Omicron's Testing Conundrum

 

On Sunday I noted that Arizona had not yet seen an explosion of cases due to the Omicron, like elsewhere. On Monday I thought there may have been detonation, with AzDHS reporting over 7000 new cases. 

But today they're reporting just 1,976 new cases. So it looks like once again we have unreliability due to delays in reporting cases.

Anyway, ASU reported over 70% of tests with s-gene dropouts (explainer here) yesterday. That means Omicron is definitely taking over in Arizona.

So what should you do about testing if you think you're infected? The current choices are antigen tests and PCR tests.

Antigen Tests

Antigen tests are the kind you can buy (if you're lucky) at the drug store and that you can administer yourself. The tests produce results in 15-30 minutes. That's great if you're trying to catch an Omicron infection because it spreads quickly and you want to stay away from other people if you have it.

The problem is that these tests are not very accurate. Worse, their sensitivity depends on the stage of disease you have.

A review of existing studies of antigen test accuracy shows a wide range of results. For people with symptomatic disease, the average sensitivity was 72% with an estimated a range* of 63.7% to 79%. But for people without symptoms, the average was a dismal 58.1% with an estimated range of 40.2% to 74.1%.  

The tests also have rather high false positives. The reviewers sum it up this way:

[I]n symptomatic people...between 1 in 10 and 1 in 6 positive results will be a false positive, and between 1 in 4 and 1 in 8 cases will be missed. ...[A]pplying the same tests in asymptomatic people...between 7 in 10 and 9 in 10 positive results will be false positives, and between 1 in 2 and 1 in 3 cases will be missed. 

In other words, for asymptomatic disease these tests may be no better than flipping a coin! They are also very likely to scare you with a false positive. In my book that makes them worse than useless. Even for symptomatic people there's a one in four chance that they will say you're free of infection when you're really not.

PCR Tests

Fine, you say, I'll just get a PCR test. These are done in labs based on samples taken from a nasal swab or saliva. The analysis process amplifies any genetic material from the virus contained in the sample, making it extremely sensitive. 

This means it can detect very small traces of the virus, so it is highly unlikely to produce false negatives—i.e., to say you are virus-free if you are really infected. One study showed that the sensitivity of PCR tests ranges from 82.6% to 98.9%.

The problem with PCR tests is that they're slow. The samples have to be transported to a lab, and then the amplification process takes time. ASU's lab reports an average 33.5 hours to return results as of this morning. That is a problem if you have Omicron because you could spread it while you're waiting to get your results.

The Conundrum

So, you can get fast but inaccurate results with the DIY antigen tests, or slow but accurate results with a Lab PCR test. Neither is acceptable.

There are some possible workarounds. They are pricey and may be unaffordable for a lot of people.

First, there are companies offering new at-home kits. They claim to use a "molecular" amplification process similar to PCR to achieve tests with equally high sensitivity. 

Detect charges $75 for a "hub" device and test kit, with additional kits available for $49. Unfortunately they are sold out.

Lucira sells a device good for one test. It costs $75 on their website and $89 on Amazon (though I'd avoid buying stuff from Amazon because they fund anti-vaxxers). Neither the Lucira site nor Amazon lists the product as sold out. I can also find no indication that you can get additional test kits once you have paid for the device.

Finally, if you do a Google search on "covid rapid test" you will find local labs promising you results in 20-90 minutes. I haven't investigated them, and I'll bet you pay for the speed. If you use such a lab, make sure they're using a PCR or "molecular" test, and not just an inaccurate antigen test. 

*This is a confidence interval of the mean, a statistical concept. The mean is calculated from a sample, and the interval is the range in which the true population mean (i.e. the mean if you could measure everything) lies with high confidence.