bkp_duke
Well-Known Member
How bad are the PCR tests we're using in the US? That seems really bad and there would be no way to stop the spread by testing.
This LabCorp PCR test looks like it has 100% sensitivity (small sample size): https://www.fda.gov/media/136151/download
@EVNow 's test comparison is a tad flawed in that paper, but does bring up one of the finer points of medicine. The reason why: SAMPLE COLLECTION. I don't doubt the published results (for THAT specific institution only), but there is a very large assumption being made that I didn't see the paper describe.
The PCRs are being run on nasal and throat swabs. That takes a TRAINED person to do that properly, so the quality of the results is dependent upon sample collection. Someone that doesn't know how to properly swab a throat is not going to pick up a true positive, even though the PCR test itself is nearly 100% sensitive and 100% specific.
The IgG and IgM tests have lower sensitivity and specificity, but they are run on blood samples. Blood is pretty much binary (you get it, or you don't and the person collecting it can readily see that in the sample tube).
There is also a timing effect that has to be considered. The PCRs are being run and looking at ONE point in time to see if you are infected by picking up viral RNA. There is a window of opportunity of only a few days to 2 weeks (depending upon how fast the host fights off the infection).
Conversely, the IgG and IgM test is being run at a point when you have likely produced antibodies. ANYTIME after the start of antibody production (which is later in the course of the disease), and also anytime AFTER the infection is cleared, this test should be positive for months.
So, you order the test based upon what you are looking for:
1) Active infection - order a PCR (preferably more than one sample - nasal and throat are what I would start with)
2) Looking for immunity - order the IgG / IgM antibody blood test
BTW, the false negative rate can be greatly reduced on the PCR, even by a bad sample collection, by repeating the test. This is where the training of the physician ordering it is important: clinical suspicion. If the case looks like COVID-19, and the PCR is negative, and a flu test is negative (because flu looks a lot like COVID-19), you would presumptively treat for both conditions, and repeat both tests (preferably with a different technician doing the collection