Thread by @iamgkadam

A (very long) thread on testing, the current scenario, and what ails us from testing more.

TL;DR version - unlikely to be able to do mass-scale testing w status quo. Need to experiment w currently avl test, or develop radically new tests. (1/n)

The WHO mantra has been "Test, test and test"+"test, track and trace". Test=genetic test btw, not serology. Great idea, and I'd say thats how China, Korea managed to keep the lid on. BUT they caught it early, so worked. Why are other countries struggling to ramp up tests? (2/n)

For contact tracing to work, you need to identify every last case. Easier when you have fewer cases ("catch cases early! stop the spread!!" sounds familiar?). Else, there's way too many silent, asymptomatic spreaders to track. So, end up having to test most/full populations (3/n)

Now, can you employ test at scale? No. Why? Because the test is complex (see explainer thread). Again, by "test" - I mean the genetic test where you catch *currently* infected people and tell them to quarantine so they dont spread it to anyone else. (4/n)

https://twitter.com/iamgkadam/status/1242242587999776770?s=20

This test needs many chemicals - aka "reagents" - that come in "kits", basically a set of all the reagents needed. Made mostly only by handful of companies, only 3 of them big, multinationals.

@ASMicrobiology did call this out quite early in March.
(5/n)
asm.org/Articles/Polic…

Similar issue in Europe too. (6/n)

cen.acs.org/analytical-che…

And in India, Africa and more. (7/n)

indianexpress.com/article/india/…

nature.com/articles/d4158…

How much testing do we really need? Just focusing on one country (US), experts say we need 2x-6x current testing levels to go back to "normalcy" and regular life (8/n)

politico.com/news/2020/04/2…

And remember, need to do repeat testing of same person everyday/every few days to catch infected person and their contacts early & prevent spread. So, really need a crazy lot of testing! Insane, right? Even more when you consider the whole world - thats ~20x US population. (9/n)

If you are a NZ or a HK where you havent had cases for days and weeks, then you can stop testing. Until then, you need to - to be sure its not lurking somewhere hidden/spreading asymptomatically.

Now, can we even achieve that level of testing theoretically? (10/n)

There are 3 big global manufacturers of these reagents and kits essential to the test - Qiagen, Roche and Biomerieux. Many more smaller ones (Zymergen et al), but too tiny in comparison to make much of a dent afaik.

How much can the big 3 make? (11/n)

Qiagen says they can ramp up to 6.5mn kits a month by April, and 10mn/mo by June. Say, the other two are similar-ish. Thats about 40-50mn/mo.

That is *per month*. We need to do, just for US, 6mn/day*30d = 180mn. Huge gulf!

Image source: sciencemag.org/news/2020/03/s…
(12/n)

And we havent even considered rest of the world yet. So the gulf is even bigger.

Govts are trying by asking other industries to help, but not sure this is going to suffice. Not everyone is setup to handle potentially harmful chemicals at scale. (13/n)

wkzo.com/news/articles/…

So, whats the way out of the lockdown? We are trying many different strategies, but so far they all rely on the same genetic test. For ex., pooling can help reduce reagent need, or next gen sequencing tech can step in. (14/n)

technologyreview.com/2020/04/28/100…

Other new technologies are being also developed, like LAMP and CRISPR, but these all still need the same reagents in the first part of the testing pipeline. So, still might not be good enough.. (15/n)

webwire.com/ViewPressRel.a…

We need radically new technology and ideas that completely bypass the RNA extraction step. One of them is antigen-based testing. Its a nascent area, but quickly catching up.

(16/n)
technologyreview.com/2020/04/24/100…

Its also good to see NIH finally open calls to fund new diagnostic test development, with a goal to implement by this summer.

(17/n)
nih.gov/news-events/ne…

Also, hackathons, scientist fora and other platforms have led to lots of new ideas. Here's some from the @EUvsVirus from last weekend:

(18/n)
eunitedvsvirus.devpost.com/submissions/se…

Finally, an appeal - I led a team in this hackathon, and we have a very promising idea here based on antigen-testing, looks like a pregnancy test and can be mass produced quite cheaply. Looking for partners to do R&D now, pls reach out if interested /end

Another way out is to decentralize reagent prep, and go old school. Very doable, prone to QC variability but can be controlled if done by well trained pair of hands. Personally not convinced this is a reliable globally applicable soln in current scenario.

crick.ac.uk/research/covid…

Addendum: Nice review just out listing various scientific strategies being tried and developed in our quest towards Covid19 diagnostics.

Note - a very technical read.

https://twitter.com/sgarakyaraghi/status/1256246109133029377

I do realize there's a mistake in my calculations here. Each kit can be used for ~50 tests typically, so 50mn kits/mo==2500mn tests/mo. So it's not as bad as I tweeted, my bad 🖐️

So yes, achievable for US. Not sure if implementable in poorer countries.