Modelling spatio temporal COVID-19 trends through wastewater surveillance
This is a modal window.
The media could not be loaded, either because the server or network failed or because the format is not supported.
Formal Metadata
| Title |
| |
| Title of Series | ||
| Number of Parts | 45 | |
| Author | ||
| Contributors | ||
| License | CC Attribution 3.0 Germany: You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor. | |
| Identifiers | 10.5446/55549 (DOI) | |
| Publisher | ||
| Release Date | ||
| Language | ||
| Producer | ||
| Production Year | 2021 | |
| Production Place | Wageningen |
Content Metadata
| Subject Area | ||
| Genre | ||
| Abstract |
| |
| Keywords |
MOOD Science Webinars12 / 45
9
13
24
25
26
35
36
38
40
42
43
44
00:00
MathematicsProjective plane
00:31
Axiom of choiceSample (statistics)EstimationFaktorenanalyseMassStructural loadMathematical analysisMortality rateStatistical hypothesis testingStatistical hypothesis testingGoogolStatisticsSampling (statistics)Projective planeMultiplication signTerm (mathematics)Model theoryLatent heatGroup actionDifferent (Kate Ryan album)MeasurementAreaVotingMany-sorted logicKörper <Algebra>MassPhysical systemTrailCuboidFigurate numberPoint (geometry)AverageEnergy levelMereologyComputer animation
07:02
Model theoryMaß <Mathematik>Group actionFunction (mathematics)Scale (map)Boundary value problemThermodynamisches SystemImage resolutionAreaMany-sorted logicNumerical analysisStandard deviationTheory of relativityMeasurementStatistical hypothesis testingHypothesisKörper <Algebra>PolygonPartition (number theory)Maß <Mathematik>SpacetimeLine (geometry)Graph coloringSlide ruleTotal S.A.Model theoryOrder (biology)MereologyFunction (mathematics)SubsetBoundary value problemFigurate numberLinear regressionWeight functionArithmetic meanGroup actionCurve fittingRandomizationStatistical hypothesis testingStandard errorMathematicsComputer animation
15:56
Thermodynamisches SystemBoundary value problemDatabase normalizationSmith chartWitt algebraProjective planeUniverse (mathematics)Many-sorted logicTerm (mathematics)Line (geometry)Energy levelLink (knot theory)Physical systemStructural loadNumerical analysisCountingComputabilityAverageLogical constantScaling (geometry)Functional (mathematics)Multiplication signMeasurementSampling (statistics)Theory of relativityOrder (biology)Point (geometry)SpacetimeMoment (mathematics)AreaRight angleBasis <Mathematik>Körper <Algebra>Different (Kate Ryan album)Bounded variationBuildingImage resolutionStatistical hypothesis testingStudent's t-testComputer animation
24:50
Database normalizationDifferent (Kate Ryan album)Sampling (statistics)Matching (graph theory)Link (knot theory)MeasurementUniverse (mathematics)Thermal fluctuationsPhysical systemMortality rateConsistencyMathematicsStatistical hypothesis testingStatistical hypothesis testingTheory of relativityAdditionMultiplication signBuildingModel theoryAreaVariable (mathematics)Grothendieck topologyResultantNumerical analysisObject (grammar)Mass flow rateEnergy levelMany-sorted logicGroup actionStudent's t-testCivil engineeringNoise (electronics)Meeting/Interview
Transcript: English(auto-generated)
00:10
So yes, I'm Theresa Smith. I'm from the University of Bath, which is the campus that you can see on your screen there. And I'm going to talk about a project that I've involved in,
00:20
which is looking at tracking the spread of COVID-19 by looking at wastewater. It's just a bit of an overview of this talk. So I'm not sure how familiar the general crowd is with the idea of tracking population health through wastewater. So I'll give a bit of a brief intro to that
00:41
and then talk about why it's of interest for COVID-19 specifically. And then I'm going to spend some time talking about the problem within my project that I'm most interested in, which is related a bit to handling spatial data in R. And then at the end, talk about some further challenges
01:01
about trying to model the spread of COVID-19 through wastewater. Okay, so wastewater-based epidemiology, which I'll abbreviate WBE, is a field that's been around way before COVID. And the idea of wastewater-based epidemiology is that you can measure the health of a whole community
01:22
through one sample. So people live in communities and what we consume, what's affecting us gets excreted into the sewer system. So we're all screening little compounds or bits of DNA or whatever that can say something
01:42
about what we've been doing that day or what might be affecting us that day. So what pharmaceuticals we've been taking, whether we've been drinking coffee because caffeine gets excreted or metabolites of caffeine, whether we're smokers, but also things like little snippets of DNA or RNA from pathogens that might be infecting us. So we excrete those,
02:01
they go into the sewer system en masse. So it collects up and pooled through the whole community that is contributing to that wastewater treatment plant. You can then grab a sample, and there's different ways to do the sampling, but you get a sample, take it to a lab, and then you analyze the constituents of that sample from the wastewater treatment plant, and then try to calculate whatever you're interested in
02:23
about human health. So as I've mentioned a few examples, so how many people are drinking caffeine, that sort of thing, maybe that's not the most interesting thing, but they have been using this to, say, track illicit drug uses, like looking for cocaine in the wastewater, things like looking for antimicrobial resistance,
02:42
so looking for the snippets of DNA that are related to resistant genes and bacteria. So this sort of system and pathway and pipeline was built up before COVID-19. And this is a picture that's from a recent paper of some of the people at the University of Bath team.
03:04
Okay, so now it's coming, now wastewater-based epidemiology is coming into its own for COVID-19. So I think probably one of the places where this is getting the most attention is in Australia. So this is a picture of Melbourne, and it's a screenshot of a dashboard from Melbourne, and then the areas around Melbourne.
03:22
And Australia is using wastewater-based epidemiology in the sort of what I would call qualitative sense. So they're doing presence, absence. So these orange exclamation points are whether there was a hit, basically, and then it's kind of hit and miss. So is there COVID, isn't there COVID in,
03:40
it measured in a particular, either a pumping station or a big treatment plant. And this makes sense in Australia because their cases are low. So hit and miss is something very useful to them. So this is an example of a dashboard that you could go look at today to see where are their hits and misses in terms of finding those little bits of DNA for COVID-19
04:02
in and around Melbourne. It's not just Australia. It's not just the UK where I work. It's really all over the place. So as an example, if you look on Google Scholar for wastewater and COVID-19, this claims there's 23,000 papers,
04:20
which over a year, it's only been going on for a year and a half, two years. So that's a lot of papers to be produced about this. I have not checked that every one of these 23,000 papers is actually about COVID-19 and modeling COVID-19 for the wastewater, but I did look scheme through them, but at least hundreds and hundreds of them. There's hundreds and hundreds of papers trying to accomplish this task based on,
04:41
so trying to accomplish the task of being able to figure out where the COVID is based on these pooled samples from the wastewater. And if you're really keen, and there's a whole Twitter community around this, so this aptly named COVID Poops Twitter handle, and they'll tweet the latest papers and retweet different people who are working on this.
05:00
So if you are keen after this talk, go follow one of the latest, greatest developments, follow the COVID Poops Twitter. But you can see also in the background that they collect, they have this dashboard on their website of where all the groups working on it. So there's lots of groups working on wastewater-based epidemiology for COVID-19 in Europe
05:20
and in the States in Canada, as well as populated areas of Australia, but then there's groups all over the place as well for people who are trying to do this. This is an example from the UK that is probably closest, it's not the data group that I work on,
05:40
but it's publicly available data from the government where they, for all of the wastewater treatment plants, they then measure not something qualitative, like in Australia, they measure something quantitative. So they're measuring the gene copies per liter for a particular snippet of DNA in the measurements that they get from the wastewater treatment plants. And then these are average,
06:01
these are weekly averages from June. And this box here in yellow is roughly the region, the project that I'm working on and studying. So a couple of observations about this figure. As I said, they're doing something qualitative or they're doing something quantitative, not qualitative.
06:21
So it's not like the Melbourne figure where it was kind of on, off. It's trying to measure a quantity in the wastewater here. So there could be a little bit of COVID in the wastewater or a lot of COVID in the wastewater. And everything in between. Another thing that I noticed about this picture is that the wastewater treatment plants are represented by just points basically.
06:43
I mean, they're dots big enough so that you can see the data that's been collected from there, but they're basically point level data here. And they're represented as point level data. And that's part of what I'm going to talk about today. But the main idea here
07:03
is that we can use these gene copies per liter that was being depicted on the previous slide to tell us about how many people are infected in the catchment, which is the area that contributes to the wastewater treatment plant. So the idea is that if, say,
07:22
people kind of excrete roughly the same amount of these gene snippets when they're ill with COVID-19, then there should be a rough proportionality between the number of people who are infected in a catchment area and the number of gene copies that you're measuring in the wastewater treatment plant.
07:40
I don't know what this constant of proportionality is, but the hypothesis is that maybe we can find some, that there should be some sort of relative amount that tells us if the wastewater measurements increase by this much, then we expect cases to kind of increase by that, increasing by that same amount. So the aspiration then of this field
08:01
is that we can develop a model to relate gene copies per liter this week to the future number of infections. So I'm finding it not mathematically at all. I want some model F. It really could be anything that you want. I've seen everything from random forest, which is really simple linear regressions to really complicated machine learning things
08:22
where we try to relate the number of infected next week to the gene copies per liter this week. It doesn't have to be weak. I've written it as weak, but number of gene copies per liter at the wastewater treatment plant today, what does that tell us about the number of people who are infected tomorrow? And part of the idea here is that you start excreting
08:44
these little snippets of DNA from the SARS-CoV-2 virus into the wastewater at least a couple of days before you would go and get a test yourself. So ideally this would be picking up changes in case as well before you would see an increase in cases
09:04
from the rest of the usual public health testing like people going in and seeking a test because they're feeling like they have symptoms. So someone will start putting SARS-CoV-2 genes into the wastewater before they would go and get a test. So you might see expect that the wastewater will go up
09:23
before you would see cases go up. Okay, so what ingredients do we need to train a model like this? Well, we need a number of people who are infected in the catchment area in order to train a model like this. So I showed you on the previous slide that we have measurements of the number,
09:42
we have measurements of the gene copies per liter in each wastewater treatment plan or a subset of the wastewater treatment plants in England. But how do we then relate those figures to the number of infected? Well, we need to figure out how many people are actually infected in the wastewater treatment area. That's a lot easier said than done.
10:00
And part of it is, well, a big part of it that I'm interested in is because wastewater treatment catchments and public health reporting units, whatever they might be in your area, are spatially misaligned. Meaning the geographic area that contains people who will contribute to a particular wastewater treatment plant is not the same as the geospatial polygons that the COVID data,
10:24
the clinical data is being reported on. And these catchments can be really funky. So there's like no hope of them, there's never hope of them being spatially aligned. So these are the wastewater treatment plant catchments
10:40
in London. And one, they're not a partition. So the public health reporting units are usually a partition, well, almost always a partition of your space. So they fully over, the unit of them is the whole area. So these aren't a partition, there are gaps which are probably things like,
11:01
at least in the center of London, these are probably things like parks. So nobody lives there anyway. Some, so the wastewater catchment areas are in color and then the gray lines here are the boundaries of the boroughs in London. Some of these boroughs are completely covered by one wastewater treatment catchment.
11:22
So everyone who lives in this borough is their sewer system is the green one. But then there are others where there's some misalignment. So some people in that borough will be going to one wastewater treatment plant and some would be going to another wastewater treatment plant.
11:44
So what did we do about this? Well, we've done something really not particularly complicated. So in our area in the Southwest of England, where I am, so Bath Bristol area, the smallest publicly available data
12:01
for clinical cases, COVID-19, is called the Middle Layer Super Output Area, MSOA. You can get the shape files for this from the Office of National Statistics and they're not enormous. They're sort of 5,000, 10,000 individuals in these areas. So they're not the highest resolution
12:21
area we have in the UK, but they're the highest resolution where they'll release the COVID data. And from these, we can get weekly confirmed cases. So let's let weekly confirmed cases in an MSOA be Y and then subscript W for week and M for MSOA. From the wastewater companies,
12:41
you can also get the catchment files. So these are not publicly available data, but through our relationships with the wastewater company, we know where the catchment boundaries are. Those funky shapes, we know what they are. And from this, you can calculate using pretty standard geospatial tools and are the percentage overlap between the MSOA
13:05
and the catchment. So I'm calling this percentage overlap PMC. And then we're just doing a weighted sum basically. So we sum up, if we want to know the number of cases in the catchment area, in the catchment area C and week W, then we take the original data that we have,
13:23
and then we would do a weighted sum based on the percentage of spatial overlap between the catchments and our MSOAs, our public health reporting units. Okay, so there's a problem with this,
13:41
which is it works really well in cities, at least the city that where I am sat, where most of the people are, the MSOAs are fully contained in the catchment area. So here we've got about a dozen MSOAs. So this is the vast majority of the population and they're pretty much 100% contained in the catchment area.
14:01
So if you live in one of these, your wastewater is going to the treatment plant where then I have data on SARS-CoV-2. But then there's a few more outskirts. There's some outskirts where there's a small overlap with the wastewater treatment plant catchment, but I'm not too worried about them because the majority of the data anyway
14:21
is coming from this higher density area. So remember these MSOAs, they're roughly supposed to contain the same number of populations. So the fact that this is massive means it's rural. And the fact that these are small means it's more urban. Now, if I go over here to this another area where we have some wastewater treatment data,
14:40
this is from more of a town than a city, sort of semi-rural town. We have an area where we have these four MSOAs, which is a little bit denser, but even then, none of them are 100% contained in my wastewater treatment plant catchment. And then we have more of these outskirts areas where there's a small percentage overlap. So it's not working as well there. And probably the kind of mismatches here,
15:02
this assumption, this proportionality assumption that I've made, which is pretty simplistic, the simplistic assumption maybe having a bigger effect fitting a bigger error in the total number of cases per catchment than it is over here. So this matters in part because these catchment boundaries are not arbitrary.
15:21
I can show you the catchments with the area that I work in, but this is a picture from Sweden where they have their own wastewater treatment dashboards. You can go have a look at it, but here's a picture of one of their catchments, which they do release, which is super funny looking, right? But it's funny looking for a reason. So you can see underneath here, these brown dots,
15:41
these are houses, these are building structures. So some of these weird little spindly arms, they're going out to grab a community. So if I just randomly put down points, I might not hit, I might not,
16:02
if I guess what I'm trying to say is that there's a higher chance that the catchment includes a settlement than the catchment includes some piece of land where there's nobody living there, because that's the point of it. The point of these wastewater, these sewage systems
16:21
is to go up where the houses are. So this percentage overlap of the spatial area is not really the right thing to do. So I don't want percentage overlap of the geography. I want percentage overlap of the population. I want to know the percentage of my population that is living inside my catchment area. I don't want to know the percentage of space
16:42
that's inside my catchment area, because that will probably underestimate the population because these catchments are meant to go out and grab the people. So what could we do instead? One thing that we're considering to improve this is looking at high resolution sort of remote sensing data, which can tell you where the people are, or probably won't do this ourselves
17:02
because there's loads of other people that have thought really intelligently about how you go from some remote sensing data to then getting population counts on a really, really small spatial scale. So this is a picture from University of Southampton's world pop project where they have population
17:22
that really, at least for my purposes, really tiny grids. So using these alongside our MSOA overlaps and the catchment I think is our next plan so that we can move from this geographic overlap idea more towards the population overlap idea. Okay, so that's one piece of the puzzle
17:40
and it's a piece of puzzle that I enjoy because as Tom said, I like kind of geospatial computing and I like doing these sorts of things, but there's many, many other pieces of the puzzle which we need to resolve before we can realistically do this number of infected is some model, number of infected next week is some function of the number of, the amount of gene copies
18:01
per liter this week. So, so far, this is showing data from a city in, this is publicly available data showing a city in California and then the purple is the West, the whole averages of gene copies per liter and then cases in the West. And the thing is, it looks like the peaks match the peaks and the troughs match the troughs,
18:21
but not always to the same relative amount, which is the problem. So it's not actually relative, it's not this assumption that I had that there's like kind of constant relative proportion that a person sort of chucks out the same amount of COVID-19 into the wastewater if they're ill is doesn't seem to be true, at least not the way that we've been able to measure it. So a peak matches a peak,
18:41
so maybe you can do some kind of a slope matches a slope, but I can't say that if I have this amount of SARS-CoV-2 in the wastewater, I expect there are this many cases in my community, even within this, even within one wastewater treatment plant, that doesn't seem to be consistently true. So there are many other challenges for trying to figure out how to model things
19:03
which match the peak and whether we can translate that into properly quantitative information, or we can really only do this qualitative, it's going up, it's going down, it's at a peak, it's not at a peak kind of thing. So just to wrap up, so this, as I said, I told you about one small piece
19:21
and there's loads of other pieces, so it's a pretty big team, so I've given you some of the team members here, the PI of the project is Barbara Casperse Cordon, who's been working on this wastewater-based epidemiology for a long time. And then I'd like to thank the funders of this project and you for listening.
19:42
Thank you, Teresa. Thank you so much. That was a really, really more as an introduction to the problem than a solution, I guess. So the floor is open for questions. So please, if you have a question, you can pause to chat
20:00
or you can ask directly, Teresa, but the floor is open for discussion. Maybe I can start, I can start. Oh yes, there's a question, please. Sorry, Teresa, nice talk. Thank you very much.
20:21
I wonder if you've come across something called the Toilet Board. It's something I got approached by about a year ago to look at this sort of thing, but actually not in the UK, but in India. And they've got some quite cute projects along these sorts of lines for diseases. I think COVID as well as various other diseases,
20:41
they put sensors in toilets actually, so it's a bit more immediate in terms of spatial relatedness. It might, I mean, you might, they might have something of interest to you. They're based in Geneva. They're a real thing, I promise you. I thought it was a joke. They've got quite a decent budget,
21:00
you know, several hundred million a year, the sort of UN levels. So it might be worth you having a look at and see what they do. Can you share a link to that, please? I can try. If you can share through chat. Oh yeah. It will be more interesting to look at that. Yeah, so I haven't heard of that specifically, but I do know that,
21:21
so I've talked about community level, community level projects basically in this work, but there are, I do know of, and I'm sort of intentionally involved in one project that is building level, let's say. So you can go basically pick up the manhole cover
21:43
outside a building and pick up a sample that is then only gathering data from the people who are in that building. So they've done this on some university campuses in the States and then some other places. I think not necessarily to inform any policy making
22:03
about that, any procedures that they would do on that campus, but more because you get the clearer data. So there are some examples from the US like this time last year where they were requiring all of their students to get tested before they returned to campus. And then they know the addresses of those students and then that they can test individual buildings
22:21
and it's much easier to align the data. So you don't have, well, you have much less of an alignment problem. Yes, I think these were the toilet board stuff was, the Indians are building various smart cities. They're building half a dozen of them for 250,000 people each, I think. So they're really putting some serious effort into it.
22:41
And they wanted to build in a disease monitoring system. And this was it, I can't remember offhand what diseases they were interested in, I think it was TB. Yeah, so I started a project that was on this sort of stuff just before COVID started. And so TB is something that people are interested. So I have a GCRF here
23:00
because that's one of the funders of kind of international research in the UK. So I have some collaborators in South Africa and Nigeria and we were also sort of wanting to have kind of autonomous sensors that can move around and do some sensing kind of like right on this piece of kit that's motoring around rather than having to take a sample,
23:20
take it to the lab and that sort of thing. I think for COVID, that's not at the moment possible to do that kind of autonomous sampling, but yeah, I think that is what people are, one of the things that people are quite interested. Yeah, it looks like it's a field that's taking off. And Teresa, tell us the measurements in the wastewater
23:40
in UK, for example, this is like on a daily basis or? It's a couple of times a week. So the project that I'm involved in, it's mostly, it's twice a week. We are also doing occasionally some sampling every day of the week so that we can get a sense of what are the diurnal variations
24:01
and like if there are variations in the week, like in some of these, like in London, for example, you might see something different on Saturday, Sunday and like Monday through Friday, just because there's more people in London, Monday through Friday who work there, but then don't live there. So you have these, this is also a whole field of trying to figure out
24:21
how many people are actually, just try to understand this population even by measuring things inside the wastewater. So you need many samples a week in order to do that. But I think the typical thing is a couple of samples a week. And like in Netherlands, I mean, they do the same thing. And I have a feeling when I look at the plots for Netherlands,
24:41
Netherlands, I mean, it's like a more, the catchment's even more complicated because really flat area, right? But I was looking at the measurements for the wastewater in Netherlands. And to me, it looks like it's more useful to look at these numbers from the wastewater than for the results of the test.
25:01
Because I have a feeling that they reflect much better what is the actual infection. Because the infection rates you get from test, they depend on number of tests. So it's up to people to get tested. But this is kind of objective. This is like, it's always, because if there are fixed locations
25:20
and if they do the measurements, then it's kind of objective, right? Yeah, so like in this picture where we're seeing a difference between, we're seeing peaks match peaks, but we're seeing that the sizes don't match up quite right in this figure. And one explanation for that could be changes in testing, for example, that maybe the overall test could go down,
25:43
but you still perhaps expect to see a peak. And that is a challenge here that you wouldn't expect because of changes in the way people get tested. You wouldn't expect necessarily this to be a consistent relationship over time as well. In addition to the fact that it was never exactly this relation in the first place
26:00
because you could only model with the data that we have. So I think exactly fitting this kind of relation is not realistic because we're never gonna know how many people were truly infected unless we're in these really small situations like that example I gave with the university buildings and things like that where students were required to get tested.
26:23
And you try fitting models to this data, right? You have fitted models to try to map the infection rates, I assume? Yeah, I think some of the things that are challenging is just the measurements of the virus are really heterogeneous themselves.
26:41
And I haven't talked about that at all. I'll post a paper about that if you're interested in that. But you can see here that there's like a lot of noise in some of these measurements. So it's hard to apportion kind of variability to the measures themselves or like to changes in the actual rates of COVID.
27:05
So we stuck with fairly simplistic things like sort of like lagged linear or generalized linear models, that kind of thing. Okay, let's see if there's more questions for Teresa, please. There is a question in the chat from Alan. Is the level of virus in wastewater affected by rainfall?
27:25
Yes, it is. So in the UK and I think the East Coast of the US, for example, our sewer systems are old. So the rainwater runoff goes to the same place as sewage
27:42
like sewage from our toilets and from our houses. So you can try to normalize by flow. So you know the wastewater treatment plants, they measure how much water is coming in every day. And you don't necessarily need to directly account for rainfall because you can sort of implicitly account
28:03
for it by accounting for the amount of flow. So if you have more flow through your wastewater system then the contributions from humans are diluted. So you can try to normalize by flow. In other places, it's less of an issue. So like in places that have sewer systems that were built this century, it's less of a problem
28:22
because the rainwater doesn't dilute the contribution of humans as much. You still can get some infiltration. So there's not like there's zero contribution to the wastewater treatment plants from rainfall but most modern sewer system, I'm not a civil engineer. So I think this is true, but they're segregated.
28:41
So you get the rainwater runoff going one place and then you get human contributions going another place. But it is something to think about normalizing for that most groups think about normalizing for flow to take into account the effect of rainwater.
29:02
I think for wastewater to be wastewater-based epidemiology for COVID-19 to be feasible, we need all of the other programs. You can't build and train the systems without having those original things in place. A really interesting question I think is how much money do we have to spend on the classical stuff?
29:22
The classical, everyone who feels unwell going to get tested, how much time, how much money do we need to be investing in that first so we can build a system before we can transition to, before we can really use wastewater-based epidemiology as a standalone tool for monitoring COVID-19.
29:43
There have been some cases in the UK where it was really an interesting adjunct to the classical monitoring, which had to do with variants. So in the UK or in the wintertime, if they thought that there was an uptick
30:01
in a particular variant, or they thought that there was a region that had a variant, then they would do testing for everyone in that community. So if they thought there was an area that had a variant and they would tell everyone in that community to go and get a test, if you were feeling well or unwell, they called it surge testing. And there were a couple of times where they did the surge testing because the wastewater pined a particular variant
30:21
that they were worried about. So that's one way where even already it has helped, but not in the, again, the more of a qualitative sense. We think that there is a particular variant in this borough. I think it was actually more than in the borough of London. And so we're going to go do surge testing there. But I think it's a few steps away
30:40
from being able to actually say, we don't need the case data anymore. You definitely need the case data to train the wastewater models. Definitely the time it takes for the sample to go to the lab and then the sample to be processed in the lab. And it seems like the window
31:01
between when people are shedding the virus into the wastewater and getting it, and then having symptoms and getting the test is actually really small. It's maybe a couple of days. So this time lag of how long it takes to go from the wastewater treatment site to the lab
31:22
is actually taking up a good chunk of that time. So this is one of the challenges is how, if it is to be a useful surveillance system, how early are the signal can we get in? It may actually be pretty, only a little bit ahead of the signal that you could get from the classical system. So just to give a little bit more detail,
31:41
this is the sample comes to the lab and then they have to do two steps. First, they have to extract all the particles of RNA from, or try to grab onto all the particles of RNA in that sample. And then they have to go to, I don't know the technical term, but the PCR machine, let's say, which then tells you how much is in there.
32:02
So you have to grab onto it and then you have to take it and figure out how much is actually in there. So it's two steps, each of which is pretty complicated and adds its own variability to the system. And at the moment, there's no real consensus, especially on the first step of how you should extract the DNA, sorry, in this case, RNA directly from the wastewater.
32:22
It's not my area of expertise, but sitting on the outside of like lab meetings, there's lots of talk about what's the best way to do that and differences even within the UK about how different labs are doing that. I think in the UK, the answer to that is honestly, no, I don't think that they have used
32:43
the wastewater data to decide about lockdowns in the UK, but they are using, so I guess from my research, that's a no, but they are using that kind of information in places where case rates are lower. So I think the reason that they haven't been using it in the UK is that most of the measurements, they always have COVID in them
33:03
because our cases have never really fallen too far down towards zero in the UK. And it's not so clear yet how you relate the fluctuations in the measurements in the wastewater to fluctuations in people. That's still, that link is still unclear, but there are places in the world where they have been using it where the cases are low.
33:22
So the presence of SARS-CoV-2 RNA particles in the wastewater is a really interesting signal and something that's actionable. And, but that's not quite true in the UK because of the two things. So the cases, I've never really gotten that low, unfortunately, and we're not, we haven't sorted out how to link quantitatively measures
33:41
in the wastewater to people.
Recommendations
Series of 4 media