Recently, test pooling has been suggested by the Frankfurt Goethe University where you test a group of nasal samples, all in a single test. If the test is negative this means all the samples in the group (pool) were negative, sparing a lot of unnecessary individual tests.
Unfortunately, the Frankfurt group didn't publish any recommended pool size, so anyone using this approach would make a guess for the pool size and use that, which isn't optimal.
Our simulations show that the best strategy to use for doing a pooled test on COVID-19 varies from context to context, as shown bellow:
What this table essentially says is that for the case of Italy for example, where there is an expected 0.33% infection rate among the full population, the optimal strategy for testing 100 people (cohort size), where you are allowed a maximum of 16 samples per pool, is to test:
first in groups of 15
the remaining people in the positive groups, regrouped in groups of 4
the remaining people in the positive groups, regrouped in groups of 2
the remaining people, individually
So, a 4 step strategy of (15, 4, 2, 1) where you should expect, on average, to find all the infections using only 9.47 tests!
That means an approximately 10x increase in capacity! (in some cases even 20x)
Comments