Read on Twitter
PRACTICAL Q ABOUT POOLING RT-PCR TESTS: Are there papers showing if/how sensitivity of antigen test fall if you pool more samples? Most papers we found examine statistical properties of pooling, not biological. If sensitivity falls, we want smaller pools that they recommend. 1/9
We are developing a protocol for states in India to pool tests as part of community testing. There is limited capacity to do RT-PCR. (NA tests not available yet here.) But swabs are cheap. Key question: what is optimal pool size? 2/9 @anuacharya @profmohanan
There are several pooling strategies. a/Pool N. If +, test each sample in pool. b/Pool N. If +, split into eg 2 samples of N/2. If one is +, split again & so on. c/Arrange samples in nxn matric w/ n^2=N/2. Pool each row, then each col. Identify + sample if row & column pool +. 3/
Q is what N should be for each. Common finding: size of optimal pool false with true prevalence. With very low prevalence, large samples are good. Only a few show positive. With high prevalence, smaller sample are good, otherwise all pools show positive. 4/9 @causalinf @Jabaluck
This paper shows, statistically, that the simplest strategy (a) is pretty good compared to (b) for all but very low prevalence & compared to (c). From this we infer a good size for us, statistically, is 8, which covers prevalence between .8 and 4%. https://www.medrxiv.org/content/10.1101/2020.04.06.20052159v1
5/9
5/9
This paper looks at statistical value, using (b).
https://www.medrxiv.org/content/10.1101/2020.04.03.20051995v1
@jebujohn
There are older ones not on COVID that do similar. E.g.,
https://journals.plos.org/plosntds/article?id=10.1371/journal.pntd.0007196 on heminths.
This accounts for costs of errors (incorrect quarantine).
https://www.tse-fr.eu/optimal-group-testing-exit-covid-confinement
@CGollier
6/9
https://www.medrxiv.org/content/10.1101/2020.04.03.20051995v1
@jebujohn
There are older ones not on COVID that do similar. E.g.,
https://journals.plos.org/plosntds/article?id=10.1371/journal.pntd.0007196 on heminths.
This accounts for costs of errors (incorrect quarantine).
https://www.tse-fr.eu/optimal-group-testing-exit-covid-confinement
@CGollier
6/9
The only pooling paper that does COVID tests notes sensitivity could be an issue, but does not show how sensitivity could fall with pooling.
https://jamanetwork.com/journals/jama/fullarticle/2764364
@StanfordPath
7/9
https://jamanetwork.com/journals/jama/fullarticle/2764364
@StanfordPath
7/9
There are some news reports of other groups, but I can't find articles.
https://www.labmedica.com/covid-19/articles/294781270/israeli-researchers-introduce-pooling-method-for-covid-19-testing-of-over-60-patients-simultaneously.html
https://eurekalert.org/pub_releases/2020-03/guf-pto033020.php
8/9
https://www.labmedica.com/covid-19/articles/294781270/israeli-researchers-introduce-pooling-method-for-covid-19-testing-of-over-60-patients-simultaneously.html
https://eurekalert.org/pub_releases/2020-03/guf-pto033020.php
8/9
So, are there any papers that show if/how sensitivity falls as pool size grows? (Even if for non-covid.) Intuition suggests yes, And if so, we may want to reduce pool size further.
If not, we'll go with pool size of 8 based on work by @itamarlandau @YayonNadav.
9/9 @arpitrage
If not, we'll go with pool size of 8 based on work by @itamarlandau @YayonNadav.
9/9 @arpitrage
I apologize if I have missed anyone else's papers!! If I did, please respond to this thread & tell me. Will read ASAP.
10/9
10/9
