Hacker News new | ask | show | jobs
Sewage could reveal true scale of coronavirus outbreak (scientificamerican.com)
184 points by plcancel 2265 days ago
15 comments

Before anyone goes collecting sewage samples, consider deriving your scaling factors from discrepencies in ICD-10 diagnostic encoding practices.

- Stacked bar chart: https://twitter.com/OccupyWallSt/status/1246478379064295425/...

- Govt data: https://data.cdc.gov/NCHS/Provisional-Death-Counts-for-Coron...

ICD-10 is really the heart of the matter. Earlier recommendations have asked doctors to encode corona death information using Pneumonia (J12.89) or Bronchitis (J20.8, J40) if positive test results are available, suffixed by B97.29 if authorized by the CDC to do so. We now appear to finally have the pneumonia encoded info thanks to NCHS. Hopefully govt will open source the bronchitis encoded reports too.

See also:

- https://www.cdc.gov/nchs/data/icd/ICD-10-CM-Official-Coding-...

- https://www.cdc.gov/nchs/data/icd/ICD-10-CM-April-1-2020-add...

- https://www.cdc.gov/nchs/data/icd/COVID-19-guidelines-final....

What is the implication here? Is it that the number of pneumonia deaths is the real indication of covids prevalance?
We don't know yet. These numbers are still preliminary, but it helps us to understand better the impact of the workarounds that doctors have needed to use, to tell the CDC it's COVID, when the software does not yet have a button for COVID.

The good news from the recent report (https://data.cdc.gov/NCHS/Provisional-Death-Counts-for-Coron...) is that overall mortality in America these past few months, has actually been lower than previous years. The bad news is that we should prepare for the possibilty that mortality for the new coronavirus specifically could turn out to be much worse than many non-official sources online have speculated, and that what's happening in NYC is happening in a lot more places too. Correcting for software workarounds and verifying good accurate data takes time.

I expect 400k COVID-19 deaths in the US. Anything less than that is a major win. Anything less than the 2.2M is progress.
This is likely a gross overestimation. We're at 70k worldwide deaths; turns out it's just not as bad as we were initially told.
> Before anyone goes collecting sewage samples, consider deriving your scaling factors from discrepencies in ICD-10 diagnostic encoding practices.

Did I understand correctly that there was two diagnostic codes for cause of death related to the coronavirus? One for "potentially related" to COVID-19, and one for "definitively caused by" COVID-19?

If so, how would this skew actual numbers reported by the WHO or CDC for cause of death?

Actual paper:

Presence of SARS-Coronavirus-2 in sewage / Medema, et al.

https://www.medrxiv.org/content/10.1101/2020.03.29.20045880v...

Check out Table 4.

Seeing papers like that, I really appreciate what tex did for scientific publishing. (Yes, I understand formatting was not their top concern right now)
That is the default MS Word theme on Mac (unsure if Windows also uses that color scheme as the default). No real excuse to have multiple tables labeled "Table 1“.
Maybe, equally we have every idiot going to the useless John Hopkins site because it looks pretty.

I guess it's an arms game in part.

Thank you! At least someone understand my frustration. When I try to compare countries, it somehow manages to insert New York along with a few other US states as countries. He/She/They might be a good React/GraphQL ninja(s), but definitely skipped out on the ETL course.
Is this any more useful as a comparison? http://nrg.cs.ucl.ac.uk/mjh/covid19/
What's your preferred 'dashboard' for updates?
Interesting. That's also been done for recreational drugs. And artificial vanilla flavoring.

I wonder whether sampling from sewage pumping units could be done. That might produce higher resolution data. But getting representative samples would be harder. You'd probably need to take large samples, mix, and subsample.

Here's a recent wastewater analysis study showing the prevalence of several illegal drugs in Europe.

http://www.emcdda.europa.eu/topics/pods/waste-water-analysis...

> This report is focused on illicit stimulants. No results for cannabis are reported because cannabis use is estimated by measuring its main metabolite (THC-COOH), which is the only suitable biomarker found so far, but is excreted in a low percentage. More research is needed to understand the excretion percentage of THC-COOH or find alternative biomarkers (Causanilles et al., 2017a).

this is the second sewage testing paper I’ve read today. Why are these so incredibly interesting and fun to read when compared to the other testing documents I’ve read haha.

It's a kind of hack on modern sanitation systems. It makes you realize that everyone is (literally!) leaking all sorts of information about what's in their bodies. You just need to know where to pick up that treasure trove of information. It tickles the hacker fantasies!
Also, I'm wondering at what point privacy-groups will start complaining about this sort of testing. In the future, the testing will get to a point where smaller and smaller parts of cities can be measured independently and that would yield policy-action by the government. E.g. "This area over here has lots of meth users, let's police that area more."

Or... They test everyone's sewage independently and privacy-advocate groups cry foul because sewage (and garbage) are personal information and the city can't use it to discriminate or punish people!

Part of the issue with meth specifically is that there are legal 'cousins', although the expected levels of prescriptions differing could tell.

Really the sewage issue is just proof that the system is broken on multiple layers. First off is sticking with the war on drugs when it is clear that fighting abstract concepts or inanimate objects is like getting drunk and boxing with a lamppost - you just look stupid and hurt yourself. Second the precedents and doctrines of 'in plain view' of discards including what is essentially legally required. They were done in an era where the data would be useless to the police even if they could process it. Third, for all of their new data the process isn't actually improving things on any level. Fourth they are running on tautologies enforcing because they must enforce not because they have any vision for an outcome or a connection between their actions and results beyond 'because it is their job'.

Here's a horrifying thought:

"Smart" toilet/sink/pipe + sewage testing = data

Reminds me of an old "Bastard Operator From Hell" story. https://www.theregister.co.uk/2004/03/09/bofh_protecting_bod...
I remember reading something about this about a decade ago maybe in wired or popular science?

It was interesting because some organizations didn’t want to know that level of detail, which drugs were being used when, how many students came to school drunk etc. and all by sampling the sewage system regularly

Can you talk about the vanilla a little more? I don’t get that one.
> In the area around Puget Sound, the University of Washington team found, the spicy residues that remain in wastewater end up flowing into the sound's inland waterways.

> Of all the flavors trickling downstream, artificial vanilla dominates the sound, Keil said. For instance, the team found an average of about six milligrams of artificial vanilla per liter of water sampled.

> The region's sewage runoff contains more than 14 milligrams of vanilla per liter. This would be like spiking an Olympic-size swimming pool with approximately ten 4-ounce (113.4-gram) bottles of artificial vanilla.

https://web.archive.org/web/20170225090427/http://news.natio...

I'm not sure I'd assume that all of that vanillin would come from human sources.

The decomposition of wood creates vanillin, which I assume makes up most of what these folks found.

Huh.

I see that some artificial vanillin is made from lignin, but I don't know whether natural wood decomposition creates vanillin. Do you have a reference for that?

lignin is wood (or rather it's a polymer in the cell walls of wood/plants)
It literally says how much is contained in the sewage run off. You don't even have to click the link.
Ah yes, the quantified toilets joke from SIGCHI 2014 has come to unironic fruition:

https://www.theatlantic.com/technology/archive/2014/04/what-...

It was meant as a prank, not an instruction manual.

What a wonderful symmetry to the cholera outbreak in 1854 in London, where John Snow traced the sources to drinking water wells: https://en.wikipedia.org/wiki/1854_Broad_Street_cholera_outb...
Interesting that while Snow blamed germs, 'authorities' later blamed 'miasma' instead and ignored Snow. Once again, we learn from history that...
I’ve been wondering about this. It would be especially interesting if, after we come out of lockdown, we could sample sewage at many points as the sewer comes together. You might be able for example to identify an infected block and then go around to give individual swabs to each resident if the block came up positive.
This should also allow to monitor for mutations of the Covid 19, whereas it would be impractical to do at an individual level. It would give an early warning signal.
There’s a YC company [0] already doing this.

[0] https://www.biobot.io/covid19

STFC19aaS (Sewage Testing for COVID-19 as a Service)
Serious question: are there 'pandemics' we don't hear about because the virus doesn't hurt us? Are those outbreaks also studied?
If you mean animal flus that didn’t cross over to humans then certainly. Some of them are in wild animals where we might not casually notice a drop unrelated to our activity, or in farm animals where the heavy use of antibiotics reduces the spread within some farms.
I'm sure there are great reasons, but why is it that we can't detect the virus in wastewater at a personal level? Could this be a less invasive way to test for covid19?
That'd arguably be less convenient than nasal swabs. But on the other hand, it might be more sensitive for early infections, given that TFA notes:

> Studies have also shown that SARS-CoV-2 can appear in faeces within three days of infection, which is much sooner than the time taken for people to develop symptoms severe enough for them to seek hospital care—up to two weeks—and get an official diagnosis, says Tamar Kohn, an environmental virologist at the Swiss Federal Institute of Technology in Lausanne.

Still, that'd be a lot of samples to test.

There's an Adult Swim short about that: https://www.youtube.com/watch?v=DJklHwoYgBQ :)
They just never took the Smartpipe bit to its conclusion. No punchline happened. Broomshakalaka, however, is masterclass.
It could also look at other markers as a proxy for possible covid infection. If we see that a whole block is getting an unusual amount of sicknesses, we might want to investigate/quarantine that block.
Public statement from the Dutch CDC (RIVM) on this method:

https://www.rivm.nl/node/153991

Maybe. Similar to drug testing an entire city using sewage.
That is interesting. Do we now what the rate at which the virus dies in water?
You don't need live viruses, just their RNA.
... so do we know how fast the RNA degrades in sewer conditions?
very, very fast.
Slowly enough to not matter. Unless something is degrading it, RNA can last years.
I’m pretty sure ribonucleases are found everywhere (even in sewage) so if enough are present it would break down all RNA fast. RNases come from skin and dust, for example.
lol no, rna degrades if you so much as look at it. dna is quite a bit more stable.

https://www.bioline.com/rna-hints-and-tips/

Incidentally, we can also test sewage for drugs and then swat places where people consume drugs. How very helpful. It's just because we need to fight the virus! ️
Except in areas without sewer