Post #731: EDITED: Two-Tier Testing is now “Pool Testing”; a simple solution to the problem of re-testing within a pool.

Posted on June 26, 2020

Source:  Home Depot.

Please read Post #605, from 4/6/2020.  Germany figured out how to test all of its hospital employees, efficiently, by testing them in batches.  Only when a batch tested positive would they test each employee in the batch.

Here’s my description, from that posting:

The Germans test their hospital employees in batches of ten.  They pool the swabs for 10 employees, test that, and then only when a batch is positive, they test the individual employees in that batch. Man, that’s smart.

From today’s Washington Post, here’s a radical shift in testing strategy that the US National Institutes of Health is just now thinking about:

The approach works this way: Samples from, say, 20 people are combined into a single pool. One coronavirus test is used on the entire pool. If the test comes back negative, researchers know they can move on to another pool of samples. If it comes back positive, only then would each individual be tested.

The US has finally discovered what I termed Teutonic Two-Tier Testing.  Only here the official term is Pool Testing.  (Which I think is an unfortunate choice, even if technically accurate.)

I will emphasize that the German approach was not a secret.  It’s something the German government did, and discussed.  Something that a random blogger such as myself could readily discover.  And write up math problems about.  It was clearly efficient.  It was widely reported within mainstream healthcare industry publications.  Including the a publication under the aegis of The Lancet, where they toss in a further efficiency factor by testing in three stages instead of two.

And this approach is widely used in other health care settings, such as screening donated blood.   So it’s hardly as if the idea of testing in two stages (pooled to screen out many units, then individual to find an infected person within a pool) is new to health care in general.

For the life of me, as the US government and various state governments yammered on about the lack of availability of tests, I could not figure out why we didn’t just copy the Germans.  And now, two-and-a-half months later, faced with a rapidly rising count of new COVID-19 cases, the National Institutes of Health and company are finally stumbling toward this approach.

I guess I need to finish up by pointing out the obvious:  It’s not the lack of physical tests that’s the barrier to screening, in most cases.  It’s the lack of cooperation.

In Germany, they use this approach for hospital employees.  There, presumably, they can be compelled to take a test, and take a test in this manner, as a condition of employment.

In the US, though, citizen cooperation is often the rate-limiting step.  In the case of the infected hairdressers in Missouri (Post #721), only 42 of the 140 potentially infected customers would agree to be tested.  The rest couldn’t be bothered to do it.  So it wasn’t a lack of physical tests that was the problem.  It was the lack of public cooperation.

Plausibly, this is now being brought up in the context of another shortage of physical test kits, brought about by this second wave of the US pandemic.  If so, then this is a great idea, and needs to be given all due consideration.  (The underlying test is a PCR test — that is, it greatly replicates any DNA in the sample.  To me, that nature of the test, and the success in Germany, suggest that there’s no physical barrier to combining multiple swabs in a single test).

But if the idea here is that we’re finally going to get mass screening testing going in the US, I’d say, guess again.  Heck, we can’t even get people to wear masks most places.  We can’t even get people who know they’ve been exposed to COVID-19 to take a test.

But if this were used to ramp up workplace testing of all sorts, then that would a) be completely feasible, and b) be a win-win.  Because, in a workplace, presumably you can require testing as a condition of employment.  I don’t see any barrier to using this for (e.g.) routine screening of meat packing plants.  (Except that — see Post #605 — if the infection rate is high enough, pool testing is less efficient than standard testing.)

I wish them luck with this.  I also wish they’d have just copied what the Germans were doing, months ago.  And I hope they use this concept as part of an overall workplace safety plan, something that I think we pretty clearly need.

Extras for Experts:  Here’s Scientific American’s take on it, courtesy of the Post article.  They seem to suggest that, with sufficiently low rates of positives on the tests, a more sophisticated way of pooling the test swabs can result in not having to do a second test to pin down the infected individuals.  As noted in my Teutonic Two-Tier Testing posting, my take on it is that any such algorithm will not be robust enough to be used in practice.  I would love to be proven wrong on that, but algorithms like that are generally a form of “balance puzzle”, and for those, an unexpectedly high rate of positives will disrupt the algorithm.

Edit:  I thing there’s an obvious, robust, and vastly simpler solution to the problem of having to re-test individuals:  Take two swabs in the first test.  Plausibly, you might even be able to do that in one pass, just using two swabs for the procedure, held together, rather than one.  But, if not, swab each patient twice.

Hold the second swab in reserve.  Then test those second swabs, as needed, if a batch turns out positive.  And, in fact, re-testing using a previously reserved sample is exactly what the Germans do, per this reference.

Either way, properly structured, there would be no need to call back individuals for re-testing.  A person only has to sit still and endure one session of being swabbed, even though the testing takes place in several rounds. Given low public compliance with any sort of request for testing, I think it makes vastly more sense to get what you need, while you have your hands on the patient, rather than hope that you can get the patient to come back.

However you slice it, we need to follow German’s lead on this one.  Right now, they use this approach to screen nursing home staffs.  Everybody who works at the nursing home is tested, and yet, if there are no positive cases, the number of tests required is small.  Seems like a pretty good idea to me.