All Times EDT
Keywords: testing, sensitivity, specificity, reliability, COVID-19, diagnostic
With the onset of Covid 19 in each geographic region of the world many national governments, corporations, colleges and universities, and private labs rushed to design, get at least provisional approval for, and market various types of tests. For example, in Germany a single test was developed and was adopted in Germany and a number of other European countries. A similar situation took place in South Korea. The US took a different path with the result that there are a number of tests of different types which have varying levels of sensitivity, specificity, or for that matter reliability.
Imagine a world in which there were a set of tests whose sensitivity, specificity were well known when a diagnosis is first suspected and whose results could form the basis for estimating prevalence in a population. In this same world, an accurate diagnosis could form the basis of a population wide study on topics such as length of time a test was positive and indicative of ability to infect others, immunity if, when, for how long it occurs.
Discussion questions:
What would we as statisticians
(1) define our requirements for testing?
(2) like to see done now?
(3) recommend doing to resolve the current, uneven testing picture?
(4) think are the ethical matters if any?