Thread by @michaelmina_lab

Important (but not surprising) study

Rapid Antigen tests may be MORE, not less, accurate to detect VIABLE virus (i.e. transmissible virus) compared to the more sensitive PCR.

But how?

1/

medrxiv.org/content/10.110…

We continue to see the use of the word “accurate” but accuracy depends on the target of interest. If the goal is to detect any evidence of virus at all (i.e. RNA remnants), then PCR will be more sensitive, more specific and overall more accurate.

2/x

However if the target is to detect VIABLE and thus likely transmissible virus - then it is possible that PCR can “overcall” positive results when in fact the sample contains just RNA and no viable virus.

3/x

In this study the authors did a terrific job at characterizing some PCR tests against viral load and then assessed the sensitivity and specificity of the Antigen test.

What did they find?

4/x

Not surprising they found that PCR is always more sensitive. This makes sense! They also found that PCR is more specific when the target is simply any COVID19 RNA... this too makes sense

5/x

The new piece here is that they went through the trouble to stratify the RNA positive samples based on culture positivity (whether viable virus is detected).

Here, they found that the antigen test is more specific...

6/x

They found that when the test turned +ve it was more likely to contain viable virus than when the PCR turned positive.

It did miss a small number too, but it shows that in this setting when an antigen test is truly positive it likely means that you have transmissible virus

7/x

What didn’t this paper discuss...?

It didn’t discuss that this was in a contrived environment of high percent positive samples overall. So this doesn’t get at the actual specificity of the test overall

8/x

These tests can and do have real false positives - where they turn positive for no reason having to do with the virus. Perhaps a couple %.

For repeated population screening this can be bad.

But, there are solutions

9/x

A simple solution that I have been recommending (perhaps to deaf ears) regarding rapid antigen tests is the need to package them together with an orthogonal rapid test. A test that is similarly rapid but would not be likely to turn falsely positive for the same reason.

10/x

For every pack of 30 rapid tests, they come with a pack of five orthogonal confirmatory rapid tests. Both needed to be positive to require isolation for ten days. If discrepant, still mask, distance and test again 24h later

We discuss this a bit here

wsj.com/amp/articles/b…

The point is though that the paper shows that antigen tests can do a good job at picking up viable virus and not a good job picking up RNA without viable virus. As expected. This means that these tests can have a special role in public health

11/x

If they can be deployed broadly for frequent use and can come with a confirmatory test to drive false positives to below 1:1000, then they can become crucial to detect and filter out transmitting people before transmission persists onward

nejm.org/doi/full/10.10…

12/x

We saw the benefits of frequent testing on display at the White House all the way up until October when a case got through and spread to many others because they weren’t wearing masks or distancing. We’ve also seen innumerable schools stop outbreaks through frequent tests

13/x

Many schools do not have outbreaks bc they have used frequent testing to identify / stop transmission before it gets going.

Cheap rapid antigen tests could make similar programs available to the population at large, helping to get cases under control and open the economy.

But these tests should never serve as a substitute for masks and distancing. Such an idea would be unsafe and remarkably not well thought out.

Also. The preprint referenced throughout this thread 👆is from @JohnsHopkinsSPH @JohnsHopkinsMMI & @BDandCo