UKH

Not how much Chlorine there is in your local swimming baths, but testing of large numbers of people for Covid.

Bloke on the radio this afternoon, who are doing CV19 testing, mentioned pool testing, if I've got this right, you get a large number of samples, mix them together, then test the whole lot for CV, if it's negative, then the whole sample group is negative, if not you re-test a smaller subset of the group, till you find out who has it.

One test can clear loads of people, so you can do more tests. It's simple, like a guess my number game.

We're not doing this, it's not been validated in the UK.

It means you can essentially test many numbers of people at the same time, it seem to make so much sense I can't understand why we aren't already doing it.

There was a thread on this a while ago. In short, in my experience (mainly based on PCR genotyping of rare recombination events) pooling is great if you can pool already at the first and most labour intensive steps. Here, this would be sample collection and initial processing, which is impossible for obvious reasons.

Any gains from pooling later will be lost from having to retest every single sample from every positive pool (making it counterproductive to make the pools to big). Late pooling, e.g. just for the PCR, will inevitably cause more false negative tests due to sample dilution in the first PCR round and sample degradation in the second. Neither is ideal in a scenario where sample collection efficiency is a known problem: Many false negative tests failed because the swabs did not catch enough nasal secretion, or there was little virus high enough in the throat.

Similarly, sample collection is the only likely stage where kit (nasal swabs, sample collection tubes with lysis and stabilization reagents) may run out in certain regions and pooling may therefore be helpful. Unfortunately, this is also the step where pooling is logistically impossible.

CB

Do they all have to share the same swab? 

It is kind of the same idea as this https://www.bbc.co.uk/news/science-environment-53635692

It should technically work as the test just picks up the virus RNA just depends on how dilute it can pick up the RNA.

'More or Less' on Radio four, suggested than by splitting a sample into multiple pools, it is possible to avoid retesting, if only a small percentage are positive.

It has some drawbacks regarding sensitivity however.

https://www.bbc.co.uk/programmes/p02nrss1/episodes/downloads

It would save on swabs, I suppose

I wonder if they could recycle some of the ear swabs flushed down the bog, which end up in the link you posted.

What obvious reasons? They are using this in the states, not that that's a great example, but that might be for different reasons.

I get this, but surely this is a problems with logistics rather than, testing itself.

The fact we haven't sorted out Test and Trace after the length of time we've had to make it work, and because we have been sending tests out to other countries, German and Italy to name a few, would mean I would have thought we should be doing everything to test as many as possible.

Making the false negative rate even worse?

I suppose if it allows testing of more people it might detect more cases even if the sensitivity or false negative rate is worse.

Nice one cheers.

The main logistic problem in covid testing is to get the sample collected properly, not having some serco wanker who has trouble finding his own nose in the dark pointlessly poking around peoples noses and throats.

Once you have the samples, you need to get them to the PCR facility in individual, sealed vials.

These are the rate limiting steps, and unless you want to reuse swabs, you cannot pool at that stage. This is what I meant by obvious reasons for not pooling early.

Once the sample is processed and its RNA extracted you can dilute it, and either make multiple pools with different compositions and combinatorially arrive at a small subset of samples that do need to be retested individually, or to combine parts of the samples into multiple pools whose optimal size depends on the fraction of positives you expect. This could of course be calculated, but in my experience it is not worth the bother.

The first approach requires precise tracking and complicated pooling schemes, a surefire recipe for data labelling or pipetting f*ckups. The second needs quick PCR turnarounds (to allow retesting before the sample degrades.

Neither is going to work reliably when the testing infrastructure is near capacity. It is IMO much safer to keep things simple and increase sample collection capacity, especially for the labour intensitive early steps. Everything down the pipeline is scalable by simply putting another set of machines into the pipeline.

CB

edit: forgot to address your second point: The fact that the UK sent tests to Germany and Italy suggests that there is indeed also a bottleneck at the PCR stage, but for the reasons above I would suggest that this is better addressed by expanding machine capacity than by complicating processes even more in a system that is apparently stretched to breaking point.

Have you ever done PCR genotyping using pooling? IMO if you have to do that something else went wrong in your experiment design. The ONLY time I found it worth the hassle is when I tested large sets of experimental animals for very rare genetic events. I collect a tissue sample from every individual (actually, grind up the founder fly from a genetic cross), as I can then pool, say 50 such flies already for DNA extraction (the main, labour intesive step).

If I want to test 1000 flies that means I have to do the extraction procedure maybe 20 times, and if I get one positive pool I retest the progeny from the 50 crosses that went into this pool, ending up with having to do the extraction step 70 times rather than 1000 times. If I get many different positive pools I would also only need retest only a small fraction of samples, as my chances would be much better. The actual numbers are quite a bit higher, but the principle still applies.

If that pooling were impossible at sample collection stage I would not bother pooling at all.  To do the 1000 PCRs just means loading the samples into three 384 well plates and shoving them into the pipetting and PCR robot in the neighbouring lab.... Also, I know that I can always extract enough DNA that I could dilute my samples much, much further without losing the signal, and in my scenario I do not care about false negatives as long as I catch one positive. Clearly, neither fact applies to Covid testing.

I am certain that this is the reason why pooling is - to my best knowledge - not done in any country with a well functioning testing infrastructure.

CB

The problem with putting more capacity in, is as the bloke on the radio said, they can't get funding for it.

That was his reason for suggesting pooling. He seemed to think it was a better option under the current circumstances.

I agree with your comments about Serco, but that's going to be applicable to individual tests or pooled tests, isn't it?

Of course the government could fund it (machines are generally cheaper than people), but then presumably their pals do not happen to own companies that sell PCR robots....

The problem is that false negative tests exist already, presumably because low viral load in the upper airways makes proper sampling so important. Take the swab a bit too low, and the detected virus level will disappear into the background.

Pooling under these circumstances is a recipe for disaster, especially when the system can tolerate a false positive much better than a false negative. Would you rather have a nurse or teacher have a free week of home holidays while healthy, or do you prefer that they continue working because they have been tested negative while infected?

CB