The Centers for Disease Control (CDC) have issued new guidance on serological testing and it’s pretty disappointing. CNN reports:
(CNN)Antibody tests used to determine if people have been infected in the past with Covid-19 might be wrong up to half the time, the US Centers for Disease Control and Prevention said in new guidance posted on its website.
Antibody tests, often called serologic tests, look for evidence of an immune response to infection. “Antibodies in some persons can be detected within the first week of illness onset,” the CDC says.
They are not accurate enough to use to make important policy decisions, the CDC said.
“Serologic test results should not be used to make decisions about grouping persons residing in or being admitted to congregate settings, such as schools, dormitories, or correctional facilities,” the CDC says.
“Serologic test results should not be used to make decisions about returning persons to the workplace.”
I guess that in the final analysis what that tells us is that the point I have been making all along is the one that will guide us. Reopening will depend more on recalibrating how risk is assessed rather than on testing.
BTW, take the observation about retesting with a grain of salt. When a test provides incorrect results half the time, a retest won’t necessarily give you a correct result. You should be able to convince yourself of that by flipping a coin.
I hope this didnt surprise anyone. One of the reasons it takes a while to bring new drugs or new tests to market is to make sure they work. When you skip or shorten those steps then you get a bunch of tests that work poorly. So you are correct that we will just have to guess rather than test for incidence in the community.
Steve
Is the CDC still not working with the FDA?
While some serological tests are bad; some are very very good.
The University of Washington virology department did a verification study on the test by Abbot.
https://www.seattletimes.com/seattle-news/patients-struggle-to-get-uws-touted-test-amid-misinformation-suspicion-about-reliability-of-coronavirus-antibody-tests/
UW found it had a specificity of 99.6% and a sensitivity of 100%.
At 99.6%, only places with very small outbreaks, like Hawaii or Alaska would there be an issue with false positives. Using a handy calculator, a prevalence rate of 1%; false positives is only 1% of all negative tests.
UW is a first tier university hospital system — is no one at the CDC talking to them? Brings to mind the FDA trying to shut down their unauthorized surveillance of coronavirus back in February.
If you are not part of Abbot’s supply chain can be hard to get into it now. There are also the issues proprietary testing equipment and compatibility with existing hospital systems. Dont know their capacity either. (Just got off a phone meeting where we talked this over. It looks like the only way to guarantee we have enough testing ability to be useful for antibody testing is to work exclusively with a single company. We are setting up that alliance. Meanwhile we offer the less reliable tests to those who insist on having them. Mostly executives and professionals.)
Steve
I went over the article and CNN did a terrible job.
The CDC was not saying the general test on the market has a 50/50 of producing a false positive. The CDC was giving an *example* of a test with low specificity of 90%, prevalence of 5%, produces false positives 50% of the time.
Checking the list of approved FDA tests; only 2 of 11 had a specificity where the lower bound of the 95% confidence interval got to 0.9 — those were Cellex and Chembio.
https://www.fda.gov/medical-devices/emergency-situations-medical-devices/eua-authorized-serology-test-performance
The CDC also says you can repeat the antibody tests for positives to drastically reduce the incidence of false positives.
CNN misses the key reason why the CDC is not recommending them for policy decisions. It is this point in the guidance.
“Although the presence of anti-SARS-CoV-2 antibodies when detected using a testing algorithm with high positive predictive value for the context of use likely indicates at least some degree of immunity, until the durability and duration of immunity is established, it cannot be assumed that individuals with truly positive antibody test results are protected from future infection.”
Maybe I’m missing something from that last quotation but doesn’t it also say that a vaccine cannot work (at least not work in the sense that you assume you are protected from future infection)? Isn’t that how a vaccine works? By getting the body to produce its own antibodies without actually contracting the disease?
Yes; that quote casts skepticism on the effectiveness of potential vaccines.
I believe the “until durability and duration of immunity is established” is in reaction to reports of recovered COVID patients testing negative then positive on PCR tests for weeks after illness; and the speculation that SARS-COV2 will mutate like the flu, rendering those with antibodies vulnerable to infection.
The advice is over-cautious. Because there is overwhelming evidence that recovered COVID patients are shedding dead virus — so antibodies did kill the virus. And there is no known mutation of SARS-COV2 that makes those antibodies useless.
‘Because there is overwhelming evidence that recovered COVID patients are shedding dead virus — so antibodies did kill the virus. And there is no known mutation of SARS-COV2 that makes those antibodies useless.’
If Kung Flu cannot mutate to the point of being able to reinfect people, that would be great news. However what about the conflicting reports of people getting reinfected by virulent strains of the beast? Some say it’s possible, others say the test are just false positives.
South Korea’s CDC published a study proving people were not getting reinfected; but just shedding dead virus.
https://www.npr.org/sections/goatsandsoda/2020/05/22/861061727/south-korean-study-shows-no-evidence-recovered-covid-patients-can-infect-others
Don’t credit the South Korean CDC too much, they were the ones who raised the possibility of reinfection in the first place (publishing before throughly investigating what they saw, sigh).