the case before, even if I stretch the numbers, it’s hard to believe that could be the driver behind this increase in admissions.

Yet a second way we might see a rising U.S. hospitalization rate without any true increase in hospitalizations would be via a shift in the geographic location of the new cases.  States vary wildly in terms of their COVID-19 hospitalization case rates.  If we’ve simply seen a shift in cases from low-rate to high-rate states, that could explain a rising U.S. hospitalization rate with no rise in the new case rate.

I can rule that out just by looking at a few large states, all directly from the CDC COVID data tracker.  This divergence between cases and hospitalizations is present in each of California, Florida, and Texas.  So this is happening within states, and the national number is not merely an artifact of a shift in cases across states.

3:  I remain baffled at the breakdown between cases, hospitalizations, and deaths.

First, I can’t believe it’s some real phenomenon.  I don’t understand how there could be (e.g.) twice as many people per day who are sick enough to need hospitalizations for COVID-19, but no increase in deaths out of that much larger hospitalized population.  Generally speaking, for any disease, severity of illness is a spectrum.  If there’s an increase in the number of people who require an inpatient stay for a disease, there will be a corresponding increase in the number of people who die during those inpatient stays.

Second, I can’t believe it’s just some artifact of methods, such as an increase in false positive tests among individuals hospitalized for non-COVID reasons.  Of the things that might result in a sharp increase in persons merely hospitalized with COVID, the one that makes the most sense to me is an increase in false positives on COVID-19 PCR (DNA) tests.  But the numbers just don’t work out.

Beats me.  And the implications for health care risks are vastly different if this is a true increase in hospitalizations for COVID, as compared to some artifact of how these admissions are tested or counted.

I wish somebody with access to the detailed data would try to sort this out.