Update thread triggered by comments from @patrick_sharkey

1. The simulations were my way of exploring possibilities.

2. I do not draw any conclusions from them other than this:

A policy of “test and isolate,” can reduce R even if the test has a high false negative rate
3. The main goal of the annimations was to give people a way to understand features of these models known to experts but not to most others.

E.g. why is it that a virus that is spreading very rapidly will typically not infect everyone.
4. In showing how infections end, I hoped to make it clear that we do not need to catch every infected peson to control a pandemic.

I was worried that the discussion about contract tracing could encourage the opposite impression — that we must root out every infected person
5. In any evaluation of “test and isolate” Pat is right that we need to estimate the rate of transmission by those in isolation b.c. it will not be zero. We must also allow for possibility that some people may not agree to be tested.
6. Separate from my simulations I took up a line of argument proposed by Sten Linnarsoon and Jussi Taipale — you can put a bound on the number of tests that you need to be sure that “test and isolate” will push R below 1. https://medium.com/@sten.linnarsson/to-stop-covid-19-test-everyone-373fd80eb03b
6. (cont.) The very attractive feature of this argument is that you can come up with an estimate of the required scale of testing based on a small set of parameters:

- the false negative rate of the test
- R0
- Rq (rep num for those in quarantine)
- and compliance rate
7. However, as some, e.g. @AdamJKucharski have observed, this bound is sensitive to the assumption that the mean interval of infection can be treated as a single period.

This is an important point, one that will require more thought and a fuller response.
8. This is like battlefield medicine so I do not feel that I have the freedom to take as much time as I would like to explore all these issues.

That said, I remain confident that there is a rate at which we could test such that test and isolate by itself yields R < 1
9. Moreover, I remain confident that the cost of the required level of testing is at least one, probably 2 orders of magnitude lower than the cost that lockdown is now imposing.

As an aside — no one seems to dispute this assertion, even those who say we “can’t do the tests.”
10. So as we refine the argument about the bound, I will continue to resist any attempt at forcing a premature consensus that:

To get R < 1 we must use

a) lockdown, or
b) digital contact tracing, or
c) both
You can follow @paulmromer.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: