Covid detectable in sewage sample back in March 2019

Covid-19 discussion, bring your own statistics
Post Reply
monkey
Stargoon
Posts: 95
Joined: Wed Nov 13, 2019 5:10 pm

Covid detectable in sewage sample back in March 2019

Post by monkey » Mon Jun 29, 2020 4:18 pm

Hello all,

I found a random on the internet using a preprint study of Barcelona sewage samples that shows that SARS-CoV-2 was in people last March (at least in Barcelona). I was immediately skeptical of this claim, because I think we would have noticed by other means if that was the case, and the theory in the paper is that it was masked by seasonal flu, which seems implausible to me. The random was using the paper to claim that the virus is nothing to worry about, "stands to reason, if it's been around that long, innit" (not actual quote, but should give the impression of what was said).

The claim seems to rest on a small amount of Covid RNA being detected in a frozen sample of sewage. There doesn't seem to be any statistical analysis done on the key result for the conclusion that it was around last march. However, I don't know enough about the methods used to know what might have happened, apart from a possibility of accidental contamination, or if you're doing around 50 tests, a couple might give a positive by accident.

I'm hoping someone here who knows stuff can explain things to me so's I can tell someone they're wrong. I am also happy to be told they are right, but not as happy, because then I'm wrong but at least I learnt something.

The preprint is here: clicky.


Abstract:
SARS-CoV-2 was detected in Barcelona sewage long before the declaration of the first COVID-19 case, indicating that the infection was present in the population before the first imported case was reported. Sentinel surveillance of SARS-CoV-2 in wastewater would enable adoption of immediate measures in the event of future COVID-19 waves.

monkey
Stargoon
Posts: 95
Joined: Wed Nov 13, 2019 5:10 pm

Re: Covid detectable in sewage sample back in March 2019

Post by monkey » Mon Jun 29, 2020 5:30 pm

Just found a post in another thread by Woodchopper with a link to a blog discussing this. Thanks Woodchopper!

post: viewtopic.php?p=38859#p38859

blog: https://scienceintegritydigest.com/2020 ... na-sewers/

User avatar
Woodchopper
After Pie
Posts: 1770
Joined: Sat Oct 12, 2019 9:05 am

Re: Covid detectable in sewage sample back in March 2019

Post by Woodchopper » Mon Jun 29, 2020 6:02 pm

monkey wrote:
Mon Jun 29, 2020 5:30 pm
Just found a post in another thread by Woodchopper with a link to a blog discussing this. Thanks Woodchopper!

post: viewtopic.php?p=38859#p38859

blog: https://scienceintegritydigest.com/2020 ... na-sewers/
Ta :D

User avatar
Bird on a Fire
Stummy Beige
Posts: 3476
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Covid detectable in sewage sample back in March 2019

Post by Bird on a Fire » Mon Jun 29, 2020 7:35 pm

monkey wrote:
Mon Jun 29, 2020 4:18 pm
However, I don't know enough about the methods used to know what might have happened, apart from a possibility of accidental contamination, or if you're doing around 50 tests, a couple might give a positive by accident.
That's the case with statistical tests (with alpha=0.05), but I don't think the same is necessarily true for testing genetic samples. There are techniques people use for checking for false positives, like running negative controls (e.g. blank or known-negative samples), testing in duplicate/triplicate, etc. It ought to be the case that a positive test result means the sample is positive for coronavirus, though it doesn't tell you how or when coronavirus got in the sample.

From the blog:
The authors tested the presence of the SARS-CoV-2 virus using different Real-Time PCRs (rtPCRs) on five different viral regions, i.e. IP2, IP4, E, N1, and N2. The IP2 and 4 PCRs both detect parts of the viral RNA-dependent RNA polymerase (RdRp) gene. They followed the WHO / Institute Pasteur RT-PCR protocol to detect the SARS-CoV-2 virus.
In theory, that's a pretty sensible protocol with a lot of redundancy built in - it ought to be a lot more sensitive/specific than a 5% error rate.

However:
As expected, most wastewater samples were positive between January and May 2020, in most of the PCR assays. The first COVID-19 case in Barcelona was reported on February 25, so it seems the virus might have already been circulating a couple of weeks before that.

But then, the authors looked at some older wastewater samples. One of these, from March 2019 (a year before the first COVID-19 cases in the city) appeared to be positive as well.

It was just one sample, though, and it does not appear that the authors repeated the test at all. No replicates as far as I can tell. But also note that only 2 of the 5 RT-PCR tests were positive, i.e. on the IP2 and IP4 targets.

Now, I am not a viral gene specialist, but several SARS-CoV-2 PCR assays have been described to give false positives. Even the WHO protocol to detect the virus, which was the protocol followed by the Barcelona group, states that a positive IP2/IP4 PCR, although specific in a small panel of other viruses, should be confirmed by the “E gene assay”. That test was negative in the Barcelona sample.

But the authors did not do any other target PCRs to confirm the positive sample. They also did not do any genomic sequencing. They did not do any confirmation to rule out cross-reactivity with some other virus. They also did not report on negative controls to rule out contamination of the 2019 sample with one of the positive 2020 samples. There is nothing in the paper about negative controls or strict separation of pre- and post-PCR rooms, so it is not clear how stringent they were in preventing cross-sample contamination.
So it sounds like they have no way to rule out contamination of the sample, and can't exclude the possibility that some other virus - possibly even another unknown coronavirus of animal origin - ended up in that particular sample of sewage.
Born at 356.32 ppm CO2 #ShowYourStripes

User avatar
sTeamTraen
Catbabel
Posts: 756
Joined: Mon Nov 11, 2019 4:24 pm

Re: Covid detectable in sewage sample back in March 2019

Post by sTeamTraen » Mon Jun 29, 2020 8:33 pm

Apparently they have traces of two of the five markers that they look for, in sample from March. Then nothing until 2020. We are expected to believe that this highly contagious virus showed up in one of the most densely populated cities in Europe in March 2019 in quantities sufficient to last over a year and be detectable in a lab, then disappeared only to return nine months later.

Alternatively, they have some sort of contamination of the sample with another related virus.

Hmm, what shall we do? Run some more tests, or put out a preprint and have our university press office tell the whole f.cking world that the virus was circulating for nearly a year without anyone noticing, thus confirming the claims of everyone on Facebook who definitely had COVID-19 last November, ooh it were awful, I couldn't get out of bed for three days, you know what, I reckon this lockdown is bollocks, I'm going to the pub?

Of *course* they're going to go for the preprint and the press release. c.nts.

This reminds me of the "faster than light neutrinos" thing a few years ago. Only then, the basic attitude from the scientists was "Er, hi everybody, we've found this, it's probably an artifact but help us out here because otherwise the laws of physics are in danger". And sure enough it was a bad cable.
Careering along in my creosote car

monkey
Stargoon
Posts: 95
Joined: Wed Nov 13, 2019 5:10 pm

Re: Covid detectable in sewage sample back in March 2019

Post by monkey » Mon Jun 29, 2020 9:14 pm

Bird on a Fire wrote:
Mon Jun 29, 2020 7:35 pm
monkey wrote:
Mon Jun 29, 2020 4:18 pm
However, I don't know enough about the methods used to know what might have happened, apart from a possibility of accidental contamination, or if you're doing around 50 tests, a couple might give a positive by accident.
That's the case with statistical tests (with alpha=0.05), but I don't think the same is necessarily true for testing genetic samples. There are techniques people use for checking for false positives, like running negative controls (e.g. blank or known-negative samples), testing in duplicate/triplicate, etc. It ought to be the case that a positive test result means the sample is positive for coronavirus, though it doesn't tell you how or when coronavirus got in the sample.
I was assuming that it was only a problem because they were only finding small numbers of doodads. I thought that if there's loads of the doodads, you don't have to worry much, but you might have to if you're not getting many. I also assumed that there's be a test you could do to see how confident you could be in your result (presumably based on how many doodads you found and the sensitivity of the test).

But again, It just seemed iffy to me even I don't know much about all this, which is why I came here :)

User avatar
Bird on a Fire
Stummy Beige
Posts: 3476
Joined: Fri Oct 11, 2019 5:05 pm
Location: nadir of brie

Re: Covid detectable in sewage sample back in March 2019

Post by Bird on a Fire » Mon Jun 29, 2020 10:12 pm

monkey wrote:
Mon Jun 29, 2020 9:14 pm
Bird on a Fire wrote:
Mon Jun 29, 2020 7:35 pm
monkey wrote:
Mon Jun 29, 2020 4:18 pm
However, I don't know enough about the methods used to know what might have happened, apart from a possibility of accidental contamination, or if you're doing around 50 tests, a couple might give a positive by accident.
That's the case with statistical tests (with alpha=0.05), but I don't think the same is necessarily true for testing genetic samples. There are techniques people use for checking for false positives, like running negative controls (e.g. blank or known-negative samples), testing in duplicate/triplicate, etc. It ought to be the case that a positive test result means the sample is positive for coronavirus, though it doesn't tell you how or when coronavirus got in the sample.
I was assuming that it was only a problem because they were only finding small numbers of doodads. I thought that if there's loads of the doodads, you don't have to worry much, but you might have to if you're not getting many. I also assumed that there's be a test you could do to see how confident you could be in your result (presumably based on how many doodads you found and the sensitivity of the test).

But again, It just seemed iffy to me even I don't know much about all this, which is why I came here :)
In general, false negatives are a bigger concern than false positives - in theory it should be impossible to amplify a sequence that isn't present in the sample (and it practice it is at least very difficult), but it can be difficult to amplify something only present in very tiny amounts, or highly fragmented.

For instance, according to the linked doc on the WHO protocol which this study supposedly used has a 95% sensitivity/"hit rate" - ie a 5% chance of false negative (I'm no expert on qPCR but I think this results from the threshold they apply to distinguish 'signal' from 'noise' in the fluorescence stage of the reaction). However, false positives should be rare, as they've excluded cross-reactivity with a bunch of other (presumably human?) respiratory viruses (influenza A(H1N1)pdm09, A(H3N2), B-Victoria, B-Yamagata; influenza C; RSV A, B; hBoV; hPIV; hMPV; HRV/enterovirus; adenovirus; hCoV (HKU1, OC43, 229E and NL63); MERS-CoV).

False positives could result from a lack of sequence specificity, e.g. if the particular regions they amplified (two of a possible five, remember) also occur in other viruses. The WHO protocol seems designed for testing human rather than environmental samples, so possibly there's cross-reactivity either with non-respiratory virus, or a respiratory virus in other species - perhaps some other coronavirus circulating in sewer rats, or a sample collected under a bat roost. Or it could come from contamination (probably more likely, but also more boring).
Born at 356.32 ppm CO2 #ShowYourStripes

Post Reply