Really worrying information on false positives and negatives: "if an LFIA is applied to a population with a...prevalence of 10%, for every 1000 people tested, 31...will be incorrectly told they are immune, and 34...will be incorrectly told that they were never infected."
-
-
Show this thread
-
The review also found that the evidence used to generate sensitivity/specificity was woefully inadequate and at high risk of biaspic.twitter.com/ldugtRdp0Y
Show this thread -
This is really bad for serological surveys (many of which have used these point-of-care tests) Raises the question about infection-fatality estimates, particularly those based on populations with low rates of infection
Show this thread -
-
The problem for many serosurveys is that they used ELISA tests for IgG. According to this study, the pooled estimate for specificity of these tests is 98.9% and sensitivity of 80.6%
Show this thread -
Now in a population with a 10% prevalence, that's pretty bad. In 1000 people, you miss 20 true positives and get 10 false positives, so you underestimate prevalence substantially
Show this thread -
Conversely, if you have 1% prevalence, you miss 2 true positives but get 11 false ones, so you overestimate the prevalence of COVID infections enormously This is a big problem!
Show this thread -
Now, some serosurveys have corrected for issues like this, but these new results suggest that commercial tests are less reliable and that their figures are more likely to be wrong That's a worry
Show this thread -
Also, apologies, the initial tweet was incorrect - this SR/MA was for ALL serology tests for COVID-19 not just POC ones
Show this thread
End of conversation
New conversation -
-
-
Great short thread
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.