Category: COVID through 2022

 

FWIW, the U.S. now stands at 33 new cases per 100K population per day, based on the official counts of tests.  Plus-or-minus reporting variations due to Memorial Day, it has been at that level for the past three weeks. Continue reading Post #1530, COVID-19 trend to 6/7/2022, now 33 cases, still feeling the effects of disturbed data reporting.

 

The official U.S. new COVID-19 case count still stands at 31 per 100K per day, up 7% in the past seven days.  The U.S. average has been at more-or-less the same level for two weeks now.

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 6/4/2022, 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

The CDC shows us at well under 300 deaths per day, still.  Hospitalizations appear to be topping out around 3900 per day.

Surely I can’t let a day pass without spreading a bit of gloom and doom.  Today’s nugget of negativity is that we’re less than a month from what ought to be the low point for the year, for new cases, based on the first two years of the pandemic.

So I look at that and I say, well, maybe this is about as good as it gets.  We’ve got this ludicrously contagious disease — it’s been a while since I’ve seen anybody even venture a guess as to the R-nought, but the last one I recall was somewhere around 21.  (Compared to maybe 1.5 for seasonal flu.)   We’re at the point where reinfections are common and vaccines do almost nothing to prevent infection (though they are still quite effective against severe infection).

But eventually, I come to my senses and look around.  If there is some “natural” rate of new infections, it surely isn’t apparent by looking internationally.   Below is a view of the world’s daily COVID-19 infection rates since 1/1/2022.  Although the new infection rates appear to be stabilizing around the world, they are stabilizing at vastly different reported rates in different areas.  The U.S. is the middle of the pack — a very spread-out pack.

Source:  Our world in data, annotations mine.

You can’t rationally look at that and conclude that we’re stuck at our current new case rate forever.  My conclusion is that plenty of industrialized nations have lower rates than we do right now.  And there’s no obvious reason why our rate should remain permanently high.

 

Today, the U.S. new COVID-19 case count rounds up to 31 per 100K population per day.  That’s down 7 percent in the past seven days.  Give all the reporting noise surrounding the Memorial Day holiday, the best way to characterize it is to say that new cases remain more-or-less flat. Continue reading Post #1527: COVID-19 trend to 6/2/2022, still topping out.

 

Officially, the U.S. now has 30 new COVID-19 cases per 100K population per day.  That’s down 7 percent over the past seven days.

That’s up a bit from yesterday’s figure of 29, due to some catch-up post-holiday reporting in California and a few other states.

That said, it still appears that the official count of daily new COVID-19 cases has peaked, for now.

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 6/2/2022, 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

No clue what the real count is.

It sure would be helpful if we could get some better estimate of the actual number of new cases in circulation.  I bring this up because I’ve heard reports of new infections from two different people in my small circle of acquaintances.  So I’m back to wondering just how thick on the ground actual new infections are, compared to the official statistics.

The cluster-y nature of new infections makes it hard to based any estimate on your own personal observations.  Not only are there relatively few new infections, but they tend to come in large, related batches.  In that sort of situation, you can easily see an infrequent “spike” within any small population.  I can’t really conclude much from the two sets of cases I heard about recently.

There’s no doubt that official count is far lower than the true count of new cases.  That’s been true throughout the pandemic.  The number of cases captured via the official test data has always been a substantial underestimate of total cases.  As of a few months ago, the official count was about half of what the CDC estimated from the presence of antibodies in blood samples (“seroprevalence”).

There’s a couple of factors suggesting that the degree of underestimate should have grown in recent months.  The first is the widespread availability of over-the-counter COVID-19 tests.  The second is the reduced severity of the average new case, as an ever-increasing fraction of cases are occurring in individuals with some pre-existing immunity (either re-infections or breakthrough cases in the vaccinated).  Together, those mean that a) more people can figure out their COVID-19 status on their own, and b) a small fraction of new cases require medical intervention and the associated official test results.

Wouldn’t it be nice to have some estimate of what that actual new-case rate is doing?

Falling average case severity means that we can no longer rely on either deaths or hospitalizations as a good measure of new case prevalence.  The rising fraction of cases that are re-infections and breakthrough infections means that the death and hospitalization case rates (deaths per new infection, hospitalizations per new infection) ought to be falling.

The CDC’s most recent seroprevalence estimate is still from February 2022.  That’s now four months old, and so predates the entire Omicron-II wave.  I should point out that that’s far later than in the past.  Looking back in my records, I see a snapshot of the October 2021 survey dated early December 2021, and the January 2022 survey dated March 1, 2022.  So, historically, CDC was able to get that information out about a month after they got the blood samples.  I have no idea why that’s now a month late.  But the bottom line is that we have no new information from that source.  And if that eventually shows up, it’ll tell us what things looked like back in March 2022.

I looked again at Tompkins County, NY — the only jurisdiction I found that tracked self-reported positives from home tests.  They stopped collecting that information months ago.

To sum it up:  Zilch.  Nothing.  There appears to be no data source from which to guess what the actual new-case rate is in my area, or even for the U.S. population as a whole.

 

The trend numbers for this week will be perturbed by  the Memorial Day holiday.  So you don’t want to over-analyze this most recent development.

That said, nominally, the U.S. now stands at 29 new COVID-19 cases per 100K population per day.  That’s down from 33 just before the holiday, and that’s down 7 percent over the past seven days.

Some chunk of that is due to clear under-reporting in California, so I would expect to see a bit of a rebound.   That said, it does appear that outside of the Mountain region, new case counts are flat-to-declining in most parts of the country.  It this isn’t the peak of the Omicron-II wave, it’s close. Continue reading Post #1525: COVID-19 trend to 5/31/2022, over the hump?

 

If you take the available research at face value, a huge fraction of the U.S. population now suffers from debilitating long-term consequences of COVID-19.

Just last week, a CDC-sponsored study (reviewed below) seemed to imply that 13 percent of the entire U.S. adult population has some form of long COVID.

(They estimated that ~22 percent of persons who had COVID had some long-term health consequences, above-and-beyond the baseline rate at which those conditions arose in the non-COVID population.  (Literally, 16% of non-COVID patients had some mention of a relevant health condition,versus 38% of COVID-19 patients, for a difference of 22%. When multiplied by the most recent CDC seroprevalence estimate that almost 60% of U.S. adults have had COVID, you would come to the conclusion that (22% x 60% =~) 13% of the U.S. adult population has long COVID.)

One-in-eight U.S. adults has long COVID? 

Really?   This is very much at odds with my perception of reality. 

Surely there are some cases of serious long-term health consequences from COVID.  To be clear, any acute condition that nearly kills a person will likely result in some long-term effects.

What I don’t believe is that serious long-term health outcomes from COVID infection could possibly be that common.  If so, U.S. health care rehab facilities, outpatient departments, and physician offices would be overflowing with followup care for COVID survivors.

So, what’s the answer?  How can a seemingly-legitimate CDC-sponsored study find such huge fraction of COVID-19 cases have “long COVID”, and yet, there appears to be no proportionate impact on the U.S. health care system.

I think the answer has two parts.

---

The CDC analysis is a study of COVID-19 cases sick enough to require hospital-based intervention.

First, if you read it closely, that CDC study is clearly a study of U.S. adults who were so sick with COVID-19 that they sought medical intervention.  The underlying data are from electronic medical records.  Such records are only generated when individuals use the health care system in some fashion.  Aside from a handful of individuals who might have gone to their health care provider to get a COVID-19 test, that means we’re looking (almost exclusively) at people who sought treatment for their COVID.

But it gets worse.  This study was based on a proprietary database of electronic medical records.  You then have to pursue some description of what, exactly, is in that database.  Unfortunately, the descriptions of the proprietary data source are marketing-oriented, not technically-oriented, but a) the data underlying this study come almost exclusively from hospital systems, and b) it certainly appears that the overwhelming majority of the information is from hospital inpatient and ER encounters.

In other words, this most recent CDC study is primarily a study of U.S. adults who had such a bad case of COVID-19 that they sought help at a hospital. 

To be honest, the underlying data source is just sufficiently ambiguous enough that that’s not an air-tight claim.  But that’s certainly how it appears to me.

Assuming I’m correct, now the results make sense.  Sure, a high fraction of those who had a severe or life-threatening case of COVID-19 have significant long-term aftereffects.  That’s completely unsurprising, because almost anything that results in a near-death experience (think stroke or heart attack) will have serious long-term consequences.

But, as will become obvious from the numbers below, under no circumstances should you interpret the latest CDC study as showing outcomes for the typical adult COVID-19 case. 

And, to be clear, I think it’s hugely misleading of the CDC to fail to make that crystal-clear.  Because that’s exactly how the popular press is reporting it.  And that’s exactly how U.S. citizens are going to be interpreting it.  And the last thing the CDC needs, in the context of COVID-19, is to be caught out at what appears to be one last round of scare-mongering.  Even if, technically, they have done the appropriate amount of vague CYA (caveats) at the end of the research paper.

A quick walk through the numbers that almost everyone ignores

I have read and done a lot of health services research over the years.  I have learned the hard way that in some cases, the most important numbers are buried in the footnotes.  In particular, there’s a reason that most scholarly journals will make you show exactly how you got down to your final sample of data.  What did you start with, whom did you throw out, and how did you justify throwing them out.

And in this case, looking at the starting population is the key to understanding these results.  This CDC study ends up with about 350,000 U.S. adults with COVID.  But the kicker is, that’s drawn from database of >63,000,000 U.S. adults, a fact that is just casually mentioned in passing in this research.

This, over a time period when 31% of U.S. adults had a case of COVID-19, based CDC data.

Those of you capable of doing some crude arithmetic on the fly will realize that magnitude of the discrepancy here.  That 350,000/63,000,000 ain’t nowhere near 31%.  It’s not even close to 3.1%.  It’s more like 0.5%.

So, from the get-go, without any further analysis, you immediately realize that this is a study of a tiny, tiny subset of adult COVID-19 cases.  I’ll do the math a little more precisely below but in round numbers, this is a study of less than 1 percent of all the adult COVID-19 cases that should have occurred in that population of 63 million.

But tiny is tiny no matter how you slice it. This ends up being a study of a tiny fraction of COVID-19 cases.

And my key point is that it’s  not some randomly-chosen set of cases.  This tiny subset consists of a) everyone who was hospitalized for COVID, b) probably most people who were sick enough to end up in the hospital ER for COVID, and c) an unknown number of others who sought formal medical care for COVID at a physician’s office, but required no more intense care.

In fact, I’ll just go ahead and say it.  Based on U.S. prevalence and hospitalization data, I would be unsurprised if this database contained much more than the COVID-19 hospitalized population.

Here’s the citation for the study.

Bull-Otterson L, Baca S, Saydah S, et al. Post–COVID Conditions Among Adult COVID-19 Survivors Aged 18–64 and ≥65 Years — United States, March 2020–November 2021. MMWR Morb Mortal Wkly Rep 2022;71:713–717. DOI: http://dx.doi.org/10.15585/mmwr.mm7121e1.

The CDC study starts from a proprietary for-profit database provider who claims to have medical records for 63 million U.S. adults.  If you then read the description of the data source and see the resulting research, it’s immediately obvious that this is a hospital-inpatient-centric database.  (See here, for example).  So they surely have hospital inpatient encounters, likely have hospital OPD and ER encounters.  And the extent to which they have anything beyond that is unclear.  But, for sure, if you know what to look for, virtually all the research shown on their website, using this database, is research about inpatient care.

The researchers then identified 350,000 adults with (what I interpret as) some health care encounter with a diagnosis of COVID-19, from the start of the pandemic through November 2021.  But in addition, they appear to have screened out about one-fifth of all otherwise eligible cases, due to some pre-existing condition.  Accounting for that, they would have found about (350K x 5/4 =) 440K adults with (what I assume to be) some COVID-19 related treatment.

Doing the long division, they appear to have found about 440,000 adults in their database, with COVID, in a claimed population of 63 million adults.   Or (440K / 63M =) roughly 0.7 percent of their adult population appears to have had a case of COVID-19.  Of which some where tossed out for having a relevant pre-existing condition.

But as of November 2021, the CDC’s seroprevalence surveys show that >31% of the U.S. adult population had already had COVID-19.

Source:  CDC COVID data tracker.

Now here’s where this gets just plain ugly.  As of November 2021, the CDC estimates that about 0.9% of the entire U.S. adult population had been hospitalized for COVID-19.   (Source, CDC COVIDnet, accessed 5/31/2022).  But of those, about 15% to 20% died in the hospital.  That would suggest that about (0.9% x 85% =~) 0.75% of the entire U.S adult population survived a COVID-19 hospitalization over this period.  That’s an estimate based on a sample of hospitals, but it’s the best estimate available for the U.S.

In other words, on the face of it, based on the case counts and underlying population as-reported, this study appears to be primarily, perhaps almost exclusively, a study of individuals who were hospitalized with COVID-19.  There simply are not enough cases in their database to account for anything much above-and-beyond the expected number of COVID-19 hospitalization survivors in a population of 63 million U.S. adults.

OK, now it makes sense.  If you told me that being hospitalized for COVID had a 20% risk of long-term consequences, then sure, I’d believe that.  Being hospitalized for any condition with a 20% in-hospital mortality rate is likely to leave you with some long-term consequences.  (Stroke, heart attack, …)

But under no circumstances should this research be interpreted as showing that 20% of typical adult COVID-19 cases result in some debilitating form of long COVID. 

This is so misleading, and being so commonly mis-interpreted, that I think the CDC ought to be burdened to issue a clarification.  Or, at the very least, require the researchers to go back into their database and calculate the percent of their sample of cases that was hospitalized with COVID.  If, as I absolutely expect, that turns out to be almost all of their cases, then they absolutely need to issue a clarification.

Otherwise, I’m just going to chalk this one up to a long stream of CDC fails in this area.  They screwed up the initial set of COVID-19 tests.  They refused to acknowledge that COVID-19 was spread via aerosol (airborne) spread.  They incorrectly assured Americans that we didn’t need to wear masks.  Only after their arms were twisted (by the head of the Chinese CDC, no less) did the CDC grudgingly encourage Americans to wear inferior cloth masks.  They were dead wrong about the importance of fomite transmission, leading to the entire U.S. going through what amounted to two years of hygiene theater.  And they still have never acknowledge the benefit of wearing N95 respirators, despite a completely normal and adequate supply of those in the U.S. now.

To which we can now add this study, uniformly misinterpreted in the popular press, as the coda on that masterful performance.

Otherwise — aside from that one tiny little issue — this is a pretty good study.  They flagged the presence of plausibly COVID-related diagnoses in the medical records.  They tossed everybody who had one of those as a pre-existing condition.  They only looked at individuals who had some use of the health care system in both of the two years of the study.  And so on.  Everything else was more-or-less up to snuff, as these observational data studies go.  The only problem I see is that, by default, and never clearly stated, this is more-or-less a study of persons hospitalized for COVID.  But that only little issue has huge implications for how these results should be interpreted.

Separately, having done (plausibly) thousands of small health-care-claims-based analyses in my lifetime, I can even guess the technical issues underlying this omission.  Electronic medical records are verbose, consume vast amounts of storage, and are just-plain-hard to process when you are talking about tens of millions of cases.  It’s a good bet that the researchers themselves did not have access to the raw data, but instead were given a “patient-level abstract” of the information.  Further, under typical HIPAA-driven data privacy rules, that abstract contained only the minimum information required to do the study.  The researchers probably had to justify every data element that they requested.  That would be the patient ID, maybe some demographics, and then a string of dates and diagnosis codes.  But not the site-of-service information.  In other words, I’d bet good money that the researchers performing this study are literally unable to tell what fraction of their COVID-19 diagnoses come from hospital inpatient stays.

One final extras-for-experts, or why this would require a little bit of care to untangle.  Medical record coding rules require that follow-up visits must be coded with the reason for the original health-care visit.  Here, a person who is seen in the hospital outpatient department, a month after discharge for a COVID-19 hospitalization, will have COVID-19 coded on that outpatient followup visit claim, and so will appear to have had an outpatient visit for COVID-19.  The only quick-and-correct way to estimate the fraction of non-inpatient COVID-19 cases in such as situation is to flag anyone with any mention of COVID, then subtract out all persons with any mention of COVID-19 on any inpatient claim.

Addendum:

I am (or was) an old-school health services researcher.  I worked on large health care claims (bills) databases.  The first thing I want to know about any data source is where it comes from.  What proportion of the claims are from hospital inpatient stays, from emergency room visits, from physician offices, from independent laboratories, and so on.

That’s the only way to get a handle on what you’re looking at.   In the traditional Medicare program, for example, there are about 20 physician claims for every hospital inpatient claim.  If you only look at the hospital inpatient claims, you get a vastly skewed notion of the state of health of the average Medicare beneficiary.  And the closer you get to that inpatient-centric view, the more skewed your estimates will be.

Near as I can tell, nobody involved in this database or this study seems to have bothered to ask that basic question:  what mix of health care encounters does this information come from?   Or, possibly, they may not be able to tell from their data.  So I have had to make some inferences.

Based on the counts, my guess is that this analysis is based primarily on hospital admissions.  But the real point, coming from someone with extensive experience in claims-based (bill-based) health services research, the reader should not be required to guess about such a fundamentally important question.

---

Other sources of confusion, or, what do inquiring minds want to know about long COVID?

Ultimately, I have a really simple question that I’d like to have answered.  I’m old, fat, vaccinated, and-double-boostered.  If I manage to catch the current prevalent strain of COVID-19 (BA.2.12.1), what are the odds that I’m going to end up with some debilitating long-term condition as a result of a COVID-19 infection now?

This is the knowledge gap that I’m trying to fill.  I’ve already done some round-numbers estimates for likelihood of hospitalization or death, in several prior posts.  So now I’m asking for the odds that it won’t kill me, it’ll just cripple me some.

And it turns out that, aside from the problem identified in the first section of this analysis, this research (and most of the rest that I have read) provides absolutely no guidance in answering this question.

First, near as I can tell, most of the “long COVID” cases come from self-reports of vague, non-life-threating conditions.  It’s hard to say, because studies like this one don’t even bother to show the prevalence of the various conditions that they are counting.  But elsewhere, in other long-COVID studies, you can at least find qualitative statements like the following (this is from a different long-COVID analysis):

Symptoms most commonly associated with long COVID include fatigue, headache, dyspnea, hoarse voice and myalgia

Fatigue, headache, muscle pain? Not to make too light of this, but that sounds like a typical day for me.  The point being that the serious or life-threatening conditions — heart conduction disorders, pnemoembolism, and so on — those appear to be quite rare.  Those extremely serious conditions are not the main drivers of the count of long COVID cases.

This is almost certainly why the same methodology that showed “long COVID” in 40+% of cases showed “long flu” in 30%.  As reported here:

The researchers found evidence that Covid-19 patients are more likely to suffer from long-term symptoms than flu patients, with around 42% and 30% respectively reporting at least one symptom three-to-six months after infection.

Source:  Forbes, reporting on the underlying research, emphasis mine.

There’s no way that a third of U.S. adult flu cases have resulted in long-term debilitating illness each year.  I can only conclude that much of what gets counted as symptoms of long COVID is conditions that are fairly prevalent in general U.S. adult population.

Second, as far as I can tell, all of the analysis is based on prior strains of COVID.  This CDC analysis, for example, includes loss of taste and smell as a COVID-related condition.  But, by and large, that no longer happens with the current strain BA.2.12.1, and almost never within the vaccinated population getting the current strain.

Third, little of the analysis distinguishes vaccinated from unvaccinated, and none of it distinguishes boosterered from merely vaccinated.  So, to the extent that the vaccination process offers some protection from the dangerous long-term complications of COVID, that’s not reflected in any research.

Fourth, little-to-none of it distinguishes short-term or immediately-post-infection complication, from presumed permanent organ damage.  My understanding of much of the talk about (e.g.) heart damage in young athletes post-COVID turned out to be temporary conditions that had largely disappeared within six months.

---

So what’s an old fat guy supposed to do?

I could go on, but at the end of the day, I just toss up my hands.  The fact is, if my worry is that I’ll get permanently disabled from a COVID-19 infection now, there is no information out there to allow me to assess my odds.

So, I have to take a guess.  And my guess is that, vaccinated-and-boostered, with BA.2.12.1, the likelihood that if I get infected with COVID-19 now, I’m going to end up with serious, permanent damage, is really, really small.  How small, I can’t say.  But I’m willing to say that the cases you read about in the newspapers are the true outliers.  I’m betting that most of them date to Delta and earlier variants, in the pre-vaccine era.  Many but not all are post-hospitalization outcomes.  And so on.

If you read this blog, you know I’m not exactly the most cheerful and upbeat person in the world.  But in terms of long COVID — despite the fear-mongering of the popular press — I’m just not going to worry about it.  I’m going to continue to take reasonable precautions against infection when I think that’s warranted.  (Because, hey, I already own a more-than-lifetime supply of N95s).  I’m not going to avoid activities for fear of long COVID. 

 

 

A trend is a trend until it ceases to be a trend.   The U.S. remains at 33 new cases per 100K population, essentially unchanged from seven days ago.  This is starting to look like a typical turning point in one of these COVID-19 waves.  Case rates are rising in some areas and falling in others, and the net (average) just happens to work out to be flat.  We have not yet reached the point where new cases are falling uniformly across the U.S., but it looks like we’re headed there.

Continue reading Post #1523: Ending the COVID-19 data week flat, still 33 new cases per 100K

 

The U.S. remains at 33 new COVID-19 cases per 100K population, up about 5 percent over the past seven days.  By eye, there’s now a pronounced “flat spot” at the end of the U.S. new-cases curve.

Continue reading Post #1522: COVID-19 trend to 5/26/2022, still 33/100K

 

The U.S. is back up to 33 new COVID-19 cases per 100K population per day, up 8% over the past seven days.  It now looks as if the Northeast region — the leader for the Omicron-II wave — has peaked.  That suggests that the U.S. as a whole should not be far behind.  The current, stable rate of new cases in the U.S. is the result of continued increases in most of the country being offset by reductions in the Northeast and Midwest regions.

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 5/26/2022, 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

Here’s the Northeast by state, where the peak continues to shape up nicely.  As they were the leaders on the up-slope of this wave, they should be the bellwethers for the peak.

---

Extreme aerosol emissions from intense aerobic exercise

 

 

One of the earliest U.S. COVID-19 super-spreader events occurred at a church choir practice in Washington State.  More than 50 persons (of the roughly 60 in attendance) were infected, and two died.

The terrible thing about this event is that they did everything “right” according to then-current CDC guidance.  But this was still the period during which the CDC was in denial about aerosol (airborne) transmission of COVID-19.   At that time, the CDC was still telling people that COVID spread via large droplets that rapidly fell out of the air.  Accordingly, the official party line was that if people stayed 6′ apart, all would be well and there would be no transmission of COVID-19.  No need for masks, ventilation, or other measures.  If you read the CDC’s official write-up about that event, they are still in denial, and per the existing, absolutely implausible writeup, that an entire roomful of persons was sickened by one index case because they failed to stay 6′ apart.

Putting the CDC’s past intransigence behind us, it’s now well-established that COVID-19 spreads via aerosols (tiny droplets that can float through the air).  We know that certain people (“superemitters”) and certain activities greatly increase emission of those aerosol droplets.  And, in particular, we know that singing produces as much aerosol emissions as coughing.  This is why some countries (e.g., Germany) and many U.S. mainstream churches banned singing in church for the duration of the emergency phase of the pandemic (see e.g. Post #678, or search this site for “singing”).

In hindsight, then, it’s no mystery why this choral practice created a superspreader event, as did many similar situations around the world.  It as if nearly every first-world country had at least one major church-based super-spreader event, e.g. Post #679).

You can extend that to any situation with crowds and loud talking.  Plausibly, this is why going to a bar seems to have been the most dangerous activity possible for spread of COVID-19, and why bars were always the first sites to be shut down when COVID-19 restrictions were in place.  Followed by indoor dining.  As it turns out, sipping a drink and talking loudly is just about the worst thing you can do in terms of aerosol emissions (Post #723).

If talking loudly increases aerosol emissions (Post #585), and singing does the same, it should come as no great shock that breathing heavily likely increases aerosol emissions as well.  This is probably why many lists rated “working out at a gym” as a relatively high-risk activity during the pandemic (Post #811).

Based on reporting in yesterday’s NY Times, research now shows definitively just how much intense aerobic exercise increases aerosol emissions.  This research, published in the Proceedings of the National Academies of Science,found the aerosol emissions increased well over 100-fold at maximal exercise rate, compared to emissions at rest.  My recollection is that this is modestly higher than was found for singing or coughing.  Thus, I believe that working out at your maximum aerobic capacity — breathing as hard as you can breathe — increases your aerosol emissions more than any other activity measured so far.  The upshot is that high-intensity indoor group exercise is likely among the most dangerous activities for spread of COVID-19. 

The research explicitly notes that emissions during intense exercise are far higher than emissions during loud speech.  For a given density (persons per square yard), and given level of ventilation, from the standpoint of COVID-19 transmission, you’re safer in a bar than in an aerobics class.

The article shows that the biggest increase occurs when you push really hard.  Moderate aerobic exercise pushes up aerosols somewhat.  But the big increase occurs when you are pushing yourself as hard as you can.  More-or-less, emissions are a function of how hard you breathe.  And, interestingly, trained athletes — with higher aerobic capacity — are capable of emitting more aerosols.

Source:  Adapted from “Aerosol particle emission increases exponentially above moderate exercise intensity resulting in superemission during maximal exercise”, Benedikt Mutsch, Marie Heiber , Felix Grätz , Rainer Hain et al, PNAS May 23, 2022m https://doi.org/10.1073/pnas.2202521119

There’s a clear policy implication here for Virginia and likely many other states.  When mask mandates were imposed for indoor public spaces, Virginia made an exception for individuals using cardio equipment in a gym.  In the gym my wife and I (used to, and maybe will again) use, you had to wear a mask walking around, using the weights, and so on.  But you did not have to wear one while using the cardiovascular equipment or otherwise engaging in intense aerobic exercise.  (Though, to be clear, I continued to wear one in that situation for my own protection, because breathing deeply is also the best way to inhale aerosolized virus directly to the locations where it can take hold most effectively.)

Based on this research, that’s entirely backwards.  It’s the people using the cardio equipment that pose the greatest threat to others, and for whom mandatory masking would generate the greatest public health benefit per person.

Given that we’ve all put this behind us, I don’t expect to see any changes in policy.  But if — as many expect — we have a winter COVID-19 wave, this research provides a reason for the cautious among you to skip the indoor aerobics when case counts are up.  A room full of people breathing heavily from intense cardio exercise is probably not a room you want to be in during an airborne pandemic.