Thread by @dr_dmorgan

We set out to identify the sensitivity & specificity of common tests for #COVID19 along w/ @dan_diekema , @Anthony98947615 w/ @CDCgov support

A simple enough task, right?

I’ve seen tweets by @DrSidMukherjee @drjohnm @BenMazer @PaulSaxMD @BradSpellberg and others interested
1/n

Looking for comments/criticism

What are we missing? No industry adverts please!

Important papers?
2/n

The @US_FDA has a test comparison site that is incomprehensible to me… but @ASMicrobiology types tell me it reports on analytical sensitivity and LoD for tests
3/n

First, terms:
As a clinician, I want to know if a patient does or doesn’t have #covid19.
I’m less concerned by internal laboratory QA unless it translates
We wanted what has been called CLINICAL or diagnostic sensitivity/ specificity NOT analytical sen/sp.
4/n

Described in 1997:
Clinical “diagnostic sensitivity is defined by the percentage of persons who have the disorder of interest who have positive results on the assay”

And more recently early in COVID by @akesselheim

nejm.org/doi/full/10.10…

5/n

Another note, PCR has largely been the laboratory gold standard.

So, most reports are for sen/sp relative to PCR—not to a clinical gold standard.

We adjusted reported results to reflect:
e.g. PCR Sen 90% x Antigen Sen of 50% = overall antigen Sen of 45%

6/n

Finally, we are NOT talking about being infectious for #covid19.
That is another topic with even LESS data I hope to share soon.
Suffice to say, many people with #COVID19 disease are NOT infectious, especially after a week or 10 days, but some tests still +, esp PCR

7/n

PCR—the gold standard
Clinical sensitivity ~ 90% on day 4 of symptoms w/ common instruments
sciencedirect.com/science/articl…

dx.plos.org/10.1371/journa…

jcm.asm.org/content/58/8/e…

academic.oup.com/ofid/article-a…

8/n

-earlier reports of lower PCR sensitivity had atypical gold standard or were not reporting optimal sampling on day 4

-Sensitivity IS lower before/ after day 4 of symptoms per acpjournals.org/doi/10.7326/M2…

9/n

Reasons for PCR false-negatives (10% false – rate)

lack of virus in sampling site
inadequate sampling
lack of instrument optimization or variation between instruments

10/n

PCR Clinical Specificity ~99%
this was hard!
Scant data, authors suggest between 95%-99.5%

bmj.com/content/369/bm…
gov.uk/government/pub…
cdc.gov/csels/dls/locs…
m.koreaherald.com/view.php?ud=20…
jcm.asm.org/content/58/8/e…

11/n

Reasons for PCR false-positives (~1% false + rate)
-past infection with residual RNA
-differences in testing between instruments
-glitches in instrument reading of Ct/Cq values
-lack of laboratory optimization normally required by the FDA
-contamination
12/n

Next, the Point of Care NAAT tests
Limited information published and in some ways very similar to lab based PCR
Abbott IDNow seems less sensitivity/more specific than Cepheid Xpert Xpress/Roche Cobas

13/n

POC NAAT tests
—Abbott IDNow
Clinical Sensitivity 54% / Clinical specificity 97.5%
(final numbers after adjustment for comparison to PCR)

Misses high CT value PCR + which are often NON-infectious. But no data vs. cell culture
cochranelibrary.com/cdsr/doi/10.10…
amazing work @deeksj

14/n

other POC NAAT sources
jcm.asm.org/content/58/8/e…
jcm.asm.org/content/58/8/e…
ncbi.nlm.nih.gov/pmc/articles/P…
sciencedirect.com/science/articl…

15/n

We estimated
Cepheid Xpert Xpress /Roche

Overall
Clinical sensitivity 90%
Clinical Specificity 95%

jcm.asm.org/content/58/8/e…
pubmed.ncbi.nlm.nih.gov/32417674/

(and other refs for IDNow)

16/n

Last, but not least...

Antigen tests! The great promise.

I would say useful but not great.

Overall
Clinical sensitivity 45%
Clinical specificity 97%

cochranelibrary.com/cdsr/doi/10.10…
sciencedirect.com/science/articl…
sciencedirect.com/science/articl…

17/n

False +s with Antigen tests are real
cleveland.com/coronavirus/20…

dpbh.nv.gov/uploadedFiles/…

ahcancal.org/Data-and-Resea…
18/n

Antigen tests better detect live virus but VERY limited data vs. cell culture @michaelmina_lab

And newer antigen tests will likely be better (but need clinical data, not just lab comparisons)
19/n

So, in conclusion, a group of @IDSAInfo docs, epidemiologists and @ASMicrobiology estimated the following CLINICAL Sensitivity & specificity for #COVID19 diagnostic tests

--feedback welcome