Hello, I'm Chris Anderson. Welcome to The TED Interview. We're gearing up for season four with some extraordinary guests, but I don't want to wait for that for today's episode, because we're in the middle of a pandemic, and there's a guest I really wanted to talk to now.
He is Adam Kucharski, an infectious diseases scientist who focuses on the mathematical modeling of pandemics. He's an associate professor at the London School of Hygiene and Tropical Medicine and a TED Fellow.
(TED Talk) Adam Kucharski: So what kind of behavior is actually important for epidemics? Conversations, close physical contacts? What sort of data should we be collecting before an outbreak if we want to predict how infection might spread? To find out, our team built a mathematical model ...
Chris Anderson: When it comes to figuring out what to make of this pandemic, known technically as COVID-19, and informally as just the coronavirus, I find his thinking unbelievably helpful. And I'm excited to dive into it with you. A special callout to my friends on Twitter who offered up many suggestions for questions. I know this topic is on everyone's mind right now. And what I hope this episode does is give us all a more nuanced way of thinking about how this pandemic has unfolded so far, what might be to come and what we all collectively can do about it. Let's dive in.
Adam, welcome to the TED Interview.
Adam Kucharski: Thank you.
CA: So let's just start with a couple of basics. A lot of skeptical people's response — certainly over the last few weeks, maybe less so now — has been, "Oh, come on, this isn't such a big deal, there's a relatively tiny number of cases. Compare it to the flu, compare it to anything else. There are much bigger problems in the world. Why are we making such a fuss about this?" And I guess the answer to that fuss is that it comes down to the mathematics. We're talking about the mathematics of exponential growth, fundamentally, right?
AK: Exactly. And there's a number that we use to get a sense of how easy things spread and the level of transmission we're dealing with. We call that the reproduction number, and conceptually, it's just, for each case you have, on average, how many others are they infecting? And that gives you a sense of how much is this scaling, how much this growth is going to look like. For coronavirus, we're now seeing, across multiple countries, we're seeing each person on average giving it to two or three more.
CA: So that reproduction number, the first thing to understand is that any number above one means that this thing is going to grow. Any number below one means it's going to diminish.
AK: Exactly — if you have it above one, then each group of people infected are going to be generating more infection than there was before. And you will see the exponential effect, so if it's two, you will be doubling every round of infection, and if it's below one, you're going to get something that's going to decline, on average.
CA: So that number two or higher, I think everyone here is maybe familiar with the famous story of the chessboard and the grains of rice, and if you double the number of grains for every square of the chessboard, for the first 10 or 15 squares nothing much happens, but by the time you've got to the 64th square, you suddenly have tons of rice for every individual on the planet.
Exponential growth is an incredible thing. And the small numbers now are really not what you should be paying attention to — you should be paying attention to the models of what could be to come.
AK: Exactly. Obviously, if you continue the exponential growth, you do sometimes get these incredibly large, maybe implausibly large numbers. But even looking at a timescale of say, a month, if the reproduction number is three, each person is infecting three on average. The gap between these rounds of infection is about five days. So if you imagine that you've got one case now, that's, kind of, six of these five-day steps in a month. So by the end of that month, that one person could have generated, I think it works out at about 729 cases. So even in a month, just the scale of this thing can really shoot up if it's not controlled.
CA: And so certainly, that seems to be happening on most numbers that you look at now, certainly where the virus is in the early stages of entering a country. You've given a model whereby we can much more clearly understand this reproduction number, because it seems to me this is almost like the core to how we think of the virus and how we respond to it and how much we should fear it, almost. And in your thinking, you sort of break it down into four components, which you call DOTS: Duration, Opportunities, Transmission probability and Susceptibility. And I think it would be really helpful, Adam, for you to just explain each of these, because it's quite a simple equation that links those four things to the actual reproduction number. So talk about them in turn. Duration, what does that mean?
AK: Duration measures how long someone is infectious for. If, for example, intuitively, if someone is infectious for a longer period of time, say, twice as long as someone else, then that's twice the length that they've got to be spreading infection.
CA: And what is the duration number for this virus, compared with, say, flu or with other pathogens?
AK: It depends a little bit on what happens when people are infectious, if they're being isolated very quickly, that shortens that period of time, but potentially, we're looking at around a week people are effectively infectious before they might be isolated in hospital.
CA: And during that week, they may not even be showing symptoms for that full week either, right? So someone gets infected, there's an incubation period. There's a period some way into that incubation period where they start being infectious, and there may be a period after that, where they start to show symptoms, and it's not clear, quite, how those dates align. Is that right?
AK: No, we're getting more information. One of the signals we see in data that suggest that you may have that early transmission going on is when you have this delay from one infection to the next. So that seems to be around five days. That incubation period, the time for symptoms to appear, is also about five days. So if you imagine that most people are only infecting others when they're symptomatic, you'd have that incubation period and then you'd have some more time when they're infecting others. So the fact that those values seem to be similar, suggesting that some people are transmitting either very early on or potentially before they're showing clear symptoms.
CA: So almost implies that on average, people are infecting others as much before they show symptoms as after.
AK: Potentially. Obviously these are early data sets, but I think there's good evidence that a fair number of people, either before they're showing clear symptoms or maybe they're not showing the kind of very distinctive fever and cough but they're feeling unwell and they're shedding virus during that period.
CA: And does that make it quite different from the flu, for example?
AK: It makes it actually similar to flu in that regard. One of the reasons pandemic flu is so hard to control and so feared as a threat is because so much transmission happens before people are severely ill. And that means that by the time you identify these cases, they've probably actually spread it to a number of other people.
CA: Yeah, so this is the trickery of the thing, and why it's so hard to do anything about it. It is ahead of us all the time, and you can't just pay attention to how someone feels or what they're doing. I mean, how does that happen, by the way? How does someone infect someone else before they're even showing symptoms themselves, because classically, we think of, you know, the person sneezing and droplets go through the air and someone else breathes them in and that's how infection happens. What is actually going on for infection pre-symptoms?
AK: So the level of transmission we see with this virus isn't what we see, for example, with measles, where someone sneezes and a lot of virus gets out and potentially lots of susceptible people can get exposed. So potentially, it could be quite early on that even if someone has quite mild symptoms, maybe a bit of a cough, that's enough for some virus to be getting out and particularly, some of the work that we've done trying to look at sort of close gatherings, so very tight-knit meals, there was an example in a ski chalet — and even in those situations, you might have someone mildly ill, but enough virus is getting out and somehow exposing others, we're still trying to work out exactly how, but there's enough there to cause some infection.
CA: But if someone's mildly ill, don't they still have symptoms? Isn't there evidence that even before they know that they're ill, something is going on? There was a German paper published this week that seemed to suggest that even really early on, you take a swab from the back of someone's throat and they have hundreds of thousands of these viruses already reproducing there. Like, can someone just literally just be breathing normally and there is some transmission of virus out into the air that they don't even know about and is either infecting people directly or settling on surfaces, is that possible?
AK: I think that's what we're trying to pin down, how much that [unclear]. As you said, there's evidence that you can have people without symptoms and you can get virus out their throats. And so certainly there's potential that it can be breathed out, but is that a fairly rare event for that actual transmission to happen, or are we potentially seeing more infections occur through that route? So it's really early data, and I think it's a piece of the puzzle, but we're trying to work out where that fits in with what we know about the kind of other transmission events we've seen.
CA: Alright, so, duration is the duration of the period of infectiousness. We think is five to six days, is that what I heard you say?
AK: Potentially around a week, depending on exactly what happens to people when they're infectious.
CA: And there are cases of people testing positive way, way later, after they've got infected. That may be true, but they are probably not as infectious then. Is that basically right, that's the way to think of this?
AK: I think that's our working theory, that a lot of that infection is happening early on. And we see that for a number of respiratory infections, that when people obviously become severely ill, their behavior is very different to when they may be walking around and going about their normal day.
CA: And so again, comparing that D number to other cases, like the flu, is flu similar? What's the D number for flu?
AK: So for flu, it's probably slightly shorter in terms of the period that people are actively infectious. I mean, for flu, it's a very quick turnover from one case to the next, actually. Even a matter of about three days, potentially, from one infection to the person that they infect. And then at the other end of the scale, you get things like STDs, where that duration could be several months, potentially.
CA: Right. OK, really nothing that unusual so far, in terms of this particular virus. Let's look at the O, opportunity. What is that?
AK: So opportunity is a measure of how many chances the virus has to spread through interactions while someone is infectious. So typically, it's a measure of social behavior. On average, how many social contacts do people make that create opportunities for transmission while they're infectious.
CA: So it's how many people have you got close enough to during a day, during a given day, to have a chance of infecting them. And that number could be, if people aren't taking precautions in a normal, sort of, urban setting, I mean, that could run into the hundreds, presumably?
AK: Potentially, for some people. We've done a number of studies looking at that in recent years, and the average, in terms of physical contacts, is about five people per day. Most people will have conversation or contacts generally with about 10, 15, but obviously, between cultures, we see quite a lot of variation in the level of physical greetings that might happen.
CA: And presumably, that number again is no different for this virus than for any other. I mean, that's just a feature of the lives that we live.
AK: I think for this one, if it's driven through these kind of interactions, and we've seen for flu, for other respiratory infections, those kinds of fairly close contacts and everyday physical interactions seem to be the ones that are important for transmission.
CA: Perhaps there is one difference. The fact that if you're infectious pre-symptoms, perhaps that means that actually, there are more opportunities here. This is part of the virus's genius, as it were, that by not letting on that it's inside someone, people continue to interact and go to work and take the subway and so forth, not even knowing that they're sick.
AK: Exactly. And for something like flu, you see when people get ill, clearly, their social contacts drop off. So to have a virus that can be infectious while people are going around their everyday lives, really gives it an advantage in terms of transmission.
CA: In your modeling, do you actually have this opportunities number higher than for flu?
AK: So for the moment, we're kind of using similar values, so we're trying to look at, for example, physical contacts within different populations. But what we are doing is scaling the risk. So that's coming on to the T term. So that between each contact, what's the risk that a transmission event will occur.
CA: Alright, so let's go on to this next number, the T, transmission probability. How do you define that?
AK: So this measures the chance that, essentially, the virus will get across during a particular opportunity or a particular interaction. So you may well have a conversation with somebody, but actually, you don't cough or you don't sneeze or for some reason, the virus doesn't actually get across and expose the other person. And so, for this virus, as I mentioned, say people are having 10 conversations a day, but we're not seeing infected people infect 10 others a day. So it suggests that not all of those opportunities are actually resulting in the virus getting across.
CA: But people say that this is an infectious virus. Like, what is that transmission probability number, again, compared with, say, the flu?
AK: So, we did some analysis looking at these very close gatherings. We looked at about 10 different case studies, and we found that about a third of the contacts in those settings subsequently got infected in these early stages, when people weren't aware. So if you had these, kind of, big group meals, potentially, each contact had about, a kind of, one in three chance of getting exposed. For seasonal flu, that tends to be slightly lower, even within households and close settings, you don't necessarily get values that high. And even for something like SARS, those values have, kind of — the risk per interaction you had was lower than what we seem to be getting for coronavirus. Which intuitively makes sense, there must be a higher risk per interaction if this thing is spreading so easily.
CA: Hm. OK, and then the fourth letter of DOTS is S for susceptibility. What's that?
AK: So that is a measure of the proportion of the population who are susceptible. If you imagine you have this interaction with someone, the virus gets across, it exposes them, but some people may have been vaccinated or otherwise have some immunity and not develop infection themselves and not be infectious to others. So we've got to account for this potential proportion of people who are not actually going to turn into cases themselves.
CA: And obviously, there's no vaccine yet for this coronavirus, nor is anyone, at least initially, immune, as far as we know. So are you modeling that susceptibility number pretty high, is that part of the issue here?
AK: Yeah, I think the evidence is that this is going to fully susceptible populations, and even in areas, for example, like China, where there's been a lot of transmission but there's been very strong control measures, we estimated that up to the end of January, probably about 95 percent of Wuhan are still susceptible. So there's been a lot of infection, but it hasn't really taken much of that component, of the DOTS, of those four things that drive transmission.
CA: And so the way the mathematics works, I have to confess, amidst the stress of this whole situation, the nerd in me kind of loves the elegance of the mathematics here, because I'd never really thought about it this way, but you basically just multiply those numbers together to get the reproduction number. Is that right?
AK: Exactly, yeah, you almost take the path of the infection during transmission as you multiply those together, and that gives you the number for that virus.
CA: And so there's just a total logic to that. It's the number of days, duration that you're infectious, it's the number of people you're seeing on average during those days that you have a chance to infect. Then you multiply that by the transmission probability, is virus getting into them, essentially, that's what you mean by crossing over. And then by the susceptibility number. By the way, what do you think the susceptibility probability is for this case?
AK: I think we have to assume that it's near 100 percent in terms of spread, yeah.
CA: Alright, you multiply those numbers together, and right now, it looks like, for this coronavirus, that you say two to three is the most plausible current number, which implies very rapid growth.
AK: Exactly. In these uncontrolled outbreaks, we're seeing now a number of countries in this stage — you are going to get this really rapid growth occurring.
CA: And so how does that two to three compare with flu? And I guess, there's seasonal flu, in the winter, when it's spreading, and at other times during the year drops well below one as a reproduction number, right? But what is it during seasonal flu time?
AK: During the early stage when it's taking off at the start of the flu season, it's probably, we reckon, somewhere between about maybe 1.2, 1.4. So it's not incredibly transmissible, if you imagine you do have some immunity in your population from vaccination and from other things. So it can spread, it's above one, but it's not taking off, necessarily, as quickly as the coronavirus is.
CA: So I want to come back to two of those elements, specifically opportunity and transmission probability, because those seem to have the most chance to actually do something about this infection rate. Before we go there, let's talk about another key number on this, which is the fatality rate. First of all, could you define — I think there's two different versions of the fatality rate that maybe confuse people. Could you define them?
AK: So the one that we often talk about is what's known as the case fatality rate, and that's of the proportion who show up with symptoms as cases, what proportion of those will subsequently be fatal. And we also sometimes talk about what's known as the infection fatality rate, which is, of everyone who gets infected, regardless of symptoms, how many of those infections will subsequently be fatal. But most of the values we see kicking around are the case fatality rate, or the CFR, as it's sometimes known.
CA: And so what is that fatality rate for this virus, and again, how does that compare with other pathogens?
AK: So there's a few numbers that have been bouncing around. One of the challenges in real time is you often don't see all of your cases, you have people symptomatic not being reported. You also have a delay. If you imagine, for example, 100 people turn up to a hospital with coronavirus and none have died yet, that doesn't imply that the fatality rate is zero, because you've got to wait to see what might happen to them. So when you adjust for that underreporting and delays, best estimate for the case fatality is about one percent. So about one percent of people with symptoms, on average, those outcomes are fatal. And that's probably about 10 times worse than seasonal flu.
CA: Yeah, so that's a scary comparison right there, given how many people die of flu. So when the World Health Organization mentioned a higher number, a little while back, of 3.4 percent, they were criticized a bit for that. Explain why that might have been misleading and how to think about it and adjust for that.
AK: It's incredibly common that people look at these raw numbers, they say, "How many deaths are there so far, how many cases," and they look at that ratio, and even a couple of weeks ago, that number produced a two percent value. But if you imagine you have this delay effect, then even if you stop all your cases, you will still have these kind of fatal outcomes over time, so that number will creep up. This has occurred in every epidemic from pandemic flu to Ebola, we see this again and again. And I made the point to a number of people that this number is going to go up, because as China's cases slow, it will look like it's increasing, and that's just kind of a statistical quirk. There's nothing really kind of, behind a change, there's no mutations or anything going on.
CA: If I have this right, there are two effects going on. One is that the number of fatalities from the existing caseload will rise, which actually would boost that 3.4 even higher. But then you have to offset that against the fact that, apparently, huge numbers of cases have just gone undetected and that we haven't, due to bad testing, that the number of fatalities don't — they probably reflect a much larger number of early cases. Is that it?
AK: Exactly. So you have one thing pulling the number up and one thing pulling it down. And it means that on these kind of early values, if you actually just adjust for the delay and don't think about these unreported cases, you start getting really very scary numbers indeed. You get up to 20, 30 percent potentially, which really doesn't align with what we know about this virus in general.
CA: Alright. There's a lot more data in now. From your point of view, you think the likely fatality rate, at least in the earlier stage of an infection, is about two percent?
AK: I think overall, I think we can put something probably in the 0.5 to two percent range, and that's on a number of different data sets. And that's for people who are symptomatic. I think on average, one percent is a good number to work with.
CA: OK, one percent, So flu is often quoted as a tenth of a percent, so it's five to 10 times or more more dangerous than flu. And that danger is not symmetric across age groups, as is well known. It primarily affects the elderly.
AK: Yeah, we've seen that one percent on average, but once you start getting into the over 60s, over 70s, that number really starts to shoot up. I mean, we're estimating potentially in these older groups, you're looking at maybe five, 10 percent fatality. And then of course, on top of that, you've got to add what are going to be the severe cases and people are going to require hospitalization. And those risks get very large in the older groups indeed.
CA: Adam, put these numbers together for us. In your models, if you put together a reproduction rate of two to three and a fatality rate of 0.5 percent to one percent and you run the simulation, what does it look like?
AK: So if you have this uncontrolled transmission, and you have this reproduction number of two or three and you don't do anything about it, the only way the outbreak ends is enough people get it and immunity builds up and the outbreak kind of ends on its own. And in that case, you would expect very large numbers of the population to be infected. It's what we see, for example, with many other uncontained outbreaks, that it essentially burns through the population, you get large numbers infected and with this kind of fatality rate and hospitalization rate, that would really be hugely damaging if that were to occur. Certainly at the country level, we're seeing — Italy is a good example at the moment, that if you have that early transmission that's undetected, that rapid growth, you very quickly get to a situation where your health systems are overwhelmed. I think one of the nastiest aspects of this virus is that because you have the delay between infection and symptoms and people showing up in health care, if your health system is overwhelmed, even on that day, if you completely stop transmission, you've got all of these people who have already been exposed, so you're still going to have cases and severe cases appearing for maybe another couple of weeks. So it's really this huge accumulation of infection and burden that's coming through the system on your population.
CA: So there's another key number, actually, is how does the total case number compare to the capacity of a country's health system to process that number of cases. Presumably that issue makes a huge difference to the fatality rate, the difference between people coming in with severe illness and a health system that's able to respond and one that's overwhelmed. The fatality rate is going to be very different at that point.
AK: If someone requires an ICU bed, that's a couple of weeks they're going to require it for and you've got more cases coming through the system, so it very quickly gets very tough.
CA: So talk about the difference between containment and mitigation. These are different terms that we're hearing a lot about. In the early stages of the virus, governments are focused on containment. What does that mean?
AK: Containment is this idea that you can focus your effort on control very much on the cases and their contacts. So you're not causing disruption to the wider population, you have a case that comes in, you isolate them, you work out who they've come into contact with, who's potentially these opportunities for exposure and then you can follow up those people, maybe quarantine them to make sure that no further transmission happens. So it's a very focused, targeted method, and for SARS, it worked remarkably well. But I think for this infection, because some cases are going to be missed or undetected, you've really got to be capturing a large chunk of people at risk. If a few slip through the net, potentially, you're going to get an outbreak.
CA: Are there any countries that have been able to employ this strategy and effectively contain the virus?
AK: So Singapore have been doing a really remarkable job of this for the last six weeks or so. So as well as some wider measures, they've been working incredibly hard to trace people who have come into contact. Looking at CCTV, going through to find out which taxi someone might have gotten, who might be at risk — really, really thorough follow-up. And for about six weeks, that has kept a lid on transmission.
CA: So that's amazing. So someone comes into the country, they test positive — they go to work, and with a massive team, and trace everything to the level of actually saying, "Oh, you don't know what taxi you went in? Let us find that out for you." And presumably, when they find the taxi driver, they then have to try and figure out everyone else who was in that taxi?
AK: So they will focus on close contacts of people most at risk, but they're really minimizing the chance that anyone slips through the net.
CA: But even in Singapore, if I'm not mistaken, numbers started to trend back down to zero, but I think recently, they've picked up again a bit. It's still unclear whether they will actually be able to sustain containment.
AK: Exactly. If we talk in terms of the reproduction number, we saw it dipped to maybe 0.8, 0.9, so under that crucial value of one. But in the last week or two, it does seem to be ticking up and they're getting more cases appearing. I think a lot of it is, even if they are containing it, the world is experiencing outbreaks and just keeps throwing sparks of infection, and it becomes harder and harder with that level of intensive effort to stamp them all out.
CA: In the case of this virus, you know, there was warning to most countries in the world that this thing was happening. The news out of China very quickly became very bleak, and people had time to prepare. I mean, what would ideal preparation look like if you know that something like this is coming and you know that there's a lot on the line if you can successfully contain it before it really escapes?
AK: I think two things would make a really big difference. One is having as thorough a follow-up and detection as possible. We've done some modeling analyses, looking at how effective that kind of early containment is. And it can be, if you're identifying maybe 70 or 80 percent of the people who might have come into contact. But if you're not detecting those cases coming in, if you're not detecting their contacts — and a lot of the early focus, for example, was on travel history to China, and then it became clear that the situation was changing, but because you were relying on that as your definition of a case, it meant a lot of maybe other cases that matched the definition weren't being tested because they didn't seem to be potentially at risk.
CA: So I mean, if you know that early detection is key to this, an essential early measure, I guess, would be to rapidly ensure that you had enough tests available and where needed, so that you could respond, be ready to swing into action as soon as someone was detected, you then have to very quickly, I guess, test their contacts and so forth, to have a chance of keeping this under control.
AK: Exactly. In my line of work, we say there's value in a negative test, because it shows that you're looking for something and it's not there. And so I think having small numbers of people tested doesn't give you confidence that you're not missing infections, whereas if you are doing really thorough follow-up on contacts, we've seen countries even like Korea now, huge numbers of people tested. So although there are still cases appearing, it gives them more confidence that they have some sense of where those infections are.
CA: I mean, you're in the UK right now, I'm in the US. How likely is it that the UK is going to be able to contain, how likely is it that the US is going to be able to contain this?
AK: I think it's pretty unlikely in both cases. I think the UK is going to have to introduce some additional measures. I think when that happens obviously depends a bit on the current situation, but we've tested almost 30,000 people now. Frankly, I think the US may well be moving beyond that point, given how much evidence of extensive transmission that has, and I think without clear ideas of how much infection there is and that level of testing, it's quite hard to actually see what the picture currently is in the US.
CA: I mean, I definitely don't want to get too political about this, but I mean, does this strike you as — you just said that the UK has tested 30,000 people — the US is five or six times bigger and I think the total number of tests here is five or six thousand, or it was a few days ago. Does that strike you as bizarre? I don't understand, honestly, how that happened in an educated country that has so much knowledge about infectious diseases.
AK: It does, and I think there's obviously a number of factors playing in there, logistics and so on, but there has been that period of warning that this is a threat and this is coming in. And I think countries need to make sure that they've got the capacity to really do as much detection as they can in those early stages, because that's where you're going to catch it and that's where you're going to have a better chance of containing it.
CA: OK, so if you fail to contain, then you have to move to some kind of mitigation strategy. So what comes into play there? And I think I almost want to bring that back to two of your DOTS factors, opportunity and transmission probability, because it seems like the virus is what it is, the actual duration when someone is potentially infectious, we can't do much about. The susceptibility side, we can't do much about until there's a vaccine. We could maybe talk about that in a bit. But the middle two of opportunity and transmission probability, we can do something about. Do you want to maybe talk about those in turn, of what that looks like, how would you build a mitigation strategy? I mean, first of all, thinking about opportunity, how do you reduce the number of opportunities to pass on the bug?
AK: And so I think in that respect, it would be about massive changes in our social interactions. And if you think in terms of the reproduction number of being about two or three, to get that number below one, you've really got to cut some aspect of that transmission in half or in two-thirds to get that below one. And so that would require, of the opportunities that could spread the virus, so these kind of close contacts, everybody in the population, on average, will be needing to reduce those interactions potentially by two-thirds to bring it under control. That might be through working from home, from changing lifestyle and kind of where you go in crowded places and dinners. And of course, these measures, things like school closures, and other things that just attempt to reduce the social mixing of a population.
CA: Well, actually, talk to me more about school closures, because that, if I remember, often in past pandemics has been cited as an absolutely key measure, that schools represent this sort of coming together of people, children are often — certainly when it comes to flu and colds — they're carriers. But on this case, children don't seem to be getting sick from this particular virus, or at least very few of them are. Do we know whether they can still be infectious? They can be the unintended carriers of it. Or actually, is there evidence that school closures may not be as important in this instance as it is in others?
AK: So that point on what role children play is a crucial one, and there's still not a good evidence base there. From following up of contacts of cases, there's now evidence that children are getting infected, so when you're testing, they are getting exposed, it's not that somehow they're just not getting the infection at all, but as you said, they're not showing symptoms in the same way. And particularly for flu, when we see the implications of school closures, even in the UK in 2009 during swine flu, there was a dip in the outbreak during the school holidays, you could see it on the epidemic curve, it kind of comes back down in the summer and goes back up in the autumn. But of course, in 2009, there was some immunity in older groups. That kind of shifted more the transmission into the younger ones. So I think it's really something we're trying to work to understand. Obviously, it will reduce interactions, with school closures, but then there's knock-on social effects, there's potential knock-on changes in mixing, maybe grandparents and their role, in terms of alternative carers if parents have to work. So I think there's a lot of pieces that need to be considered.
CA: I mean, based on all of the different pieces of evidence you've seen, if it were down to you, would you be recommending that most countries at this point do look hard at extensive school closures as a precautionary measure, that it's just worth it to do that as a sort of painful two, three, four, five-month strategy? What would you recommend?
AK: I think the key thing, given the age distribution of risk and the severity in older groups is reduce interactions that bring the infection into those groups. And then amongst everyone else, reduce interactions as much as possible. I think the key thing is we've got so much of the disease burden in the kind of 60-plus group that it's not just about everyone trying to avoid everyone's interactions, but it's the kind of behaviors that would drive infections into those groups.
CA: Does that mean that people should think twice before, I don't know, visiting a loved one in an old people's home or in a residential facility? Like that, we should just pay super special attention to that, should all these facilities be taking great care about who they admit, taking temperature and checking for symptoms or something like that?
AK: I think those measures definitely need to be considered. In the UK, we're getting plans for potentially what's known as a cocooning strategy for these older groups that we can really try and seal off interactions as much as possible from people who might be bringing infection in. And ultimately, because as you said, we can't target these other aspects of transmission, it is just reducing the risk of exposure in these groups, and so I think anything at the individual level you can do to get people reducing their risk, if either they're elderly or in other risk groups, I think is crucial. And I think more at the general level those kind of more large-scale measures can help reduce interactions overall, but I think if those reductions are happening and not reducing the risk for people who are going to get severe disease, then you're still going to get this really remarkably severe burden.
CA: I mean, do people have to almost apply this double lens as they think about this stuff? There's the risk to you as you go about your life, of you catching this bug. But there's also the risk of you being, unintentionally, a carrier to someone who would suffer much more than you might. And that both those things have to be top of mind right now.
AK: Yeah, and it's not just whose hand you shake, it's whose hand that person goes on to shake. And I think we need to think about these second-degree steps, that you might think you have low risk and you're in a younger group, but you're often going to be a very short step away from someone who is going to get hit very hard by this. And I think we really need to be socially minded and think this could be quite dramatic in terms of change of behavior, but it needs to be to reduce the impact that we're potentially facing.
CA: So the opportunity number, we bring down by just reducing the number of physical contacts we have with other people. And I guess the transmission probability number, how do we bring that down? That impacts how we interact. You mentioned hand-shaking, I'm guessing you're going to say no handshaking.
AK: Yeah, so changes like that. I mean, another one, I think, handwashing operates in a way that we might be still be doing activities that we've previously done, but handwashing reduces the chance that from one interaction to another, we might be spreading infection, so it's all of these measures that mean that even if we're having these exposures, we're taking additional steps to avoid any transmission happening.
CA: I still think most people don't fully understand or don't have a clear model of the pathway by which this thing spreads. So you think definitely people understand that you don't breathe in the water droplets of someone who has just coughed or sneezed. So how does it spread? It gets onto surfaces. How? Do people just breathe out and it goes on from people who are sick, they touch their mouth or something like that, and then touch a surface and it gets on that way? How does it actually get onto surfaces?
AK: I think a lot of it would be that you cough in your hand and it ends up on a surface. But I think the challenge, obviously, is untangling these questions of how transmission happens. You have transmission in a household, and is it that someone coughed and it got on a surface, is it direct contact, is it a handshake, and even for things like flu, that's something that we work quite hard to try and unpick, how does social behavior correspond to infection risk. Because it's clearly important, but pinning it down is really tough.
CA: It's almost like embracing the fact that for a lot of these things, we actually don't know and that we're all in this game of probabilities. Which, in a way, is why I think the math is so important here. That you have to think of this as these multiple numbers working together on each other, they all have their part to play. And any of them that you can take down by a percentage is likely contributing, not just to you but to everyone. And people don't actually know in detail how the numbers go together, but they know that they probably all matter. We almost need people to, somehow, you know, embrace that uncertainty and then try to get some satisfaction by acting on every single part of it.
AK: I think it's this idea that if on average, you're infecting, say, three people, what's driving that and how can you chip away at that value? If you're washing your hands, how much might that chip away in terms of the handshakes, you know, you may have had virus and you no longer do, or if you are changing your social behavior in a certain way, is that taking away a couple of interactions, is that taking away half? How can you really chip into that number as much as you possibly can?
CA: Is there anything else to say about how we could reduce that transmission probability in our interactions? Like, what is the physical distance that it's wise to stay away from other people if we can?
AK: I think it's hard to pin down exactly, but I think one thing to bear in mind is that there's not so much evidence that this is a kind of aerosol and it goes really far — it's reasonably short distances. I don't think it's the case that you're sitting a few meters away from someone and the virus is somehow going to get across. It's in closer interactions, and it's why we're seeing so many transmission events occur in things like meals and really tight-knit groups. Because if you imagine that's where you can get a virus out and onto surfaces and onto hands and onto faces, and it's really situations like that we've got to think more about.
CA: So in a way, some of the fears that people have may actually be overstated, like, if you're in the middle of an airplane and someone at the front sneezes, I mean, that's annoying, but it's actually not the thing you should be most freaked out about. There are much smarter ways to pay attention to your well-being.
AK: Yeah, if it was measles and the plane was susceptible people, you would see a lot of infections after that. I think it is, bear in mind, that this is, on average, people infecting two or three others, so it's not the case of your maybe 50 interactions over a week, all of those people are at risk. But it's going to be some of them, particularly those close contacts, that are going to be where transmission's occurring.
CA: So talk about, from a sort of national strategy point of view. There's a lot of talk about the need to "flatten the curve." What does that mean?
AK: I think it refers to this idea that for your health systems, you don't want all your cases to appear at the same time. So if we sat back and did nothing and just let the epidemic grow, and you had this growth rate that, at the moment, in some places is looking like maybe three to four days, you're getting doubling. So every three or four days, the epidemic is doubling. It will skyrocket and you'll end up with a whole bunch of really severely ill people needing hospital care all at the same time, and you just won't have capacity for it. So the idea of flattening the curve is if we can slow transmission, if we can get that reproduction number down, then there may still be an outbreak, but it will be much flatter, it will be longer and there will be fewer severe cases showing up, which means that they can get the health care they need.
CA: Does it imply that there will be fewer cases overall, or — When you look at the actual images of people showing what flattening the curve looks like, it almost looks as if you've got the same area still under the graph, i.e. that the same number of people, ultimately, are infected but over a longer period. Is that typically what happens, and even if you adopt all these strategies of social distancing and washing hands and etc. that the best you can hope for is that you slow the thing down, you actually will get as many people infected in the end?
AK: Not necessarily — it depends on the measures that go in. There are some measures like, shutting down travel, which typically delay the spread rather than reduce it. So you're still going to get the same outbreaks, but you're stretching out the outbreaks. But there are other measures. If we talk about reducing interactions, if your reproduction number's lower, you would expect fewer cases overall. And eventually, in your population, you will get some buildup of immunity, which would help you out if you think about the components, reducing susceptibility, alongside what's going on elsewhere. So the hope is that the two things will work together.
CA: So help me understand what the endgame is here. So, take China, for example. Whatever you make of the early suppression of data and so forth that seems pretty troubling there. The intensity of the response come January time or whatever, with the shutdown of this huge area of the country, seems to have actually been effective. The number of cases there are falling at a shockingly high rate in some ways. Falling to almost nothing. And I can't understand that. You are talking about a country of, whatever, 1.4 billion people. There have been a huge number of cases there, but it was a tiny fraction of the population have actually got sick. And yet, they've got the number way down. It's not like every other person in China has somehow developed immunity. Is it that they have been absolutely disciplined about shutting down travel from the infected regions and somehow really dialed up, massively dialed up testing at any sign of any problem, so that literally, they are back in containment mode in most parts of China? I can't get my head around it, help me understand it.
AK: So we estimated, in the last two weeks of January, when these measures went in, the reproduction number went from about 2.4 to 1.1. So about 60 percent decline in transmission in the space of a week or two. Which is remarkable and really, a lot of it is likely to be driven by just fundamental change in social behavior, huge social distancing, really intensive follow-up, intensive testing. And it got to the point where it took enough off the reproduction number to cause the decline, and now, of course, we're seeing, in many areas, a transition back to more of this kind of containment, because there's few cases, it's more manageable. But we're also seeing them face a challenge, because a lot of these cities have basically been locked down for six weeks and there's a limit to how long you can do that for. And so some of these measures are gradually starting to be lifted, which of course creates the risk that cases that are appearing from other countries may subsequently go in and reintroduce transmission.
CA: But given how infectious the bug is, and how many theoretical pathways and connection points there are between people in Wuhan, even in shutdown, or relatively shut down, or the other places where there's been some infection and the rest of the country, does it surprise you how quickly that curve has gone down to nearly zero?
AK: Yes. Early on when we saw that flattening off in cases in those first few days, we did wonder whether it was just they hit a limit in testing capacity and they were reporting 1,000 a day, because that's all the kits they had. But it continued, thankfully, and it shows that it is possible to turn this over with that level of intervention. I think the key thing now is seeing how it works in other settings. Italy now are putting in really dramatic interventions. But of course, because of this delay effect, if you put them in today, you won't necessarily see the effects on cases for another week or two. So I think working out what impact that's had is going to be key for helping other countries work on how to contain this.
CA: To have a picture, Adam, of how this is likely to play out over the next month or two, give us a couple of scenarios that are in your head.
AK: I think the optimistic scenario is that we're going to learn a lot from places like Italy that have unfortunately been hit very hard. And that countries are going to take this very seriously and that we're not going to get this continued growth that's going to overwhelm totally, that we're going to be able to sufficiently slow it down, that we are going to get large numbers of cases, we're probably going to get a lot of severe cases, but that will be more manageable, that's the kind of optimistic scenario. I think if we have a point where countries either don't take this seriously or populations don't respond well to control measures or it's not detected, we could get situations — I think Iran is probably the closest one at the moment — where there's been extensive widespread transmission, and by the time it's being responded to, those infections are already in the system and they are going to turn up as cases and severe illness. So I'm hoping we're not at that point, but we've certainly got, at the moment, potentially about 10 countries on that trajectory to have the same outlook as Italy. So it's really crucial what happens in the next couple of weeks.
CA: Is there a real chance that quite a few countries end up having, this year, substantially more deaths from this virus than from seasonal flu?
AK: I think for some countries that is likely, yeah. I think if control is not possible, and we've seen it happen in China, but that was really just an unprecedented level of intervention. It was really just changing the social fabric. I think people, many of us, don't really appreciate, at a glance, just what that means, to reduce your interactions to that extent. I think many countries just simply won't be able to manage that.
CA: It's almost a challenge to democracies, isn't it — "OK, show us what you can do without that kind of draconian control. If you don't like the thought of that, come on, citizens, step up, show us what you're capable of, show that you can be wise about this and smart and self-disciplined, and get ahead of the damn bug."
CA: I mean, I'm not personally superoptimistic about that, because there's such conflicting messaging coming out in so many different places, and people don't like to short-term sacrifice. I mean, is there almost a case that — I mean, what's your view on whether the media has played a helpful role here or an unhelpful role? Is it actually, in some ways, helpful to, if anything, overstate the concern, the fear, and actually make people panic a little bit?
AK: I think it's a really tough balance to strike, because of course, early on, if you don't have cases, if you don't have any evidence of potential pressure, it's very hard to get that message and convince people to take it seriously if you're overhyping it. But equally, if you're waiting too long, and saying it's not a concern yet, we're OK for the moment, a lot of people think it's just flu. By the time it hits hard, as I've said, you're going to have weeks of an overburdened health system, because even if you take interventions, it's too late to control the infections that have happened. So I think it's a fine line, and my hope is there is this ramp-up in messaging, now people have these tangible examples like Italy, where they can see what's going to happen if they don't take it seriously. But certainly, of all the diseases I've seen, I think many of my colleagues who are much older than me and have memories of other outbreaks, it's the scariest thing we've seen in terms of the impact it could have, and I think we need to respond to that.
CA: It's the scariest disease you've seen. Wow. I've got some questions for you from my friends on Twitter. Everyone is obviously super exercised about this topic. Hypothetically, if everyone stayed home for three weeks, would that effectively wipe this out? Is there a way to socially distance ourselves out of this?
AK: Yeah, I think in certain countries with reasonably small household sizes, I think average in the UK, US is about two and a half, so even if you had a round of infection within the household, that would probably stamp it out. As a secondary benefit, you may well stamp out a few other infections, too. Measles only circulates in humans, so you may have some knock-on effect, if, of course, that were ever to be possible.
CA: I mean, obviously that would be a huge dent to the economy, and this is in a way almost, like, one of the underlying challenges here is that you can't optimize public policy for both economic health and fighting a virus. Like, those two things are, to some extent, in conflict, or at least, short-term economic health and fighting a virus. Those two things are in conflict, right? And societies need to pick one.
AK: It is tough to convince people of that balance, the thing we always say of pandemic planning is it's cheap to put this stuff in place now — otherwise, you've got to pay for it later. But unfortunately, as we've seen with this, that a lot of early money for response wasn't there. And it's only when it has an impact and when it's going to get expensive that people are happy to take that cost on board, it seems.
CA: OK, some more Twitter questions. Will the rising temperature in coming weeks and months slow down the COVID-19 spread?
AK: I haven't seen any convincing evidence that there's that strong pattern with temperature, and we've seen it for other infections that there is this seasonal pattern, but I think the fact we're getting widespread outbreaks makes it hard to identify, and of course, there's other things going on. So even if one country doesn't have as big an outbreak as another, that's going to be influenced by control measures, by social behavior, by opportunities and these things as well. So it would be really reassuring if this was the case, but I don't think we can say that just yet.
CA: Continuing from Twitter, I mean, is there a standardized global recommendation for all countries on how to do this? And if not, why not?
AK: I think that's what people are trying to piece together, first in terms of what works. It's only really in the last sort of few weeks we've got a sense that this thing can be controllable with this extent of interventions, but of course, not all countries can do what China have done, some of these measures incur a huge social, economic, psychological burden on populations. And of course, there's the time limit. In China they've had them in for six weeks it's tough to maintain that, so we need to think of these tradeoffs of all the things we can ask people to do, what's going to have the most impact on actually reducing the burden.
CA: Another question: How did this happen and is it likely to happen again?
AK: So it's likely that this originated with the virus that was circling in bats and then probably made its way through another species into humans somehow, there's a lot of bits of evidence around this, there's not kind of single, clear story, but even for SARS, it took several years for genomics to piece together the exact route that it happened. But certainly, I think it's plausible that it could happen again. Nature is throwing out these viruses constantly. Many of them aren't well-adapted to humans, don't pick up, you know, there may well have been a virus like this a few years ago that just happened to infect someone who just didn't have any contacts and didn't go any further. I think we are going to face these things and we need to think about how can we get in early at the stage where we're talking small numbers of cases, and even something like this is containable, rather than the situation we've got now.
CA: It seems like this isn't the first time that a virus seems to have emerged from, like, a wild meat market. That's certainly how it happens in the movies. (Laughs) And I think China has already taken some steps this time to try to crack down on that. I guess that's potentially quite a big deal for the future if that can be properly maintained.
AK: It is, and we saw, for example, the H7N9 avian flu, over the last few years, in 2013, it was a big emerging concern, and China made a very extensive response in terms of changing how they operate their markets and vaccination of birds and that seems to have removed that threat. So I think these measures can be effective if they're identified early on.
CA: So talk about vaccinations. That's the key measure, I guess, to change that susceptibility factor in your equation. There's obviously a race on to get these vaccinations out there, there are some candidate vaccinations there. How do you see that playing out?
AK: I think there's certainly some promising development happening, but I think the timescales of these things are really on the order of maybe a year, 18 months before these things be widely available. Obviously, a vaccine has to go through these stages of trials, that takes time, so even if by the end of the year, we have something which is viable and works, we're still going to see a delay before everyone can get ahold of it.
CA: So this really puzzles me, actually, and I'd love to ask you as a mathematician about this as well. There are already several companies believing that they have plausible candidate vaccines. As you say, the process of testing takes forever. Is there a case that we're not thinking about this right when we're looking at the way that testing is done and that the safety calculations are made? Because it's one thing if you're going to introduce a brand new drug or something — yes, you want to test to make sure that there are no side effects, and that can take a long time by the time you've done all the control trials and all the rest of it. If there's a global emergency, isn't there a case, both mathematically and ethically, that there should just be a different calculation, the question shouldn't be "Is there any possible case where this vaccine can do harm," the question surely should be, "On the net probabilities, isn't there a case to roll this out at scale, to have a shot at nipping this thing in the bud?" I mean, what am I missing in thinking that way?
AK: I mean, we do see that in other situations, for example, the Ebola vaccine in 2015 showed, within a few months, very promising evidence and interim results of the trial in humans showed what seemed very high efficacy. And even though it hadn't been licensed fully, it was employed for what is known as compassionate use in subsequent other outbreaks. So there are these mechanisms where vaccines can be fast-tracked in this way. But of course, we're currently in a situation where we have no idea if these things will do anything at all. So I think we need to accrue enough evidence that it could have an impact, but obviously, fast-track that as much as possible.
CA: But the skeptic in me still doesn't fully get this. I don't understand why there isn't more energy behind bolder thinking on this. Everyone seems, despite the overall risk, incredibly risk-averse about how to build the response to it.
AK: So with the caveat that, yeah, there's a lot of good questions on this, and some of them are slightly outside my wheelhouse, but I agree that we need to do more to get timescales out. The example I always quote is it takes us six months to choose a seasonal flu strain and get the vaccines out there to people. We always have to try and predict ahead which strains are going to be circulating. And that's for something we know how to make and has been manufactured for a long time. So there is definitely more that needs to be done to get these timescales shorter. But I think we do have to balance that, especially if we're exposing large numbers of people to something to make sure that we're confident it's safe and that it's going to have some benefit, potentially.
CA: And so, finally, Adam, I guess going into this — There's another set of infectious things happening around the world at the same time, which is ideas and the communication around this thing. They really are two very dynamic, interactive systems of infectiousness — there's some very damaging information out there. Is it fair to think of this as battle of credible knowledge and measures against the bug, and just bad information — You know, part of what we have to think about here is how to suppress one set of things and boost the other, actually, turbocharge the other. How should we think of this?
AK: I think we can definitely think of it almost as competition for our attention, and we see similarly, with diseases, you have viruses competing to infect susceptible hosts. And I think we're now seeing, I guess over the last few years with fake news and misinformation and the emergence of awareness, more of a transition to thinking about how do we reduce that susceptibility if we have people that can be in these different states, how can we try and preempt better with information. I think the challenge for an outbreak is obviously, early on, we have very little good information, and it's very easy for certainty and confidence to fill that vacuum. And so I think that is something — I know platforms are working on how can we get people exposed to good information earlier, so hopefully protect them against other stuff.
CA: One of the big unknowns to me in the year ahead — let's say that the year ahead includes many, many more weeks, for many people, of actually self-isolating. Those of us who are lucky enough to have jobs where you can do that. You know, staying home. By the way, the whole injustice of this situation, where so many people can't do that and continue to make a living, is, I'm sure, going to be a huge deal in the year ahead and if it turns out that death rates are much higher in the latter group than in the former group, and especially in a country like the US, where the latter group doesn't even have proper health insurance and so forth. That feels like right there, that could just become a huge debate, hopefully a huge source of change at some level.
AK: I think that's an incredibly important point, because it's very easy — I similarly have a job where remote working is fairly easy, and it's very easy to say we should just stop social interactions, but of course, that could have an enormous impact on people and the choices and the routine that they can have. And I think those do need to be accounted for, both now and what the effect is going to look like a few months down the line.
CA: When all's said and done, is it fair to say that the world has faced, actually, much graver problems in the past, that on any scenario, it's highly likely that at some point in the next 18 months, let's say, a vaccine is there and starts to get wide distribution, that we will have learned lots of other ways to manage this problem? But at some point, next year probably, the world will feel like it's got on top of this and can move on. Is that likely to be it, or is this more likely to be, you know, it escapes, it's now an endemic nightmare that every year picks off far more people than are picked off by the flu currently. What are the likely ways forward, just taking a slightly longer-term view?
AK: I think there's plausible ways you could see all of those potentially playing out. I think the most plausible is probably that we'll see very rapid growth this year and lots of large outbreaks that don't recur, necessarily. But there is a potential sequence of events that could end up with these kind of multiyear outbreaks in different places and reemerging. But I think it's likely we'll see most transmission concentrated in the next year or so. And then, obviously, if there's a vaccine available, we can move past this, and hopefully learn from this. I think a lot of the countries that responded very strongly to this were hit very hard by SARS. Singapore, Hong Kong, that really did leave an impact, and I think that's something they've drawn on very heavily in their response to this.
CA: Alright. So let's wrap up maybe by just encouraging people to channel their inner mathematician and especially think about the opportunities and the transmission probabilities that they can help shift. Just remind us of the top three or four or five or six things that you would love to see people doing.
AK: I think at the individual level, just thinking a lot more about your interactions and your risk of infection and obviously, what gets onto your hands and once that gets onto your face, and how do you potentially create that risk for others. I think also, in terms of interactions, with things like handshakes and maybe contacts you don't need to have. You know, how can we get those down as much as possible. If each person's giving it to two or three others, how do we get that number down to one, through our behavior. And then it's likely that we'll need some larger-scale interventions in terms of gatherings, conferences, other things where there's a lot of opportunities for transmission. And really, I think that combination of that individual level, you know, if you're ill or potentially you're going to get ill, reducing that risk, but then also us working together to prevent it getting into those groups who, if it continues to be uncontrolled, could really hit some people very, very hard.
CA: Yeah, there's a lot of things that we may need to gently let go of for a bit. And maybe try to reinvent the best aspects of them.
Thank you so much. If people want to keep up with you, first of all, they can follow you on Twitter, for example. What's your Twitter handle?
AK: So @AdamJKucharski, all one word.
CA: Adam, thank you so much for your time, stay well.
AK: Thank you.
CA: Associate professor and TED Fellow Adam Kucharski. We'd love to hear what you think of this bonus episode. Please tell us by rating and reviewing us in Apple Podcasts or your favorite podcast app. Those reviews are influential, actually. We certainly read every one, and truly appreciate your feedback.
This week's show was produced by Dan O'Donnell at Transmitter Media. Our production manager is Roxanne Hai Lash, our fact-checker Nicole Bode. This episode was mixed by Sam Bair. Our theme music is by Allison Layton-Brown. Special thanks to my colleague Michelle Quint.
Thanks for listening to the TED Interview. We'll be back later this spring with a whole new season's worth of deep dives with great minds. I hope you'll enjoy them whether or not life is back to normal.
I'm Chris Anderson, thanks for listening and stay well.