Below the Radar Transcript
Episode 117: Charting the Pandemic with Data Modelling 鈥� with Caroline Colijn
Speakers: Melissa Roach, Am Johal, Caroline Colijn
[theme music]
Melissa Roach 0:01
Hi, I'm Melissa Roach with Below the Radar. A knowledge democracy podcast. Below the Radar is recorded on the territories of the Musqueam, Squamish, and Tsleil-Waututh peoples. This time on Below the Radar, our host, Am Johal, is joined by 丁香园AV mathematics professor Caroline Coljin, who's been working throughout the pandemic to model the spread and evolution of COVID-19. I hope you've enjoyed the conversation.
[theme music fades]
Am Johal 0:27
Thank you so much for joining us on Below the Radar again, we are very lucky to have Caroline Colijn with us, 丁香园AV professor, welcome, Caroline.
Caroline Coljin 0:38
Hello.
Am Johal 0:39
Caroline, I wonder if we can just begin by you introducing yourself a little bit.
Caroline Coljin 0:43
Absolutely. So, I'm Caroline Coljin. I'm a professor and in the Department of Mathematics at 丁香园AV and been working on pandemic modeling since the start of the pandemic, when I organized a hackathon back in February 2020. Just to get my research group and 丁香园AV's community familiar with some of the data that were coming out early on in the pandemic. More broadly than that, my research program is an infectious disease and in particular, diversifying or diverse pathogens and how to model them mathematically and how to get the most out of out of data for those pathogens and understand how their populations spread and change.
Am Johal 1:24
Caroline, how did you get into this area of work? Obviously, it requires like hard mathematics, but in the epidemiology area, how did you find yourself in this zone of work? How did you come into it?
Caroline Coljin 1:38
Yeah, so I actually did a PhD on foundations of quantum mechanics, which is quite distinct from this field. And then at the postdoc stage, I wanted to move into something more applied in terms of data. And I did some dynamical modeling at McGill around diseases of the blood that were cyclical, so populations of blood cells. And then I did postdoctoral work at Harvard and MIT where I worked in a TB group, tuberculosis epidemiology research group with Professor Meghan Murray at Harvard. And that was really where I got interested in infectious disease. And in particular, the relationships and interactions between drug resistant forms of disease and drug sensitive forms of a pathogen. And there's some really interesting and nice and really important questions around you know, what happens when we use antibiotics? Do we drive antibiotic resistance? Or do we just have a benefit on treating antibiotic sensitive bacteria. So that was when I got interested in infectious disease. And it's a great field for me because it does link mathematics and statistics and data, but it also connects with policy with public health, with people with biology. And so, it's a great field, if you're interested in a lot of different things and fitting them together into a coherent picture.
Am Johal 2:56
Global pandemics are so fascinating, and we've had global era health crises before but if something of this magnitude and size, you almost have to go back to the Spanish flu pandemic. It was quite natural when the first wave hit to go back to popular accounts. And you see how epidemiology was looked at after the fact and people trying to piece the story back together. And stories of the second wave are taught in epidemiology. And there's some famous case studies from that era. But I'm wondering if we can sort of begin the conversation with sort of the first wave, and a number of you researchers getting together to start doing this work. And I'm wondering, looking back at that first wave now, what was the kind of mode of response you were sort of placed into, and having to take all of these things you had learned and actually putting it into deep practice?
Caroline Coljin 3:53
Yeah, in a way, it was an exciting time. And it did start from this hackathon event where we identified the somatic areas for six teams, and actually, they turned out to be quite relevant. One was around borders and flights, and could we figure out, you know, what's the daily risk of bringing an infection into BC? How many flights are there? Where are people coming from? What do we know, right now today about infection levels, and we didn't know a lot, for example, before Iran was known to be an affected area. People were detecting COVID-19 in people who had been to Iran. And so, you know, modeling flights from Iran if they weren't reporting cases. That was one area, and another was just in estimating how effective social distancing was, that grew out a little bit of setting up some models from that hackathon. So, there were different themes that were that were already emerging in February of 2020 that were going to be important.
Caroline Coljin 4:48
Another was kind of risk within a population. How much contact did people have, how much risk are they at. Another was in estimating the infection fatality rate or the case fatality rate. And that's of course incredibly important because we know now how steeply that rises with age. So, there was some early data on that. So, we started working in partnership with , at that stage, and we got some funding from the . And we built up a modeling team that had people from UBC, people from my group at 丁香园AV, a couple people who were very talented modelers, and programmers from the Department of Fisheries and Oceans, from the Ministry of Health and from BC Children's at the time. And we built up a team to develop different kinds of models to answer questions that were important to BC CDC and to the provincial control. And we also started participating in the broader conversation, of course, about COVID-19, and how it was spreading and social distancing and those topics.
Am Johal 5:48
And of course, during this time, as people watch the pandemic timeframe, move along, and press conferences every day, getting a sense of what was going on globally, it's like everyone became an armchair epidemiologist, so people were weighing in with popular accounts of what was going on. And now looking back on the strategies that different nation states followed from Taiwan to South Korea to Japan to New Zealand and Australia to Sweden, you really get a sense of how it landed down in different places by their strategies with it. But are there any sort of observation you have from that first wave in terms of how countries approached it in terms of looking at it from a mathematical or modeling point of view, if there's any observations you have?
Caroline Coljin 6:37
Yeah, I think one thing that I've learned in hindsight is really about border measures, and the importance that they can play. There was at the time, a meta-analysis, worked by the WHO suggesting that in a pandemic, you know, border measures would really only delay things and they hadn't suggested closures of borders as a major tool at that time. But in the meantime, of course, places like New Zealand and Taiwan and even Atlantic Canada have really found that border measures have made a huge difference. And you know, there are economies that have been largely reopened, supported by, you know, border measures, and also very rapid responses when if they do have introductions, so keeping those introductions low, but then really being able to jump up and down all over them if they do happen. And that's probably been more effective than people might have anticipated at that early stage. And more worth it in the sense of maybe we didn't know quite the level of economic disruption that it was going to take and for how long.
Caroline Coljin 7:37
One of the things that got me involved with this actually was I had seen the modeling out but from Imperial College London, that was a model that's very rich in detail and had many 10s of 1000s of lines of code, and people were thinking about it, but its overall behavior was very similar to simpler models. And so, I had developed this quite simple model of social distancing. And it looked kind of just like this more complicated model. So, I think I've put it out as a tweet or a blog post or something. And then I ended up with , at the Globe and Mail, to put out a little explainer for social distancing and "Flatten the Curve". So, what did flattening the curve look like? And what's amazing is if you put that in context, now, we didn't flatten that curve, we actually prevented that curve from happening. So when we talk about the first wave, all we really mean is, cases went up, and then we did something and they went down, it's more of the first time we cranked up the fire hoses, then really a wave in the sense of influenza, where the influenza, seasonal wave, where it kind of goes through the population, people get immunity, so many people get immunity that cases turn around. This is not an immunity driven wave in any way. And I think that was one of the framing of waves even rather than framing it, as you know, this is an ongoing fire. And if we don't put it out, we have to keep putting water on it the whole time for a year. And that came from even that flattening the curve. But I think that's something that we saw happening then and that kind of is a lesson learned throughout that these waves are not waves that come and go like a wave on the beach, that you know that wave on the beach doesn't recede because you jump up off your beach towel and push it back away from the sand. It comes and it goes. This is more like a forest fire that we are not putting out. And when we cease our efforts, it rises again. And that's what we've seen throughout the West, time and time again, as we have to reintroduce distancing measures in the face of rising transmission.
Am Johal 9:42
And that tension between stronger measures and more moderate measures seems to have, that tension continues to play out and in so rather than call it a second wave, I'll call it the second expansion of the fire, let's say in the BC context which, you know, roughly around mid October to late January, February, let's say, we walked into a context where, you know, in other parts of the country or in other parts globally, there were certainly timelines that showed where things could plausibly go here. I'm wondering from your research point of view, in terms of how we looked at the Fall, how we took measures, what your observations are of that that period?
Caroline Coljin 10:30
Yeah. So, it is unfortunate that, you know, hindsight is 2020, but in the early Summer, we had 10 or 11 cases a day. We, at that point, could have taken an Atlantic Canada route, we could have had a Pacific Canada or Mountain bubble, if we had wanted to, and we didn't, because we proceeded cases were very low, we proceeded to reopen, we move to phase three. And actually, exponential growth started early on, and we could see it in the models, because exponential growth doubling from 10 to 20, and then 20 to 40, and then 40 to 80 is still exponential, when it's doubling at a consistent time. It's just that when 80 becomes 160 to 300, over 300, and then over 600, that's when it starts to look really scary. But that slow growth was already starting in in the summer. And I think at that point, we did kind of lose an opportunity to go the New Zealand path, maybe it wouldn't ever have been practical here with the US border, which BC doesn't have jurisdiction over, you know what's happening, we didn't have the rapid tests yet. So, we couldn't have said, for example, if you want to step across this border, you have to do this rapid test, and then another confirming one. So maybe it wouldn't have worked.
Caroline Coljin 11:45
But I think we did see that rise. And it was clear that with COVID around, if we reopened and established a lot of contact, since nothing had changed, since the wave hadn't receded because of immunity and we knew that nothing would have changed and it would still rise into the Fall. And so, I think that's something people should have been more aware of, then they were, thinking, "Oh, it's over, we had this pandemic, and now it's gone." But nothing had changed. And so, and in fact, now we know from viral sequence data that we were still getting a lot of importations into Canada across international and the US border, throughout. And so, we were not keeping COVID out entirely. And we, you know, as we reopened without doing that, we knew there would be transmission.
Am Johal 12:31
And certainly, in the Spanish flu pandemic, which is studied by epidemiologists, a very different virus and pandemic, really was the second wave that caused the most damage in many ways. Now fast forward to the present moment of increases, it seemed to be that only a month ago, there was really positive news about the rollout of vaccines, there was positive news that people could gather outside your homes and there seems to be this kind of perfect storm of positive news with the vaccine and other pieces, the warmer weather perhaps coming as well. And then the variants of concern both the UK version, but particularly the Brazilian one, which has struck at a younger demographic, and with the age phased rollout of vaccines, we seem to be heading into a moment of big increases over the next couple of months. And I'm wondering if you can speak a little bit to your data modelling and kind of what you've been seeing happening over the past couple of months. I know in the public sphere, there's a lot of both criticisms, the restaurants were shut down suddenly, but for some people, the measures are actually not going far enough as the growth is going, in terms of concerns about ICU and other places, so.
Caroline Coljin 13:58
Yeah, and in a way, they're both right. You know, it's terrible that we have to close restaurants. And it's also essential that we stop transmission. So those two things, you know, there's definitely truth in both of them. So, you asked about modeling over the last couple of months, and I think on everyone's mind right now is variants of concern. And I think that's one where again, in a way we somewhat missed the message or we were late in understanding the message in the policy and public health world as modelers. We tried to sound this alarm in February when it was, and even in January in some areas, when it was clear that the B.1.1.7 variant in the UK had a higher transmission rate than the COVID we were used to. So, it caused a huge problem in the UK, required a very strict shutdown to get it under control, and it overwhelmed numbers of the regular COVID consistently over a period of maybe six to nine weeks. And that happened not just in one place, because it was in one group of people, and they had a lot of parties. It happened consistently in many different areas.
Caroline Coljin 15:09
And so that's a really strong sign that it had higher transmission. And that just means it's better at getting from person to person. So, if on average, we were each case infecting one other, and it started to be one and a half, that's clearly greater than one. And we would expect to see that variant, if it established itself here, start to grow exponentially. And at first, it would have really small numbers. And so compared to 500 cases, that 10, that doubles to 20, that 20, that doubles to 40 wouldn't look like a lot. And in fact, in a lot of Canada, overall cases were falling, and policymakers without data in front of them about daily numbers of variants and without, you know, wanting to prioritize modeling and science and problems they didn't have yet, did not maybe act to prevent the variants from establishing and moving around the country. And now we're seeing the rises that actually we predicted in February to say, you know, if we don't contain this now, we will see rises in late March and into April. That's what we're seeing. That's frustrating, it's frustrating to be right about predictions that are unpopular, and put predictions of variants into one of their modeling reports. And they got a lot of negative response for that. But actually, that was their job to present the science that we know and send it out to policymakers into the wider messaging so that you know that that's frustrating to see that. I think the situation now, of course, in this so-called wave is different because of variants, but also because of vaccination. And so, vaccination is, we're holding it up as the light at the end of the tunnel. I think variants are in a way that the train that's in the tunnel that's blocking us from getting to that light at the end. So, I think we're in a challenging situation for the next few weeks or a couple of months, maybe even, and then hopefully we'll catch up with vaccination.
Am Johal 17:04
Yeah, certain journalists, media commentators, like , from the Tyee have been writing about this for some time, a few months back. I'm just wondering, in terms of what you're modeling your research, if you could look into May and June, what it shows in terms of the near future, in BC?
Caroline Coljin 17:28
So that's a good question. And I don't have modelling results in my hand for May and June that are like with up-to-date data from BC. But what I do, what we did look at is the vaccine rollout and how the short-term picture changes depending on how we vaccinate. What we want to do is vaccinate everyone, as soon as possible. Ideally, we would vaccinate everyone in our population tomorrow, and three weeks later, we would give them all their second dose, and that would be great. We don't have the capacity to do that. Logistically, or, you know, most importantly, in terms of the number of doses. So, what we did was we did some modeling, that combined data on how the vaccines work, so new data coming out of Israel in the UK, about how well vaccines can prevent infection from happening in the first place, as well as preventing people from getting sick, getting symptomatic or severe disease. And we looked at patterns of how people contact each other by age in the population, but also the idea of like there are some people who can't avoid having high contact at work. So essential workers, not because the work is more essential, but because it has more high contact and its unavoidable high contact at work. And we found that in the short term, prioritizing preventing infection by vaccinating those groups who have to have high contact actually does better than the strict age-based rollout. And that's just because not everyone is going to be equally exposed.
Caroline Coljin 18:54
If everyone was going to be exposed, then you would need to start with the people who are most at risk of a bad outcome given that exposure. But if you can control that exposure, that's even better, because you know what's better than a 95% efficacy, 100%, because you weren't exposed to the risk in the first place. So modeling is a tool we can use to balance out the information or to synthesize the information that we have about how the vaccines work, how transmission works, the impact of social distancing, the numbers of cases by age, the numbers of hospitalizations, and deaths by age, the age-based risk, all of that information. We can put it all together and synthesize it and use modeling to explore what are the impacts of different policies and what we found was, vaccinating essential workers earlier actually prevents hospitalizations and deaths in the older individuals, even more than vaccinating them directly.
Am Johal 19:50
This has been such a disorienting time in the way this pandemic has emerged. I can remember back in the first part thinking, okay, a couple of months, and we'll get into the Summer, and we'll see. And as the information came through, I remember Labour Day last Fall, having a small socially distance get together with people thinking, 鈥淥kay, we need to, this is going to, there's going to be a second wave at the end of March, by April, things are going to start to look better.鈥� And then, of course, April emerges, and we're still in restrictions with still a lot of variables at play. I know that there have been communications from the chief medical health officer for BC, Bonnie Henry, around post secondary institutions being back in September, and a number of other factors with vaccines rolling out. From your mathematical modeling, is there a way to think about what next Fall and Spring would look like? Because I think that, as the vaccines were rolling out, there's this assumption, we're going to be back into some kind of normal, but other epidemiologists and others have been talking about mask wearing and social distancing going well into 2022. And I'm wondering, from your research, what you think would be the right kind of public policy approach, given the fact that the virus will still be around in some form or another?
Caroline Coljin 21:21
Yeah, absolutely. So, we have done quite a bit of modeling around this question. And we're going to be putting out a blog post in the next day or two. So, it's good timing that you asked about that. We've written a report about kind of what does September look like under different assumptions about vaccination and about variants. So first, you know, if we, if we keep to our regular COVID, or our B.1.1.7 COVID, the situation in September actually looks really good, in many ways. We can reopen a lot and likely do so without overwhelming our hospital and ICU capacity. So that's good. The cautionary note with that is that we can't reopen fully because B.1.1.7 has a higher transmission rate than the regular COVID. The regular COVID, probably between an R of between two and four. If we model something two and a half, which means we keep up our symptomatic testing and contact tracing, we keep up some of our mask wearing. But otherwise, we basically relax. That's our kind of, it's a guess, but it's kind of a best guess.
Caroline Coljin 22:25
We do see a fairly big wave of cases that occur in people who are not vaccinated and people who were vaccinated, but who the vaccine because it's not 100% efficacy against infection, those who were not protected. One of the things we can do about that is actually vaccinate kids. In that 10 to 19 age group, because, you know, we would imagine they go to high school or they go to school, even under current contact patterns, they, if we reopen, they see a lot of infection because they're not being vaccinated in the current plans. If we can get approval to vaccinate them and start vaccinating them, it makes a huge difference, it actually makes the difference to whether we can reopen at that R 2.5 level in September, safely or not.
Caroline Coljin 23:11
The other thing that makes a huge difference is if we can get more and more people interested in having a vaccine. So, in the UK, I heard the other day that over, I think 95% of those over 50 had been vaccinated, which is fantastic. If we can achieve rates like that in Canada, it will make a huge difference too. So, we really, it will depend on whether we can vaccinate kids and on whether we are how much people want to have the vaccine. I also think you know what happens if we get to September, and everyone who has, who wants a vaccine has had one. And we've vaccinated, you know, many, many people, and then we reopen, and it wasn't quite enough. And so, we start to see hospitals fill up. Again, we will probably reintroduce restrictions, because that's what we've been doing the whole time is introduce more restrictions when we need to for the hospitalizations. But also, I imagine that people who declined the vaccine the first time might take it the next time if they know that there are lots of people in hospital who have not been vaccinated, and that's why. So hopefully that will change over the course of the Fall even if things go wrong.
Caroline Coljin 24:23
The other thing I wanted to mention is variants. So, we don't know the efficacy of vaccines against the P.1 variant because P.1 has never bumped up against vaccinated populations and the data haven't been available for how efficacious the vaccines are with P.1. So, I think that's something we need to monitor for and make sure we are understanding in BC as we are a place where P.1 is hitting a vaccinated or partially vaccinated population. And then if a new variant happens in the world, somewhere, that does escape vaccine immunity, are we going to be in a position to stop it from coming here, the way P.1 and B.1.1.7 have. And to stop it from undermining our COVID control the way P.1 and B.1.1.7 have. So, I think we need to be in position to do that.
Am Johal 25:12
In the BC vaccination rollout, the decision was made for people to get one shot before people proceeded to the second shot. And I'm wondering in the mathematical modeling with that approach the facts in terms of the broader numbers in terms of waiting for that second vaccine shot and for the time for it to be available in the way that it has public health outcomes, does the timing of that affect the way things could roll out in the Fall as well.
Caroline Coljin 25:44
So, by the Fall, hopefully most of us, maybe, would have been offered a second dose. I'm not sure they're talking about first doses for everyone by June or July. From the modeling, we can look at that. And under the assumptions that you get, if you use the data from the Pfizer trial, it looks much better to do the deferred second dose the way they've done in BC. So, it's really a proactive step from that point of view. And that's just basically saying if you have, you know, 70, 80, 90% efficacy, but you can achieve that in twice the number, you know, double the number of people, that's just more protection than if you have 95% efficacy, but in only half the number of people. So that's kind of more people who are protected directly right away if you defer that second dose.
Caroline Coljin 26:36
From the population dynamics point of view, it's even better than that, because you stop transmission by approaching herd immunity or even reaching herd immunity much sooner than if you can only vaccinate half the number of people so takes you half the time to get to the same level of reopening that it would take you if you had to wait to protect people maybe just over half the time. However, if you thought that the efficacy after a single dose was not good, then that situation would change. And you would need to be waiting and doing the second dose. Because you wouldn't be effectively protecting people with that single dose. The Pfizer trial looked really promising for efficacy after one dose starting at about two weeks after that one dose. There will be more data coming through from Israel and from other vaccinated populations. And we should have more information about that, based on the information we had in December and January, when we were thinking about this, and when that decision was being made, it did look strongly like this would save lives in the short term by protecting twice as many very vulnerable people in very short order. So, I hope that's been borne out.
Am Johal 27:48
Yeah, I have a question for you around, you know, coming in, as a researcher being thrust into the media sphere, as you share your information, I'm wondering if you could reflect a little bit on how that experience has been for you. And any thoughts you might have on media coverage around the pandemic, both the positive sides of it and perhaps some critiques as well?
Caroline Coljin 28:13
Yeah, I've really enjoyed doing the media work, I do it quite a lot and I do find it really satisfying. And I think that's because it's fun and it's important to communicate science to the public as clearly as we can. And I hope that my media work and our group's media work in the scientific community too, more broadly, does contribute to that, that it's not canned press releases, that it is talking to people who are actively doing research and are cranking away at modeling or other science questions day-by-day. So, it is satisfying. I'm sure the media are exhausted by it in the sense that there's just so little else to cover, you know, there's no major events to cover. People aren't traveling around. Sometimes it feels a bit like it's all COVID news all the time. And of course, there are other events, obviously the US election got a lot of coverage and so not that nothing else has happened. But there's been so much intense media coverage.
Caroline Coljin 29:11
As a scientist, I think it's great to have the opportunity to speak directly to the public, I really welcome it and to clarify things for journalists. I think they also have a really important role to play in questioning policy and in questioning government and being part of the discourse that does ask tough questions. And so, kind of shaping their understanding and knowledge in those conversations. I think is really important in terms of getting, you know, getting the best policy and getting the best discourse, the best high-quality discourse that we can. Sometimes things get a little bit misstated. For example, last week or two weeks ago, Dr. Henry noted that not all positive cases will be sent for whole genome sequencing. And that was misunderstood, I think to be saying they're not going to be looking for variants of concern. So, there are lots of different ways to look for variants and it's not just sequencing. And so sometimes you get a bit of a spin out for things that maybe shouldn't have been worrying but are taken a little bit out of context.
Caroline Coljin 30:13
But I think overall, it's been a really, for me, it's been a really positive set of experiences. And I think a valuable one for communicating about a broad set of issues to the public at a time when the public cares. It's really nice to communicate science when everyone cares, like we do. We didn't know before the pandemic came that everyone would know about the R Nought and the basic reproductive number and the R number and mRNA vaccines and herd immunity and flatten the curve. And these things are just household phrases now that we never would have, you know, we kind of knew a pandemic would come one day and people would joke, 鈥渨ould you buy an island or would you be like, what would you do?鈥� But we never thought, oh, yeah, I know what I'll do, I'll talk flatten the curve to the public every day because everyone will care. So that's been interesting for my field to applied mathematics.
Am Johal 30:58
Thank you so much for joining us on Below the Radar and sharing so succinctly and synthesizing the complex and important research you do. Hopefully, we can check in with you again in the Fall and see where we're at. I've been sometimes right, sometimes wrong as an amateur epidemiologist as everyone is these days. But it's wonderful to hear about your research and the important role it has to play in interacting with public policy. So, thank you.
Caroline Coljin 31:31
Well, thank you so much for having me.
[theme music]
Melissa Roach 31:36
Below the Radar is a knowledge democracy podcast created by 丁香园AV's Vancity Office of Community Engagement. Thanks for listening to our conversation with Caroline Coljin. You can head to the show notes to find out more about her research group as well as some of the media interviews she's done. Thanks again and we'll see you next time on Below the Radar.
[theme music fades]