Test Week: Let's talk about Sewage
Will testing sewage for the novel coronavirus help us contain outbreaks more effectively?
Welcome to Plugging the Gap (my email newsletter mainly about Covid-19 and its economics). My goal is for several posts a week explaining economic research and the economic approach to understanding the pandemic. (In case you don’t know me, I’m an economist and professor at the University of Toronto. I have written lots of books including most recently on Covid-19. You can follow me on twitter (@joshgans) or subscribe to this email newsletter here).
It’s the end of Test Week here at this newsletter. Thus far, we have talked about testing for diagnosis, clearance and mitigation with respect to Covid-19. The fourth use of testing is surveillance. One method that has emerged has been to monitor sewage waste for the presence of the coronavirus. This allows public health officials to monitor where there might be outbreaks and, therefore, where containment efforts should be focussed. But how far can we go for surveillance? In particular, can we identify outbreaks in locations before they spread to others?
Last week I asked someone whose research had been featured in this newsletter, what other research they would like to see me cover. They suggested this paper entitled “SARS-CoV-2 RNA concentrations in primary municipal sewage sludge as a leading indicator of COVID-19 outbreak dynamics,” the pre-print of which appeared in May. Outbreak dynamics? Leading indicator? Sewage sludge? Just try keeping me away from that topic!
Not to get too technical about this but a group of scientists decided to collect and test sewage in, what they listed in the abstract as, an unnamed “northeastern U.S. metropolitan area” so you totally couldn’t figure out where these scientists were getting their hands dirty. You’d think that it would be hard to find coronavirus RNA unless you really collected a ton of sewage but no. They found it all over the place and what’s more, when adjusting for time lags (what time lags I’m not sure), it was highly correlated with health data on Covid-19 cases and hospital admissions in what they admitted was New Haven and surrounding areas. (R squares of 0.99 if you must know even though we don’t look at that variable, do we?) Basically, you could know 3 days before hospital admissions and 7 days before testing data (yep, those are slower) if there was an outbreak. If this “northeastern U.S. metropolitan area” was representative of other areas — and it is hard to know because there are no comparative poo statistics provided in the paper — then this would be potentially a very useful tool for surveillance.
You’ll be glad (I think) to know that these scientists weren’t just going through the sewers because they had happened to be finally escaping from prison. Previous doctors had the idea of testing the stool of Covid-19 patients and found it was full of coronavirus. Those papers were published in March and so this team got right to it at the beginning of April. Possibly, an April Fools Joke escalating quickly. Who knows?Anyhow, the scientists were careful to point out at several opportunities that they “uniquely utilized primary sewage sludge instead of raw wastewater” because they were not going to let this slide.
If you are still reading by this point, I am sure you are wondering what precisely did they do? Well, it turns out they went out to the treatment plant every day for 6 weeks, collected samples and then froze them. I presume they have good labelling in the lab freezer. They also tell us quite freely that the treatment plant was in New Haven which, for someone who was highly intrigued by their coyness in the abstract, came as a disappointing revelation. But it is important to know that the New Haven Starbucks was closed by that time so they were doing this every day without coffee.
How much poo did they have to collect? After all, if you have test lots of it, that is kind of going to make it a less cost-effective surveillance measure. First of all, despite their lauding it over the raw wastewater folks at the other lab, “[s]ludge collected from ESWPAF is primary sludge, sampled at the outlet of a gravity thickener, ranging in solids content from 2.6% to 5%.” That’s pretty watery. But then they used “2.5 mL of well mixed sludge” to extract RNA from. On this, they conducted a standard PCR test on the sludge rather than the usual nasal stuff. (As an aside, the researchers thank the Huffman Family Donor Advised Fund for funding the research and if they want they can append this post to their annual report to them of their activities).
Regular readers will be wondering, how sensitive is the poo test? (Irregular readers now have an image of scientists poking poo which serves you right!) Sensitivity is the degree to which the presence of coronavirus in the population means that it can be detected in poo. As it turns out, it was quite sensitive with scientists picking up concentrations of the virus at a level that was at the threshold of what PCR tests can do. That means that in 1 in 10 of the poo samples you find generated by infected people, you’ll miss picking it up with this test. That said, when it comes to surveillance you can correct for sensitivity issues so it isn’t a big deal.
The Information Value of Poo
Having established that sewage can contain information and that you can actually collect and test poo way before (ahem) people, what use is that? It isn’t a leading indicator of the virus because the virus is already there. It is just a leading indicator of other information about the virus. That would put it in the category of better information that you can use to get a jump on things.
I’ve had an interest in this for some time although it only ended up being a footnote in my book. I wrote a post back in April about using Google trends to pick up whether people had specific Covid-19 symptoms like loss of taste or smell. I actually called it a smell test but my investigations didn’t involve me leaving the keyboard let alone smelling anything like our poo scientists did. My idea got picked up by the New York Times of all places with a closer analysis by Seth Stephens-Davidowitz. What he found was quite compelling.
These trends were available world-wide and are suggestive although I don’t know how they relate to sewage investigations in terms of being a leading indicator. The issue is whether people display symptoms before they go to the bathroom infected. I suspect they do not and the virus can be picked up before an unsuspecting person let’s one go.
As information, scientists get excited about sewage. For example,
“Everybody has to poop,” says Ryerson University professor Kimberly Gilbride. “It’s something that we can’t avoid.” … there can be no secrets in the sewer, since everybody contributes. “People can’t opt out of it, so it’s going to give us a really good idea of what’s happening in the communities,” she says.
A truer scientific fact has perhaps never been stated. But the key issue is where do you have to poop? For the most part, you have choices there. This is important because, if you were interested in monitoring an outbreak in a specific area — say, for instance, a school — you would have to worry about where and when people were going. Cast your net over a wide enough area though and you are fine.
But will that give you the information you need? Some scientists have doubts.
Another researcher who has been gathering samples and conducting early experiments in the UK is David Graham, professor of ecosystems engineering at Newcastle University. He says that only taking samples from wastewater treatment plants limits the information scientists can gather about what parts of a city are worst affected by covid-19.
“Say you’ve got the city of Newcastle, which has about 400 000 people, and there’s two waste treatment plants—that doesn’t tell you very much about the epidemiology of the disease within a city of that size,” he explains.
Since March, he and colleagues have been gathering samples from localised points instead: pump stations and junctions scattered across Newcastle’s sewer system. He says that they’ve gathered between 250 and 300 vials of wastewater and are now starting to analyse the first of these.
Personally, I will take anyone who has been collecting samples regularly very seriously as they aren’t doing this for fun. The problem is, that in this situation, you are going to worry more about people’s poo choices or to put it in terms economists will appreciate, the endogeneity of poop (which would be an excellent title for a paper).
Thus, to properly understand what this is telling us, we need to explore, as my econometrician friends constantly remind me, the poo generating process. Does poo change during a lockdown? Where you poo likely changes during a lockdown? Did the toilet paper shortage change people’s pooing behaviour? None, not one, of these issues was raised in the paper. I mean, if I could think of it sitting here writing, how could they not have raised these concerns during their daily treatment plant excursions.
But scientists are aware that you can’t just collect samples. To be to the heart of the case you have to go deeper.
There are, however, challenges involved in wastewater monitoring. The proportion of the virus present in wastewater can change depending on the amount of rainwater entering the sewage system, for instance. Or, temperature and other factors could alter the amount of viral material that survives in one city’s sewers compared with another’s. Wastewater from areas with more hospitals or care homes might contain greater amounts of viral RNA. Plus, things can change over time. As people move out of lockdown, the use of residential bathrooms will likely fall, again altering what turns up at the treatment plant.
In order to be able to take samples from a wastewater facility and make a reliable estimate as to how many people are infected in that area, all of the above factors need to be taken into account, says Andrew Singer, senior scientist at the UK Centre for Ecology and Hydrology.
Researchers also need to control for the time of day at which a sample was taken, since, as with the electricity grid, demand on sewers varies between morning and night. Current methods of purification mean significant proportions of viral material—up to 80%—can be lost from samples so this needs to be considered too, says Singer. Corbishley adds that he has tried to solve this by inserting a known quantity of pig virus into his samples to see how much is left by the time the specimen is purified and ready for analysis.
“It’s like detective work,” says Singer.
Suffice it to say, if you thought the scientists were done with this topic with just a 12-page paper, they have only just got started. And I wholeheartedly agree.
The Bigger Picture
There is actually a bigger and more significant public health picture in all of this. Covid-19 aside, monitoring of waste and its properties is an opportunity to explore changing patterns of all manner of infectious diseases and other indicators of population health. Automating it and creating a platform for monitoring and monitoring for new things as they emerge seems to be something that should not simply by done in the midst of a crisis. This is the only way to understand properly the information it is giving us. This is an opportunity that we shouldn’t waste.