I noticed today that there have been comments in Nature.com and The Lancet Respiratory Medicine about the false positive problem of Covid-19 tests.
Each points to the conditional probability problem of using tests with seemingly good accuracy to screen large numbers of people lacking symptoms, among a population with a low prevalence of the disease – which can lead to extraordinarily high false positive test results.
I buried links in a tangentially related post on a social media meme over on my Social Panic blog which also explains the problem in detail.
As this has finally bubbled up to “experts” in official publications, this may be noticed by the media in a few weeks, probably right after the election on November 3rd.
Caution – this does not imply that Covid-19 is not a problem. The disease impacts some people very hard and seems to kill about 1% of those who get the disease, especially those who are elderly, have other health conditions or perhaps received very large viral loads when infected. Fortunately, death rates have dropped even as positive test results have skyrocketed; hospitalizations are also trending upwards, yet are generally well below the “early days” of the pandemic.
Update – the Kinsa thermometer data indicates the outbreaks in the midwest are likely at their peak or starting to decline. Next areas of worry are parts of the east coast and the southeast, mostly. Consequently, my satirical title of this post as “Two and half weeks” might more likely be about 4-5 weeks. But still … I think that is where this is going.