I think it's become clear that rapid antigen testing for COVID19 proponents are simply unserious. They prop up a straw man: that lab medicine impractically worships the god of sensitivity. They don't wish to engage critics. He's a brief review. https://www.healthaffairs.org/do/10.1377/hblog20200909.430047/full/

A relatively one-sided article in the NY Times brought attention to the small number of people advocating for widespread, repeated antigen testing in asymptomatic people. Luckily, @KatherineJWu published an excellent follow up article. https://www.nytimes.com/2020/09/06/health/coronavirus-rapid-test.html

The NY Times article quotes thoughtful comments by Dr. Babady, Dr. Abbott, @alexandermcada5, @Linoj_S, @SBtotheDub, @uma_karma, and others. None of these critiques were addressed by this latest Health Affairs piece.

The authors continue to inflate CT values of resp. samples with viral load. Given the inherent variability of sample collection and the lack of validation as a quantitative test, this is a big leap. Then they stretch it further with the idea that CT = viral load = infectiousness.

What article do they cite to suggest antigen tests are good at picking up high viral load? An article titled "Low performance of rapid antigen detection test as frontline testing for COVID-19 diagnosis." Here's the data. Lots of overlap, even at higher viral loads.

Here's an issue with this study. They tested the same sample with PCR and antigen testing. This allows you to directly compare. But it doesn't mean it correlates to the "viral load" in the patient - it also reflects sample quality. Some "low VL" patients could be very infectious.

This study also found that antigen testing performed poorly on asymptomatic patients - the population of interest. 10/14 asymptomatic infections were missed by antigen testing.

The authors also ignore issues of specificity. They call antigen testing "highly specific" based on an EUA reporting results on only 102 SYMPTOMATIC patients (remember, they want to test asymptomatic).

Even taking the EUA's reported 98% specificity at face value, scaled up to testing every person many times, you are talking about MILLIONS of false positives. How many people will have their lives disrupted? How many people will lose faith in the test?

The results for this EUA are also for a carefully designed test meant to be performed by healthcare professionals. Will quality control be as good if billions of tests have to be quickly produced and then performed at home by regular people?

Antigen proponents ignore valid supply chain concerns. Supply chains are unable to provide enough testing reagents for established tests now. How will we scale up to BILLIONS of brand new tests? Not saying it's impossible, but you can't just pretend "the market" will do this.

Finally, we also can't ignore social, economic, and cultural concerns for a massive public health intervention like antigen proponents continue to do. Will people willingly take daily/weekly home tests? Will they perform them correctly? Will they ignore the results?

America has a terrible safety net. Sick people can lose their jobs and income and health insurance. Few can afford to miss even a single paycheck. Yet we expect people to consistently self-test, self-report, and self-quarantine - risking their own financial well being.

COVID-19, it will come as no surprise, has also become very political. Widespread testing is quite a government intrusion into people's lives. When this intrusion is coupled with a very flawed test, it presents huge issues of trust. Expect crazy conspiracy theories within a week.

So that sums up the critiques, all ignored by the Health Affairs piece.

(1) CT does not equal viral load does not equal infectiousness
(2) sens/spec isn't established in home testing on asymptomatic people
(3) social, economic, and cultural barriers to widespread home testing

Lab people and regulators, much maligned by antigen proponents, actually think about this stuff. When home tests are approved by the FDA, they often require large, practical studies showing that performance is adequate with the way people actually use the test.

All this isn't to say antigen testing doesn't have a place - it does. "Controlled" institutional environments could address many of these concerns. But that's quite different from testing every man, woman, and child in the country multiple times per week.

Perhaps a comparison is helpful. Let's look at some of the studies the FDA cites in their approval of at-home HIV testing. Notice how THOUSANDS of people were tested (known infections, high risk, low risk). Studies of real life performance were done. Other conditions tested.

Antigen proponents talk about how their unique perspective is that they are "modelers" so they can see the big picture. I know it's boring, but much of lab medicine is small picture. Do we have the data to show the test actually works in the real life testing environment?

I'd be eager to see a trial run of widespread antigen testing. Institutional settings are NOT a valid comparison, given all the issues discussed above. But pick a town and try out population antigen tests. I'm happy to have my opinion changed. Anything to fight the virus.

Update now that I’ve had coffee: maybe I wasn’t clear on the CT issue.

When we measure validated viral load tests, say in HIV or HCV, first, it’s on usually on blood — a consistent matrix. If you draw ten samples from the same patient, you’ll get similar material to test.

Respiratory sample collections are different. Repeated samples from the same patient on the same day could have log order different viral levels. But the patient has only one “real” viral load.

You can see this in the few studies we have doing repeat testing on patients. While there are trends in CT reflecting true viral load, daily fluctuations are still significant. That doesn’t reflect pure biology. There are pre-analytic and analytic factors.

Then you have the reliability of the test. True viral load tests are quantitatively validated. They are proven to get acceptably similar CT results over and over again. Current covid tests are QUALITATIVE — they get positive/negative correct but maybe not similar CT values.

All this is to say it’s confusing to talk about viral load = infectiousness. While it’s a reasonable guess that people shedding more virus are more infectious, the “viral load” being measured doesn’t necessarily reflect the actual patient viral load.

Then we can question even these reasonable assumptions. Take HIV again. We know viral load correlates to infectiousness. This took DECADES of empiric work. The best study showing undetectable VL is non-infectious only came out in the last couple years!

So bottom line is we are making a lot of back of the envelope empiric assumptions here. They aren’t crazy, they just don’t have high quality evidence. But to turn those guesses into an unprecedented national testing drive does not seem congruent to me. Start with smaller studies.

Oh, I didn’t really get to the cherry picked studies in the Health Affairs essay used to justify sweeping conclusions. Most of the studies cited in this essay were small, non-systematic, and done early in the pandemic. Here are the number of patients in each study.