Read on Twitter
Interesting new preprint on diagnostic uncertainty - a higher quantity of COVID19 tests will probably do more harm with poor test targeting (eg. contact tracing)
i.e more "no test is better than a bad test", less "it's not perfect, but it's the best science can do" https://twitter.com/riskcdt/status/1253244638552612864
i.e more "no test is better than a bad test", less "it's not perfect, but it's the best science can do" https://twitter.com/riskcdt/status/1253244638552612864
Key take-away: false positives may be more probable than true positives when infection has low prevalence in a population; countries with early testing but low case fatality rates may report more false positives.
Researchers need caution, which makes it galling another report in Santa Clara County with poor testing got traction; policy-makers looking for evidence-based approaches may adopt sloppy findings which could *increase* lockdown measures in the long run https://twitter.com/NYTScience/status/1252704616711630850
The statistical errors in the SCC report were laid out in this great thread by @wfithian. The Liverpool preprint suggests misguided policy favouring high quantity testing, even with good accuracy, may result in a second peak with a rapid end to lockdown https://twitter.com/wfithian/status/1252692360242159616
i.e. the Liverpool preprint suggests policy shouldn't rely on more antibody tests; diagnostic uncertainty will confound results when the proportion of an infected population is low. @jjcherian has a great thread on how the SCC report mishandled uncertainty https://twitter.com/jjcherian/status/1251272333177880576
