Thread by @michaelmina_lab

So much focus on sensitivity of the test. As though all tests must match up to PCR

But could focusing so much on achieving the greatest molecular sensitivity for asymptomatic testing *when testing is infrequent* be doing more harm than good?

I’ll explain

1/

The molecular tools we use like PCR for #COVID19 tests detect the virus genome.

Like detecting DNA in a piece of hair, detecting virus RNA tells little about whether the virus is still active...

2/

When the virus is growing and someone is transmitting virus, it grows to billions of viruses. So it’s easy to detect when virus load is high using antigen tests (look for the virus proteins themselves) or PCR (look for the RNA) ...

3/

But after the virus is cleared by the immune system, all of those viruses leave little trails of RNA behind. The RNA gets stuck in vesicles one the cells and it can sit there, in the nose or mouth, for weeks or months at very low levels...

4/

So what does this have to do with a super sensitive PCR test being *too sensitive* that it can potentially cause more harm than good...?

5/

The PCR test can continue picking up that leftover RNA the whole time it is there - for weeks or months AFTER the viral infection has been essentially cleared.

So ultimately, the majority of time spent in the PCR positive state is after infection, not transmissible

6/

So this can be harmful if testing is very infrequent like it is in the US (most people get tested only once, if ever - but even if testing every few weeks)

The idea of the super sensitive PCR test is to detect people when at low viral load at the beginning of infxn

7/

But if screening tests are only every few weeks, the chances you detect someone in the very short window of time between turning PCR positive and having 10x higher viral loads (the virus grows fast and exponentially inside the body) is very slim. This window is hours.

8/

On the other hand, with a very sensitive test, people might stay positive for many weeks or even months bc it continues to detect the leftover RNA from the past infection.

But this can be bad - it leads to unnecessary quarantines of people already past their infection

9/

if doing low frequency screening of asymptomatics with PCR, the majority of people found to be positive will be detected only after their infection.

So in the US we are unnecessarily quarantining millions of ppl, assuming they are infected when they’re already recovered

10/

And we are contact tracing all of these people and quarantining and testing their contacts looking only at the two days prior to their swab for the test - meanwhile they were likely infectious 1-5 weeks earlier! So wasting resources tracing the wrong people.

11/

To sum, we are using extremely sensitive PCR tests for screening. They are expensive and limited so we do them very infrequently.

What we gain is ability to find someone a few hours earlier - only IF by chance the swab is taken in the few hours where it makes a difference

12/

What we lose is ability to more frequent tests and thus do not catch people when it matters.

Further, we mistakenly end up quarantining millions of ppl for 10d bc we erroneously assume any PCR + test was collected at the beginning of an infection. When most are after.

13/

So maybe the most sensitive PCR test is not only costly for screening, but maybe even more damaging than a cheaper more frequent lower sensitivity test.

The apparent missed cases maybe aren’t false negatives. Maybe the (+) on the PCR are False (+) for actionable results

14/

I think it’s time we take a hard look at the types of tests we are using, what they do and do not tell us about infection, and figure out how to use them accordingly, instead of just assuming that more sensitive is better.

Maybe less sensitive is better for some things.

15/15

To finish, high sensitivity PCR is great if youre a doctor and need to know what is wrong with your patient. Like a detective, you want all shreds of (RNA) evidence

But it’s not always better & sometimes a lower sensitivity test is more accurate at detecting active infection.

We published on some of these issues as well as how the viral load data from the PCR could be better used clinically earlier this year:

“To Interpret the SARS-CoV-2 year, consider the Cycle Threshold Value”
academic.oup.com/cid/advance-ar…