Testing, Testing

The Royal Statistical Society (RSS) is concerned about the lack of robust statistical evidence for evaluating the clinical and analytical performance of some of novel diagnostic tests for the Covid-19 virus or the immune response to that agent (https://www.theguardian.com/world/2020/sep/27/uk-market-flooded-by-inadequate-covid-tests-experts-suggest). They have set up a working party, as some of the tests (that are sold to UK institutions), claim impressive sounding accuracy but seems based on 'dodgy' statistics. The working party will even test claims for the 'gold standard' PCR test (which detects the virus directly by 'magnifying' the RNA of the virus. Even this, can produce false negatives, when the swabbing technique used to obtain the sample fails to pick up sufficient virus from the nose and throat. It is pointed out that many countries do not allow patients out of isolation, until they have produced 2 negative swabs. It seems that authorities are more keen on counting the number of tests carried out, than worrying about their accuracy. Accuracy is, however, a fundamental requirement if test and trace is to work properly. The RSS working party should be able to tell us how confident we can be about the systems that are in place (they vary from location to location).