Post #1276: Final COVID-19 trend for the week.

Posted on October 2, 2021


No change.  Alaska seems to have peaked and is heading downward.  Most states continue to see reductions in new cases.  The now 34% below the 9/1/2021 peak, and new cases are falling at about 11 percent per week.

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 10/2/2021, from”  The NY Times U.S. tracking page may be found at


But I think it’s worth noting that the U.S. more-or-less splits into warm-weather and cold-weather states at this point.  Circled in blue above, you can see that the Northeast, Midwest, and Mountain states have new case rates that are essentially unchanged over the past month.  Their average new-case rate are about where they were at the peak of the U.S. Delta wave on 9/1/2021.

Another way to see that is to split the states into those where cases continued to grow last week, and all others.  The dozen states with case increases are concentrated along the Canadian border, and in the U.S. Mountain states.  (With the sole exception of Maryland).

Map courtesy of

Split this way, the falling U.S. total case number is the result of steep declines in the South Atlantic and South Central states, tempered by some increases in the far north and mountain states.  That's starting to look like the end of the summer Delta wave, and the start of the winter wave.  We're seeing declines, this year, when we saw increases at this time, last year, because, Right now, the end of the summer Delta wave is more than enough to offset any increases in northern tier and Mountain region states.

What I'm fumbling toward saying is that I think the winter wave has started.  It's just being masked, at present, by the delayed end of the U.S. summer wave.  If that's true, it ought to become increasingly apparent next week.

To be completely clear, so far, there's no obvious sign that we must have a winter wave.  As of yet, the national curve continues downward.  And if it fails to materialize, I'll be the first one to note it.

But if you look at the major sources of cases declines, they ought to be played out in a week or two.

For example, Florida is the third most populous state in the U.S. The sharp decline in cases there contributes strongly to the overall negative trend for the U.S. as a whole.   But Florida only has about another week to go on the downside before it crosses last year's curve.  And maybe another two weeks before it matches this year's lowest case rate.  Somewhere in that period, presumably, new case rates in Florida will stop falling, and Florida will cease to contribute to the downward trend in U.S. cases.

Likewise, Texas is the second most populous state.  The current rapid decline in new cases will take it to this year's lows within about three weeks.  At that point, as with Florida, its large contribution to the current U.S. average decline in cases should cease.

At this point, I guess I've beaten this into the ground.  Last year's winter wave started with gradually rising cases in mid-September.  This year, cases were still falling in mid-September.  But if you look closely at it, that's because the summer Delta wave was later.  The tail-end of the Delta wave results in declining cases at the moment.  But once the downward trends in those southern states end, my guess is that the winter wave will emerge, maybe a month later than last year.

This is just my way of saying that the improvement in the U.S. figure is not sufficient cause to slack off.  When you pull it apart, the "good news" states will be at their recent lows in a couple of weeks or so.  And at that point, any emerging winter trend should be apparent.

My herd immunity estimate, or why I think this year's winter wave will be the last one.

The pandemic continues to evolve.

In a just-prior post, I noted  that the Biden administration COVID-19 vaccine mandates will be too late to prevent a winter wave this year.  On the other hand, based on my experience, it seems like the booster shots are pretty popular, and the evidence shows that booster shots more-or-less bring your immunity back up to where it was when you were first vaccinated, or better.

For that, and a couple of other reasons, I'm revamping my "herd immunity" table. The key questions being

  1. How close is the U.S. to classic (i.e., mask-free, restriction-free) herd immunity? 
  2. How close will we be, if vaccination trends continue, and this year's winter wave is a repeat of last year's?

Let me emphasize that anything that blocks the transmission of disease contributes toward the overall goal of herd immunity.  To be clear, all of our COVID-19 hygiene -- distancing, masks, limits on social gatherings -- serves to reduce COVID-19 transmission, and so contributes toward that goal.

That said, when people say "herd immunity", they really mean herd immunity via immunity alone.  The sort of herd immunity that would let life return to normal, that is, without any obvious steps for COVID-19 hygiene.

And so, that's what the table below depicts.  The way I've constructed this, the goal is to reach 80%, on the bottom line.  That's the level at which, given the estimated infectiousness of the Delta variant (R-nought of 5.0), the pandemic would die out on its own, masks or no masks.

In this edition of my herd immunity estimate, I've made three changes.

First, I added in a category for those who have gotten a booster shot, and I count them in at the just-vaccinated immunity of 95%. This is a minor change, because the number of such individuals is small.

Second, I used the CDC's seroprevalence estimate as the most-nearly-correct estimate of the number of people who already have antibodies to COVID-19 due to prior infection.  That's based on a sample of blood draws for routine lab testing.  Those numbers can be found on the CDC COVID data tracker.  Currently, that estimate shows that the number of persons with some immunity from prior infection is about 1.9 times the number of persons actually diagnosed with COVID-19.  In the past, I was taking a guess on the number of infections that went unreported, and my best guess at that time was 3 times.

I will note one oddity with the CDC seroprevalence estimate.  Mainly, their estimate of persons with post-infection antibodies has not kept pace with their estimate of the total persons with known infections.

Source:  CDC COVID data tracker.  Seroprevalence is from national lab survey, actual infections are as of 7th and 21st of each month.

I don't think this is an error.  To me, this suggests that post-infection COVID-19 antibodies fade over time, and eventually fall below levels detectable in these tests.  I interpret this to say that those who were infected early on in the pandemic are slowly losing their immunity.  This is consistent, I think, with actual measurements over time (for the same set of individuals), and with immunity derived from vaccination.

Third, and most significantly, I changed my assumption about the overlap of immunization and infection.  I need some assumption there, otherwise I'll double-count individuals who had a prior infection and then went on to get vaccinated.  That was always a guess, and it has become an increasingly outdated guess, and it's high time I changed that guess.

Mostly, I've had to change that guess at the overlap of infected and vaccinated individuals because the "direction of causality" has changed.

In the past, when vaccinations just arrived on the scene, what I was talking about was behavior.  I  was asking whether people who had gotten infected (and knew it) would go on to get a vaccine.  (For example, as Senator Rand Paul famously refused to do.)  I made the (arbitrary) assumption that those with history of infection would get vaccinated at the same rate as everybody else.  Even then, that was questionable.

But now, the main driver isn't behavior, it's straight-up medicine.  The big story isn't about behavior -- whether those with prior infection will choose to be vaccinated.  Now, by contrast, the big story is that vaccines are very good at preventing infection.  It's that almost all the new infections are concentrated among the un-vaccinated population.  As illustrated at the top of this posting.

Here's the most recent week of data from Virginia, whom I claim has the cleanest cut at this particular statistic.  (This, along with fraction of population falling into these three categories, is the source for the graph at the top of this posting.)

Source:  Virginia Department of Health (text in red was added).

Obviously, that's rather an extreme difference, and is likely to be something of a temporary fluke.  I don't really expect the difference to be this large week after week.  But something in that ballpark has held true for the past several weeks.  At some point, you can't ignore the fact that there's an almost-complete separation between the vaccinated population and the infected population. 

As a result, in the current high-vaccination-rate world, my old assumption now must be grossly wrong.   So I'm taking a new stab at it, and guessing that, over the entire course of the pandemic, the overlap between the infected and vaccinated populations is just 25%.   That is, as far as I can tell at the moment, an un-testable assumption.

With that, here's how I think the U.S. looks today.  That 71% figure is net of all adjustments for (e.g.) efficacy of vaccines.  And so, as of today, we still fall short of the 80% level of population immunity that would, in theory, finally provide herd immunity against the Delta variant.

Source:  Base data are from CDC COVID tracker, accessed 10/2/2021.  Calculations are the author's.

Now let me assume that vaccination will continue at the current pace (about 0.3M new fully-vaccinated individuals per day), and that the total new infections in the 2021-22 winter wave will match last year's winter wave.  Where would that put us on March 1 2022?

Source:  Base data are from CDC COVID tracker, accessed 10/2/2021.  Calculations are the author's.

And that, in a nutshell, is why I think there will be a U.S. winter wave, and it will be the last wave.  Toss in a bit more for whatever COVID-19 hygiene people are still willing to do, and my best guess is that we're just shy of herd immunity now.  And we will be just over the herd immunity level next spring.

I have no special insight into the future.  I'm just doing the math as accurately as I can figure it.  And as of today, this is my best guess as to what the herd immunity math looks like.  And my conclusion is that we probably face one last wave of COVID-19.