Jersey Girl wrote: ↑
Tue May 12, 2020 6:03 pm
RI I spoke to my medical practice about the antibody test. They referred me to Lab Corp with my question. Went to the Lab Corp website. Here's one of the links where it discusses the test.
p.s. She said that in order to take the test you have to have had a confirmed case. That's not what they said yesterday and the Lab Corp website doesn't say that at all. I told her I was looking right at it and the only criteria stated was if you think you may have had the virus. This was the front desk I spoke to today.
I'll just have to send a message directly to my PCP and find out about it. You would think the front desk would be well prepared to answer questions on testing. But no.
So, LabCorps doesn’t make a serology test. They use different tests that they buy from the manufacturers. From their website:
LabCorp is providing serology testing based on tests from various manufacturers. Such tests have either received an emergency use authorization (EUA) from the U.S. Food and Drug Administration (FDA) or were released for use under FDA guidance, “Policy for Diagnostic Tests for Coronavirus Disease-2019 during the Public Health Emergency - Immediately in Effect Guidance for Clinical Laboratories, Commercial Manufacturers and Food and Drug Administration Staff” that was updated on May 4, 2020. Tests being performed under an EUA have not been FDA cleared or approved and LabCorp has completed independent validation of these tests. In addition, various manufacturers have submitted or will seek EUA for their tests.
So, I’ve got no idea. Also, I combined some info on antigen tests with antibody tests last night — I shouldn’t try and post complicated stuff late at night. The two kinds of antibody tests are ELIZA and LFA. ELIZA gives you a level of antibody present in your blood and an LFA just gives you a yes, no, or indeterminate.
Here are the tests that have had an EUA and the accuracy information on a which the EUA was based: https://www.fda.gov/medical-devices/eme ... erformance
The chart for each has a line labeled PPV 5%. The value on that line gives the answer to this question: If 5% of population has COVID antibodies, what is the chance that a positive result on my test would be accurate? Some have very high median values, but the 95% confidence range is very wide. Plus, I would be shocked if the prevalence rate is as high as 5% in your area. The lower the prevalence rate, the lower that PPV number will be. There’s a link to a calculator on that FDA page that lets you input specific numbers and gives you the odds that a positive test is accurate.
Just my opinion, but I would not change any of my behavior based on the test result. At a reasonable estimate of prevalence, I think a large part of that 95% confidence level would be at 50% or below — no better than a coin flip.
On the other hand, the prevalence among the population of people who had symptoms Is certainly higher. But we don’t how much. On the other other hand, if you weren’t in contact with a confirmed case and there wasn’t community spread in your locale, you’re likely part of a population with lower prevalence.
See how nice and clear and straightforward this is?
In Bayesian terms, you have to account for the likelihood that you actually were exposed based on the characteristics of the population you are a member of before you see your test result. When prevalence in the population you are a member of is very low and the test produces false positives, the odds that a positive result is a false positive can get very high. (Rate of false positives = 100% minus that PPV percentage. For example, I gave blood today. If the HIV test on that blood is positive, the odds that the result is a false positive approach 100%.
Also, this is what I understand and I’m happy to share it, but I wouldn’t rely on it without confirmation from Lemmie or Analytics or PG or any other person who knows stats better than I do.