CORONAVIRUSPublished 5 hours ago Coronavirus antibody testing finds Bay Area infections may be 85 times higher than reported: researchers

Fox News ^ | april 17, 2020 | Alexandria Hein

[absalom01]

I don’t know what tweet you’re talking about, as this is a summary from Fox News of the initial results of a study in the SF bay area. USC and LA county are underway with similar work this week.

All is preliminary and we should expect results to vary as additional data are collected.

I agree with the rest of the points that you raise. That and 5 bucks will get you a cup of coffee though!

[exDemMom]

I am well aware of the difference between IgM and IgG.

I figure an MD knows something about the different classes of antibodies, and the activation profiles for each type. But the average reader might not, so I added a short description of each type.

Yes, there are false negatives which have a different mechanism than false positives.

This paper (not peer-reviewed and published by the same publisher as the Stanford paper that everyone is going on about) discusses both false negatives and false positives. The authors suggest that the definitive diagnosis of Covid-19 should depend on two different assays: rtPCR to detect viral RNA, and serological testing to detect total antibody, IgG, and IgM. I found this statement of particular interest: "The specificity of the assays for Ab, IgM and IgG was determined as 99.1% (211/213), 98.6% (210/213) and 99.0% (195/197) by testing of samples collected from healthy individuals before the outbreak of SARS-CoV-2." (Last sentence in the Antibody measurement section.) That suggests that there is at least a 1.4% error in antibody studies--which is very close to the rate of Covid-19 seroconversion the authors of the Stanford paper claimed to have found in a (biased) sample of non Covid-19 cases. In other words, the finding that there was a 1.5% seroconversion rate, inexplicably "corrected" to 2.49%/2.75%/4.16% using different models, is well within the margin of error of the immunoassay.

While I certainly accept that there are presymptomatic cases, I am less convinced of asymptomatic cases, especially when the models and estimates propose a prevalence that makes Covid-19 more infectious than any known respiratory virus. Whenever someone makes such a hypothetical claim, you have to weigh it against the body of knowledge within that field to determine whether it is biologically feasible. Is it plausible for any virus, even an airborne virus (which coronaviruses are not), to have an R naught up in the 50 or higher range?

[Zhang Fei]

[While I certainly accept that there are presymptomatic cases, I am less convinced of asymptomatic cases, especially when the models and estimates propose a prevalence that makes Covid-19 more infectious than any known respiratory virus. Whenever someone makes such a hypothetical claim, you have to weigh it against the body of knowledge within that field to determine whether it is biologically feasible. Is it plausible for any virus, even an airborne virus (which coronaviruses are not), to have an R naught up in the 50 or higher range? ]

---

I don’t get the impression that they’re all singing from the same hymn book. Are the tests even the same? How do you get comparable numbers when the tests are different?

[exDemMom]

I don’t get the impression that they’re all singing from the same hymn book. Are the tests even the same? How do you get comparable numbers when the tests are different?

Ideally, different tests would give identical results, because they are all testing for the same thing (the presence of disease). For clinicians, the concordance of different tests adds certainty to the final diagnosis. In actual practice, all tests have some degree of error, in terms of false negatives and false positives. This would cause a small subset of the test results to disagree with test results using different methods. The raw numbers would be different, because a virus count is not the same as an antibody titer, but the overall result should be the same. In other words, a patient who has a measured virus titer should also develop measurable antibodies.

I want to point out that the interpretation of an antibody test administered to a symptomatic patient who has known risk factors is very different from the interpretation of an antibody test to a random sampling of the overall population. In the first case, the major concern would be false negatives; in the second case, false positives would skew the results.