It's been pointed out to me that most studies comparing antigen with PCR tests either consider these as independent tests carried out over the whole sample (and report concordance/discordance that way, e.g. relative to presence or lack of symptoms) or report the probability of getting a certain antigen test result conditioned on the PCR one, i.e. consider the PCR the "true thing".

Are there any studies that estimate the "real world" (i.e. non lab) probability of getting a negative PCR result conditioned on getting a positive antigen test result first?

Note: I'm not asking this to diss PCR or something like that. Simply asking because using antigen tests to screen ahead of PCR appears to be even government-mandated practice in some countries. For example the EU recommends:

Use rapid antigen tests for population-wide screening in epidemiological situations or areas where the proportion of test positivity is high or very high;

Ensure that strategies are put in place that clarify when confirmatory testing by RT-PCR or a second rapid antigen test is required;

So what kind of concordance/discordance results have been observed (and thus are expected to be "normal") in that kind of (conditioned) scenario, when PCR is only done on the subset of positive antigen tests samples/subjects?

  • Doesn't the answer strongly depend on prevalence?
    – Bryan Krause
    Mar 22, 2021 at 13:33
  • 1
    Sure, I saw the light :-D I was curious how much it does... Mar 22, 2021 at 23:04


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.