[Goose]

If you click on that link, the article contains information about which test Wuhan uses (Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) technique) and in what situations is a test done, and the physical collection of a sample.  Apparently (only) Hubei Province is also using lung imaging to classify patients as "clinically diagnosed" (detecting "advanced symptoms" which I presume means pneumonia), when the lab tests can't be done.   The different testing would seem to make the data messier to analyze.

I haven't looked at the CDC testing protocol.

Unfortunately saying the test uses Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) technique is a bit like saying wine is made by fermentation of grapes. To really compare the tests, one would have to know what the target fragments are and what primers are used. The odds that two different groups tasked with developing a test for the virus would choose the same target fragments and the same primers without coordination in that task are essentially zero. There are lots of different possibilities that will "work", but with different specificity, different false positives, and different sensitivities, as well as different difficulties of manufacturing. Some primer choices even for the same fragment are "better" than others. While there is software that can help with this choice, it is far from an exact science to obtain an "optimal" one. It is also necessary to ensure the primer that is manufactured for general use is really the "same" as the one used to develop the test (not degraded or contaminated in some way). This isn't real hard to do, but with all the time pressure to get the test out there that step may have been skipped, which can add to the confusion.

Normally, validating a test like this is a long process with many testing steps to determine exactly how the test performs. The one test the CDC dropped out (of the original three) was probably present to increase the specificity of the test from say 95% to 99.9%. It wasn't just window dressing. Unfortunately they had trouble with it. Not a big surprise with the time pressure to get this done and in the field. The good news is that PCR testing is pretty common in medicine these days, and many hospitals have the machines that do it. Once you have a "kit" it is a mostly automated process and with validated test works very well. The current tests are being validated "on the fly", which isn't optimal, but is unfortunately necessary.

Note that I am not by any means an expert in this field and the above is derived from conversation with others.

[Goose]

If you click on that link, the article contains information about which test Wuhan uses (Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) technique) and in what situations is a test done, and the physical collection of a sample.  Apparently (only) Hubei Province is also using lung imaging to classify patients as "clinically diagnosed" (detecting "advanced symptoms" which I presume means pneumonia), when the lab tests can't be done.   The different testing would seem to make the data messier to analyze.

I haven't looked at the CDC testing protocol.

Unfortunately saying the test uses Reverse Transcriptase Polymerase Chain Reaction (RT-PCR) technique is a bit like saying wine is made by fermentation of grapes. To really compare the tests, one would have to know what the target fragments are and what primers are used. The odds that two different groups tasked with developing a test for the virus would choose the same target fragments and the same primers without coordination in that task are essentially zero. There are lots of different possibilities that will "work", but with different specificity, different false positives, and different sensitivities, as well as different difficulties of manufacturing. Some primer choices even for the same fragment are "better" than others. While there is software that can help with this choice, it is far from an exact science to obtain an "optimal" one. It is also necessary to ensure the primer that is manufactured for general use is really the "same" as the one used to develop the test (not degraded or contaminated in some way). This isn't real hard to do, but with all the time pressure to get the test out there that step may have been skipped, which can add to the confusion.

Normally, validating a test like this is a long process with many testing steps to determine exactly how the test performs. The one test the CDC dropped out (of the original three) was probably present to increase the specificity of the test from say 95% to 99.9%. It wasn't just window dressing. Unfortunately they had trouble with it. Not a big surprise with the time pressure to get this done and in the field. The good news is that PCR testing is pretty common in medicine these days, and many hospitals have the machines that do it. Once you have a "kit" it is a mostly automated process and with validated test works very well. The current tests are being validated "on the fly", which isn't optimal, but is unfortunately necessary.

Note that I am not by any means an expert in this field and the above is derived from conversation with others.