Quote:
Originally Posted by golfing eagles
(Post 2051642)
I saw that, and the researchers believed that 278 of the false positives were related to faulty tests. But then, we have to subtract 278 from both the numerator and denominator (462-278=184/1322-278=1056) which yields a 17.4% false positive rate----you still can't use 900,000 as the denominator. So, throwing out the tests the researchers thought were defective, there's still a 1 in 6 chance that your positive home test is really negative.
|
I am certain that you actually know that in a low incidence event that that positive predictive value is problematic. For the readers... If we were in a smallpox epidemic and our smallpox test had a 0.01% false positive error rate it is a great test. We test everyone with smallpox looking rash, most of whom really do have smallpox, but some don't, they have bug bites, hives, or atypical chicken pox or monkey pox or cow pox. We know from sophisticated more accurate but time consuming testing that when all rashy people are tested in the middle of our smallpox epidemic that 60% really do have smallpox.
So we test 1 million rashy people. The "real" number of smallpox cases in the example would be 600,000 and we'd like to see our rapid test report the correct 600,000 people as positive. But having a false positive error rate of .01% we instead get 600,100 positive tests. It's a big nothing burger in the middle of an epidemic. Only 100 of the 600,100 positives are wrong, 0.016%. In this situation the chance that your positive test being true is over 99.9%. Great test
But instead in today's world where thanks to immunizations and vigorous public health measures and isolation and quarantine, smallpox is gone, the last case in the world being in the 1970s, if we test 1 million rashy Americans with our smallpox rapid test, we report there are 100 positive tests. In this case the chance of your test being wrong is 100%. Same test, same 1 million rashy Americans, same number of false positives.
The false positive rate on our test did not change, but because the rate of the disease changed the odds that any individual test was wrong changes. Now with smallpox gone, every single positive test is wrong. That is not because the test suddenly became more error prone, rather the prevalence of the disease changed.
In our Covid situation, the statement to which I initially responded was that the availability of at home tests would make the number of reported cases jump 1000%. I pointed out that this was wrong because positive at home tests are not going to be reported. And in reply to the claim that the at home test is useless as it is prone to false positives.. I mentioned the Canadian study
What the Canadian study showed was that in a symptom free population of adults of working age you get essentially zero [500 out of a million] false positive tests. So that proves that the tests are NOT going to jump the numbers up because healthy people are NOT going to test positive. That's all it says. We have no idea in the Canadian study how many false negatives there might have been.
GE expressed surprise at how few tests are positive given that Covid was in the community during the study period. Again, the Canadian government, even in the very Conservative western provinces, has been strongly supportive of telling ill people to not go to work, and those exposed but not ill to quarantine and not go to work. That may in part explain the low rate of positive tests. False negatives also are clearly involved.
