This transcript has been edited for clarity.
Eric J. Topol, MD: Hello. This is Eric Topol for Medscape. I'm with my co-host Abraham Verghese for a new edition of Medicine and the Machine. We have an extraordinary guest today, Professor Christina Pagel. She is a force — a professor at University College London with an extraordinary background in math, physics, and even interplanetary space. We've never had a guest with such a diverse background. Welcome, Christina.
Christina Pagel, PhD: Thank you.
Topol: You've provided extraordinary insights throughout the pandemic. But before we get into that, you've had a unique background as a physicist and mathematician. Then, somewhere along the way, after this great training you had in the United Kingdom, you went to Boston and made a switch in your career to using math and artificial intelligence (AI) to help inform data for health. How did that come about?
Pagel: I originally studied maths at university because I wanted to be a theoretical physicist. That was my aim. I always knew the kind of maths you need to do that. Then I earned a master's in quantum theory. But by then, all the bits of quantum theory that I really liked turned out to be solved.
I asked, can I do a PhD in this? And they said no, it's been done; Feynman did it all 30 years ago. So then I thought, I'd also really like to go into space and be an astronaut. I spoke to the European Space Agency, and they said, you need to do a PhD in space physics, which I'd never heard of. But I went to Imperial College London and asked, can I do a PhD in this? And they said yes. One of the benefits of doing maths is that it's actually quite easy to switch fields. So I did my PhD in interplanetary magnetic fields. And then I came to Boston University for 3 years as a postdoctoral researcher working on interplanetary electrons.
It was an amazing 3 years. But I realized two things: First, I was quite good at math but I wasn't a great physicist. Much to my disappointment, I just didn't have a feel for it in the way that some people have. And second, what I was working on, although it was kind of cool, it didn't matter. If I got it wrong, it didn't make a difference. And I wanted to feel that my work was doing that; I wanted to feel that I was contributing more to society than paying my taxes and not committing crimes.
I was looking around, thinking, what can I do? And I found this department at University College London, where they were applying mathematics to healthcare. I could see that some people there had a physics background. I thought it sounded really interesting. I've been there ever since.
Abraham Verghese: Dr Pagel, it's a great pleasure to meet you. Most of us are so narrowly specialized that we need people like you to grasp the entirety of what's going on in this world. Before we get to COVID, I know you also have a master's in medieval history and I'm sure there's a story behind that.
Pagel: Actually, I have two master's degrees in history. I have one in classical civilization, primarily the ancient Greeks and Romans, and then I went back and I did another in medieval history.
When I was 16, I had to choose between an arts or a science path. I loved both history and science. I picked science, but I never lost my love for history. It was the first time I'd chosen to do a course purely for interest. It had no impact on my career. I was doing it only for myself. It takes you back to the meaning of learning, I think, of trying to understand and find out things, and it's just so interesting.
I realized that having a mathematical background makes you quite good at history. It teaches you to look for what's not there as well as what's there, and helps you create logical arguments and understand that causation isn't correlation, which, I think, is important in history.
Verghese: I believe that it's relevant to COVID and everything you're engaged in. Santayana said that if we don't understand history, we're condemned to repeat it. And I think we're seeing a lot of that now.
Topol: By the way, if you don't follow @chrischirp, then you're missing out in terms of COVID — she is a go-to for information and perspective. Before the pandemic started, you were and still are involved in work to understand congenital heart disease, and cardiac surgery in children and adults, and I suspect much more than that.
As a cardiologist, I cue into that stuff. Can you tell us about what that work has been like and how that was a background for some things you're doing now?
Pagel: I've been working on congenital heart disease for almost 15 years. It all starts with two facts about the United Kingdom. The first is that we have a national healthcare system, which means that you can organize things at a national scale. The second is that in 1997, there was the Bristol heart scandal. One of the hospitals where heart surgery was performed in babies had much higher than expected mortality rates when compared with other units that were doing the same kinds of operations.
That took a long time to come to light because it's a high-risk, highly specialized area, so you expect, unfortunately, that some children will die. Then you have this whole question of, how many is too many? What does "expected" even mean when you have this group of children who have such diverse health problems and diagnoses, and all these diverse procedures performed on them?
After that came to light, the National Health Service (NHS) decided to centralize the service, so only about 10-12 hospitals offer it. They did that because they wanted to ensure that there are enough operations happening every year in each of those hospitals. And they made it mandatory for everyone to submit their data.
It gets audited. The data are checked. Every year, they publish survival statistics on the grounds that we don't want this to happen again. That's been going on since 2000, and it has meant that congenital heart disease services are one of the best areas, with a long-standing dataset.
Over the years, we've refined the risk models for deciding how risky these surgeries are for different children. How do we know whether some of the units have higher death rates than average? We can show that, since they've started monitoring, results have gotten better pretty much every single year.
Now that the field has come along, we're looking at what other things matter. It's not just about survival. It's about complications, quality of life. How many surgeries do you need over your lifetime? What happens in outpatients? How are people engaging with it? What happens as children become adults?
So we're using these national datasets to really dig into it. It requires quite a lot of careful statistics and data, but also talking to clinicians, talking to patients, talking to parents. We've also built websites trying to explain what the data are and aren't showing. That's how I started thinking about how we present information that matters to people in a way that they find easy to understand and that's fair.
Verghese: Regarding your work in operational research, I'm wondering, is that a common department now in many universities? What is the day-to-day life of an operational researcher like?
Pagel: So few people know what operational research is. It's technically a branch of mathematics, but it can sit in lots of different places. In the United States, it's often called operations research. It's also called systems engineering, and then it sits within engineering faculties. It can also be called management science and is taught as part of an MBA. So it runs this whole gamut of different subjects and techniques. But at its core, you want to use mathematics and data and any kind of other analytical techniques to improve decision-making. That's the idea.
It's meant to be pragmatic, focused on working with people to understand the actual problem they have. The problem often is not what people say it is. It asks, what information do you have, and how can you use that to improve the decisions you're about to make?
That is the core of it. There are standard techniques such as optimization, queueing theory, mathematical modeling or simulation. But the heart of operational research is trying to improve things, and the techniques you use actually aren't part of it. It's just that certain techniques are more common than others.
Topol: That gets us into the pandemic, because this provides a unique background — fresh eyes, transdisciplinary experience, and an intergalactic way to understanding health systems.
We are in the third year of the pandemic. You've been a leading light, a spokesperson for your views and data interpretation, not just in the United Kingdom but throughout the world. Can you give us an overview? Obviously, it hasn't gone well in many respects. What are your key takeaways and concerns? Where do you see the United Kingdom as an outlier, if at all, or different from other places like the United States, which is perhaps even worse?
Pagel: A few things stand out in the big picture. Right at the beginning, the rich Western countries weren't prepared. I believe we thought we were, and there was an element of complacency that we knew what to do.
We have the world's leading health systems, so of course we'll be fine. And that wasn't the case. As societies become more medicalized — vaccinations, treatments, healthier lives — we've become used to a growing life expectancy, not as many infectious diseases, and not as many ubiquitous public health problems.
We became quite complacent about what an infectious disease can do to a population. We had let our public health functions lapse. Certainly, in the United Kingdom, there wasn't that much public health expertise. The public health bodies are much smaller than they used to be, and people just weren't ready. Whereas some of the middle- and lower-income countries that have strong public health systems that are trying to deliver large-scale vaccination and nutrition programs, they had a much better set-up to do things like contact tracing and supporting people who were ill, to think about how to mobilize a population response.
The other thing that struck me was how unwilling the West was to learn from other countries —particularly the East Asian countries, places like South Korea, Taiwan, and Japan. We just felt they had nothing to teach us. I think that was a big mistake, and I think that's still the case today.
People talk about the inevitability of COVID. You only have to look at those countries to see how it wasn't inevitable. We have these quite lazy stereotypes — we say that wouldn't work over here because they're different over there. Well, how do you know? What did make those countries quite different is that a lot of them had experienced SARS, and they learned from that and applied that knowledge to this new pandemic.
What worries me is that we're not trying to learn from COVID and apply it to a new pandemic.
Another thing that struck me is that you can understand the panic and the mistakes that happened in the first wave in March 2020. But by the time the Alpha wave hit in late 2020, early 2021, it felt like not that much had been learned. We didn't mitigate it, particularly factors around transmission and about it being airborne. By the summer of 2020, it was reasonably clear that it could spread through the air. It was clear that masks helped. It was clear that ventilation helped. It was clear that outside conditions were much safer than inside. But little effort was made to improve indoor air quality. To this day, I do not understand that, because good air quality comes with so many more public health benefits, beyond COVID. Clean indoor air is in the public good, along the same lines as clean water. I can't understand why we've never prioritized it.
Verghese: You were prescient about the shape of the BA.5 variant and how that might look a couple of months before we saw it. What does your crystal ball show of what we can expect in the United Kingdom and the United States in terms of variants that have not yet emerged?
Pagel: The other thing that strikes me is that people still haven't understood exponential growth 2.5 years in. With the BA.5 or BA.3 before it, or the first Omicron before that, people say, oh, how did you know? Well, it was doubling every week, and I projected forward. Then in 8 weeks, it's dominant.
It's not that hard. It's just that people don't believe it. Somehow people think, oh, well, it can't happen. But what exactly is going to stop it? You have to have a mechanism to stop exponential growth at the moment when enough people have immunity. The moment doesn't last very long, and then you get these repeated waves.
You have to have a mechanism that will stop it evolving, and I don't see that. We're not doing anything different to what we were doing a year ago or 6 months ago. So yes, it's still evolving. There are still new variants shooting up all the time.
At the moment, none of these look devastating; we probably have at least 6 weeks' breathing space. But another variant will come because I can't see that we're doing anything to stop it.
Topol: One word you have used is "complacency," which we still have a lot of — a heavy dose of complacency. The other sentiment you're invoking by the unwillingness to accept exponential growth is denialism. There's a lot of that.
Long COVID is another area of concern. It's a softer endpoint than death and hospitalizations, but nonetheless, it's much more prevalent and real. But there's still a lot of denialism. Can you talk about long COVID?
Pagel: Long COVID is interesting because it took a long time for people to accept that it existed at all. And it still is dismissed and ignored.
Vaccination has not made long COVID go away to the same extent that vaccination has massively reduced the number of deaths and of severe, acute illnesses. A recent paper in JAMA showed that vaccination reduced the incidence of long COVID in healthcare workers by up to two thirds. So it helps, but it certainly doesn't make it go away.
What worries me just as much as long COVID are the longer-term problems that are becoming evident through the great work by the Veterans Administration, looking at 1-year and longer follow-up of people who've had COVID and showing elevated levels of organ dysfunction, pretty much anywhere you look.
That's the other thing about complacency: People want to fit COVID into a box they can understand. But this is a brand-new disease, and we don't understand it. As with the pandemic flu in 1918, it's thought to have been associated with a wave of Parkinson's disease about 10-15 years later.
We have no idea what’s ahead of us. To me, this idea that we can live with widespread transmission is betting the future on thinking that we understand it, but there's no particular evidence that we do, even with all the new evidence showing longer-term issues with COVID, whether that's long COVID or other kinds of organ problems.
Verghese: One thing that strikes me is that we're not talking enough about indoor air and the difference it makes, and about long COVID in the sense that if the public really knew how fearful we are, perhaps it would change behavior.
From your perspective as someone with a deep understanding of history, this is not new, this disjunction between what we know would help and the public's willingness to accept it, and political opportunism. What can we do differently from what they did in the era of Camus' The Plague, for example? What can we do to impact the science ignorance that hurts us?
Pagel: I wish I knew. The thing about cleaner indoor air is that it works on any variant and any airborne disease; it helps against pollution; it helps against all kinds of things. And it doesn't take away anyone's freedoms.
What I find difficult is that people in the "COVID isn't a problem" spectrum still don't seem to be pro-ventilation. I just cannot understand why, because it doesn't impact anybody's choices for living their lives. All it does is make people slightly healthier and mitigate some harm. So, I can't quite see why we wouldn't do that. The various studies, and there have been many, showed that it's extremely cost-effective.
One of the issues is that a lot of the experts in ventilation are not doctors or clinical experts. They're engineers, architects, chemists, and physicists. They're coming from a different evidential paradigm, if you like. So you get these calls for trials of ventilation. And these nonclinical experts say, because we understand the physics, we know these things work.
It's quite a clash of cultures along those lines that is an issue. People see the pandemic and think the solution has to be medical when the solution is actually engineering. It doesn't fit into how governments prime themselves to respond.
Also, now that most people have had COVID and have recovered from COVID, including me — I've had it twice now — when people tell you this is a massive issue, it's easy to think, well, but I'm fine.
It's hard to overcome that and explain that you're fine now, but you don't know what the long-term implications are. Is it a question of getting COVID three times in 5 or 6 years? Or getting it twice a year? That can make quite a difference on your health. And do you really want to be taking a week to 10 days off work every single year from COVID? Can you add that to the amount of illness you were suffering before? So there's a strong economic argument for it as well.
Maybe we have to start making these arguments, because it is a different situation now. We're not in the situation we were 2.5 years ago. I think sometimes people believe that when you say anything about controlling COVID, they think you're arguing for lockdowns or something else, when it's not about that.
It's about how we maintain our economy and our quality of life and our health in a way that allows us all to enjoy our lives the way that we want to.
Topol: You are on the SAGE committee. Tell us about how that group reviews science and makes recommendations to your government.
Pagel: There's a bit of a misunderstanding there. SAGE is the Scientific Advisory Group for Emergencies, which is the government's group of experts that does advise on the pandemic, and it was disbanded in March. At some point, probably hundreds of people contributed to that.
There is also a group, which I belong to, called Independent SAGE. That was launched in May 2020 specifically because at the beginning of the pandemic, membership of SAGE, which was advising the government, and all of the minutes of the decision-making were secret. So we had the government saying, we're making decisions based on the science, but we had no idea what that science was. Independent SAGE was formed to bring some of those discussions into the public. Then, SAGE started making their minutes public.
The government used to hold daily press briefings on COVID. They stopped in June 2020. Although the United Kingdom has published lots of data — we have strong national data surveillance — it's not necessarily published in a particularly friendly form for nonexperts. So what Independent SAGE has been doing is holding weekly online briefings.
I've done a lot of this as well, where I try to collate all of the data that are out there and explain what's happening right now and what it means. We've evolved into being a bit more of a public-facing science body, where we're trying to explain and interpret what's happening for the public. We take public questions and focus on particular issues such as schools or inequality; we've talked about long COVID a lot, and children.
Topol: I wish we had something like this. Right, Abraham?
Verghese: Indeed. But I must say, both of you are doing a great job of explaining the intricacies of the data to the rest of us. It's just that people with fixed mindsets can easily tune out the most obvious facts.
Pagel: I rely on Eric, especially for the literature filtering. It's incredible how you read the papers and then condense them so I can understand them. I'm not an expert in a lot of the science.
I go to my expert friends and say, you're going to have to tell me what this abstract means because I do not understand biology. Then they explain it, and I say, so is this right? And they say yes or no. Unfortunately, on academic Twitter, there are a lot of people who are amazing scientists but cannot understand how much knowledge they have that other people don't have. So I read these long Twitter threads, but I don't understand them.
Or there are people who are very much into performative cleverness. You would think that their job as an academic is to show how clever they are. Their information is not useful. I always try to provide source data and to make sure I'm not excluding people from that knowledge.
Topol: That gets to the dissemination of the information. What is so important about your background is you are into the hard data and evidence, and the interpretation of that — the analytics, if you will. Then you go to social media and you get into the cesspool of toxic responses.
By putting a lot of time into trying to help, you get this ridiculous backlash of people who are just there to cause insult and trouble. How do you deal with that? Every day I wonder, what the hell am I doing? There's so much negativism. You're trying to help, and you're putting in the effort and time, and look what you get. Of course, a lot of people appreciate that. But I'm sure you're subject to similar issues.
Pagel: It's gotten a lot easier over time for me, and I imagine for you. Certainly, once you get to a certain number of followers, there's no way you can read all the comments. In fact, mostly, I don't read them.
It means that Twitter stops being a conversation. It becomes more of a broadcast medium. You lose something from that, but at the same time, you can't get too worried about what random people are saying. Right at the beginning, one of my friends who already had a large following — he's in politics —told me to never, ever engage in Twitter arguments. You will always regret it, and they'll never work. I mainly followed his advice. A couple of times when I didn't, I regretted it, and I still regret it because those conversations are still used against me now.
Never get engaged. If you do, always pretend that the person means well. A lot of people are doing it in bad faith, but if you respond as if they aren't, there's not that much that people do. Also, you have good mute settings.
For the first year, I didn't block anyone on Twitter. That changed when I read more about how Twitter works, and how people see tweets, and how they get amplified. I had a few quite high profile, quite horrible commentators on Twitter throwing pile-ons on me, people with a million followers calling me certifiable and saying I should be in prison. If you block them, their followers can't see it and you just stop it. Someone told me to think of it as if you're walking down the street and someone starts shouting at you. You don't have to listen to it. You have a right to walk away from it.
Now I have quite a low threshold for blocking people because if they're clearly not trying to engage or be helpful and are just shouting at me, I don't have to listen to it. I've never taken it that personally because these people don't know me.
It's been a hard pandemic. People are angry and upset for all kinds of reasons. I get that. I find it much harder when other scientists are attacking me, maybe more politely. I find that harder because they're my peers, and they're basically trying to damage my reputation.
Verghese: I see a lot of undergraduates at Stanford very narrowly focused on computer science, which is the flavor du jour right now. Your career is a wonderful testament to the importance of a broad perspective, not just math but astronomy and history. I think we're going to need more of your kind of educated people for the complexity of what we're dealing with, and it's not going to be possible otherwise to have civilized conversations.
Pagel: A lot of the time, I end up feeling like a complete idiot. I'm not this super-specialized person. And I've always got my mom, as well, telling me, why is it they are listening to you, Christina? Haven't you said it all already?
Topol: I had the privilege of doing an NHS review just before the pandemic in 2018 or 2019. And I was struck by the strong data-driven culture of the team I got to work with.
I saw it during the pandemic with the UK Health Security Agency Office for National Statistics (ONS) reports that would come out every week, and the Intensive Care National Audit and Research Centre (ICNARC) reports of all the ICU admissions, which we don't have here in the United States. Just extraordinary. And then the government just gave up on all this stuff. It seems premature. They were leading the world. We were all learning every day, every week from the United Kingdom because the data were extraordinary. Can you comment about that?
Pagel: It’s sad because I think you're right. One area in which the UK has been leading the world is in surveillance, particularly random testing surveillance, which the ONS is still doing, and also the Imperial REACT study, which, unfortunately, was not funded and ended in April this year. I do think it's short sighted.
Some national data collections, like ICNARC, existed before the pandemic and it exists now. It routinely collects data on every single intensive care admission of adults in England and publishes reports on it; the same with pediatric-intensive care; the same with every hospital admission. We have those data. What happened during the pandemic is that they quickly added extra information relevant to COVID. Within days to weeks, they started collating it, so that capability is still there.
Our surveillance has been so good that it's often felt that we were watching and reporting on the pandemic incredibly carefully, but not actually acting on any of the information that it was showing. So in that sense, we have a very sensitive system that's not informing any decisions.
I'm guessing that's what prompted it closing down. If we're not going to do anything different, why are we measuring it? I hope that the ONS Infection Survey, which measures prevalence in a random subgroup of tens of thousands of people every week, carries on because that's the only thing we've got now. And it's one of the only global measures of national long COVID rates as well.
Topol: I also want to ask about your sense of the field of artificial intelligence (AI) in healthcare. You obviously have the grounding with all aspects of machine learning and the complexities of health data, which is not just from health records but all the different layers of data sensors, genomics, and the microbiome. There are lots of different datasets. Where do you see AI? In your view what can it do to improve healthcare in the future?
Pagel: I believe that it can do both more and not as much as what people think. People sometimes see it as this great savior. It isn't. Sometimes what you need is enough nurses and equipment and just old-fashioned doctoring. AI is never going to give you that. AI is incredibly good at taking complex data and using them to understand what's happening, but only if the data contain what is happening.
It's good at things like imaging. The only place where AI is commonly used in the NHS is for things like imaging. If we're looking at scans and you want to assess a tumor or a difference in a tumor, then it's very good at pattern recognition. That's what it's built for. In a sense, all the information is contained in that image.
AI is less good at routine healthcare, which is messy and dirty and subject to lots of biases as to who put it in, at what time, in which place. How has that changed over time? What is that actually telling you about the patient or the system? There it isn't as good. Also, the information of what's in there doesn't necessarily tell you about the future, which is what people are trying to use it for.
We can't predict, say, occupancy in a year's time because what's going to cause occupancy in a year's time is not contained in those data. And AI can't make that happen for me. So sometimes you have to realize there are limitations to what it can do. I believe that some important areas where it could still help more are in things like physiologic time-series data.
We're doing a project on that now, with incredibly complex data. Patients in ICUs have measurements taken multiple times per second for days on end, but we still don't understand that much about how patients get better when they're severely ill. You can see a situation there where AI could help with spotting early deterioration or spotting when patients are ready to leave the ICU, all kinds of other things.
As it turns out, when I'm working with the AI specialists on time series, it just adds a whole extra level of complexity and makes a lot of the normal ways of doing things not work because your assumptions are robbed of independence. Everything is correlated, so it becomes really difficult. But it could become a big thing.
Topol: We've enjoyed this conversation so much. The contributions you've made in the past few years, beyond everything else in your career, Christina, have been what I consider a positive outlier, even momentous.
We've all learned from you, and we're lucky to have someone of your background to be a leading force. We'll keep following you well beyond the pandemic for things that you're going to do to help healthcare. We're glad you switched from physics to healthcare. It's already made a big difference. And it's going to make even more difference in the decades ahead. Thanks so much for joining us today.
Medscape © 2022 WebMD, LLC
Any views expressed above are the author's own and do not necessarily reflect the views of WebMD or Medscape.
Cite this: Eric J. Topol, Abraham Verghese, Christina Pagel. We Are Failing to Use What We've Learned About COVID - Medscape - Sep 09, 2022.