Testing, Testing, 1, 2, 3, Testing

The testing issue is testing my patience.  Governments have just ignored the now obvious issues in false positives, low positives that don’t identify infectious people, PCR tests that pickup seasonal coronavirus, inadequate antibody assays, and so on.  Here are a few more papers on the topic.

First up, a paper which examines various PCR infection tests.  (Medrxiv Paper)   The authors discuss the components of a typical infection test kit.  They then attempt to assess performance in the real world clinical setting.  The average laboratory using these test kits has a limited ability to understand and address performance issues.  Prepared known samples of CV-19 positive, other coronavirus positive and negative composition were used to test the tests.  All negative samples or those containing another coronavirus were accurately detected by all test kits analyzed.  Four positive samples were identified as positive, but the tests had a variation of 9 cycles before detection occurred.  There was a fair amount of difference in results for various combinations of test kits and instruments.  The authors warn that this could lead to misunderstanding about cycle numbers and conclusions about infectiousness of patients.

This study compared saliva versus nasal swabbing for infection testing.  (Medrxiv Paper)   The study was done in French Guiana among 776 patients, 162 of whom were positive.  40% of the positives were asymptomatic; 60% had mild symptoms.  The sensitivity of saliva versus nasal swab depended on patient group and our old friend cycle number. Ten of the positive patients were positive by saliva but negative by nasal swab; all had cycle numbers of under 25 so you would think the swab should have picked it up.  I may have misunderstood, but it appears that at higher cycle numbers, neither test was all that accurate. It also appears that errors were more likely in asymptomatic patients, likely because they have lower viral loads.  Sensitivity was also affected by how close in time the test was to development of symptoms.  Neither the positive predictive value nor the negative predictive value exceeded 90% and this was in a relatively high prevalence environment.  The authors concluded that saliva testing was not more accurate than nasal swabbing.

And here, straight from the CDC website and updated on September 10, is information about testing and evaluating testing results.  (CDC Info.)  The information was designed to be used to determine isolation periods and sumarizes the evidence about what tests may tell us regarding infectiousness.  The whole thing is worth a read, among other things it suggests reducing isolation times to 10 days from 14, but look at figures 3, 4, 5 and 6.  Very clearly shows how few samples likely are from infectious persons a certain number of days past symptoms and at higher cycle numbers.  Based on the CDC’s own work, no one should be counting a test which required a cycle number of 33 or higher as a positive test.  The likelihood of viable virus is minscule and if there was any, it would be insufficient to form an infectious dose. 

Antibody assays were examined in this study.  (Medrxiv Paper)   Coming from Pakistan, the researchers evaluated several commercial assays.  They used about 400 donors, most of whom had been CV-19 patients, the remainder of whom were health care workers or patients from the community.  There was significant variation in performance and among the CV-19 patients a somewhat significant group was found not to have antibodies, which seems wrong, so that all the assays are not sensitive enough, or perhaps the test was not performed long enough after the infection resolved.  Lasting antibodies may not show up for a few weeks.  They also found that their group of “healthy” donors had a fairly high prevalence.  And they suggested that to get a full immune response picture a fast method of detecting T cell responses was needed.