Post #1017: Trends to 2/17/2021 and a bit more on antigen tests and probable cases

Posted on February 17, 2021

No news regarding the U.S. trend.  New U.S. COVID-19 cases are now down 69% from the peak.  The most recent rate of decline (2/14/2021-2/17/2021) is somewhat faster than earlier periods.  And there’s still no explanation for this decline other than “seasonality”, which is no explanation at all.

The only noteworthy thing I’m seeing is that new cases appear to be rising in North Dakota.  That’s worthy of some attention because ND was the bellwether for the U.S. third wave.   But while ND moved in synch with the rest of the Midwest in the past (large red circle), for this latest change ND seems to be moving alone (small red circle).

I can’t even be sure this is real, as opposed to some sort of change in data reporting.  So I’m not going to make much of it yet.

One oddity is that the majority of new cases found in ND, for the past few days,  have been “probable” cases identified via antigen tests.  (As opposed to “confirmed” cases identified via PCR (DNA) testing.)  I’m not sure what the significance of that is, either.


Confirmed versus probable cases, PCR (DNA) versus antigen tests.

The more I look at the issue of confirmed versus probable cases (Post 1016), and PCR versus antigen tests, the muddier the situation seems.  There are issues in data reporting, and there are far subtler issues in testing behavior.

Data reporting issues

The key issue is whether or not the results of positive antigen tests are included in state totals.  Per reporting by CNN, only five states currently exclude those tests in their count of new cases (CA, CO, MD, MO, NV)

This matters, because antigen tests have become an increasing share of all COVID-19 testing.  For example, in Virginia, if you failed to count those cases, you’d be missing about 30% of all new cases.

In terms of what the states report, between the states and the U.S. CDC, it now appears that every conceivable variation of confirmed and probable, PCR and antigen is being used to report total cases.

California only reports confirmed cases, and does not include antigen tests in confirmed cases.  And the CDC shows blanks or zeros for all counts of probable cases for California.

Upshot:  California totals exclude probable cases.  They exclude individuals identified via antigen test.  National databases show no counts of positive antigen tests for California.

Utah, like California, only report confirmed cases.  The CDC shows blanks or zeros for all counts of probable cases for Utah.  But Utah includes antigen tests as confirmed cases, per this footnote on the Utah COVID-19 dashboard:

Case Definitions: UDOH assigns case status following the national case definition, with the exception of considering antigen positive tests as evidence of a confirmed case. A confirmed case is any person with a positive SARS-CoV2 PCR or antigen test.

Source:  Utah Department of Health

Upshot:  Utah totals exclude probable cases.  But they include cases identified via antigen testing.  You cannot separately identify the antigen test count in Utah data.

Texas separate reports confirmed and probable cases, but no combined total.  It includes antigen tests in probable cases.  The CDC shows the combined total of confirmed and probable for the Texas count, does not break that out for the total, but does break that out for the count of daily new cases.

Upshot:  Using the combined total, Texas effectively reports both confirmed and probable cases.  The Texas total includes antigen tests.  With effort, you can identify an antigen test count for new COVID cases, but there is no official cumulative total.  (Presumably, that means that they only started counting those part-way through the pandemic.)

And so, in the end, you have to rely on the CNN reporting cited above.  The only large U.S. state that does not include positives identified via antigen tests is California.  And so, presumably, this will have only a slight impact on national trends.

Behavioral issues

Now we get to the unquantifiable question:  Does (or how does) the availability of antigen tests (“rapid tests”) affect the total volume of COVID-19 testing.  Do these tests merely substitute for PCR tests, leaving the total unaffected?  Do they add on to PCR tests?  Is there an overlap among people tested first with an antigen test, and then with a PCR test?

Reading through the CDC guidance, antigen tests are cheaper, faster, and can typically be read in the physician’s office.  But they are less sensitive than PCR (DNA) tests, and so will result in more false negatives.  Near as I can tell, they are neither more nor less intrusive to the patient.  E.g, they may still require a nasopharyngeal swab.

My vague understanding is that one area where these tests are adding to total testing is in group settings.  If you wanted to screen an entire nursing home, for example, antigen tests allow you to do that quickly and cheaply, on site.

At the end of the day, given that we don’t really know what determines total testing volume in any case, the presence of antigen tests just adds another random factor.  Arguably, the increasing use of these tests may result in a greater test volume than would otherwise be the case, and so a greater count of COVID-19 positives than would otherwise be the case.   Given all the other ongoing changes — the fraction of population immune, the shift in the prevalence of different strains of COVID — I’d have to guess that any impact of this is simply lost in the noise.