There has been a lot of talk about inaccurate test results.
NAAT (nucleic acid amplification tests) for resp infections generally have very good accuracy
per se. NAAT has been a major advance over earlier techniques. There is no reason to believe NAAT will be any different for this virus.
https://www.ncbi.nlm.nih.gov/pubmed/21073292
https://bmcinfectdis.biomedcentral.c...879-017-2227-x
BUT, a big problem is adequate sample collection.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3185851/
Respiratory virus detection is highly dependent on the type of sample collected, the time of collection after the onset of clinical symptoms, the age of the patient, and the transport and storage of the sample prior to testing (48, 49). Several different upper respiratory tract specimens are applicable for testing, including nasopharyngeal (NP) washes, NP aspirates, and NP swabs placed in virus transport media (48, 49). There are limited data that support the use of combined nose-throat swabs for influenza A virus testing by NAAT (14). Detection of 12 respiratory viruses using a NAAT panel was significantly less sensitive with oropharyngeal swab specimens (54.2%) than with either nasopharyngeal swabs (73.3%) or nasopharyngeal wash specimens (84.9%) (33). This may be due to the substantially lower viral loads in the oropharynx than in the nasopharynx (23).
The more accurate sampling techniques are more likely to induce aerosolization of respiratory secretions (coughing, sneezing) and hence pose greater risk to the tester. Threat swabs are very poor (possibly no better than a coin toss), but are quick and easy and less dangerous to the tester. Nasopharyngeal washings are the opposite in every regard.
Imagine an environment where an entire province has been isolated for the greater good, where there is a sense of being abandoned, where the state has failed to provide sufficient PPE, where health care workers (who have their own families to look after) are getting infected.
Now imagine what tests are being done.
The technology per se is probably good. But like everything else, testing breaks down when the system itself starts to break down.