Please tune in to Episode 3 at Apple Podcast, Google Podcast, Amazon Music, Spotify, and Firstory

Transcripts:

Intro:

What is value of information in health research? How do they inform policy changes? How does Bayesian inference help with designing innovative clinical trials? Anna Heath will answer all these questions for you.

Dr. Heath is Canada Research Chair in Statistical Trial Design, a scientist in the Child Health and Evaluative Sciences Program at Sickkids Research Institute, an assistant professor in the division of Biostatistics in University of Toronto, and an Honorary Research Associate in the Department of Statistical Science in University College London. Other than all these fancy titles, she is also the supervisor for my masters practicum project and opened the door to the whole new, amazing world of Bayesian statistics to me.

So of course, I am so thrilled to be hosting Dr. Heath on the show today! Also, I believe you would be as thrilled as I was when learning that, Dr. Heath is a very keen ballroom dancer that participates in international contests, a violin musician and a cricket lover! Now let’s dive into this episode to see what she shared with us!

Jocelyn: Welcome, Anna, to the Biostatistics Podcast. It’s really nice to have you with us here today.

Anna: Thank you for having me.

Jocelyn: So can you start by telling us a bit about your background and how you became interested in biostatistics?

Anna: Of course, yes. So I did my first degree, my undergraduate degree in maths and French. So I started out as a mathematician and I wanted to do maths just because I thought it was really broad and it would lead to a lot of different topics. And then as I became kind of more involved in my degree, I started doing more statistics. I found it really, really interesting. I liked the application. And so I started thinking about what I wanted to do next and decided to do a PhD in statistics. So my PhD is in statistics with application to health economics. And that’s what I was really kind of interested in and still I’m very interested in. And then at the end of my PhD, there was a really interesting role in Toronto to kind of apply the methods that I was working on in my PhD in practice, particularly working with lots of other clinicians and statisticians to implement them to design clinical trials. And that’s really when I started, I guess, working like as a biostatistician. So where I did much more kind of applied work and worked with lots of different people. So I’d say I kind of slowly over my career became more and more applied and probably started being a biostatistician in 2018. So that’s what I’ve done at SickKids since then.

Jocelyn: I see. That’s very interesting. Can you tell us a bit more about what you currently do?

Anna: Yeah. So my current role is as a scientist at SickKids Hospital in Toronto. So what I do is I have a really great team of people who work on developing different statistical methods for the design and analysis of clinical trials. So clinical trials broadly in the context that we work in them are any kind of prospective study in which we try to assess the safety or efficacy of interventions in humans. We work in the team across a lot of different disease areas, particularly at the moment we focus on interventions in the pediatric emergency department. So children who arrive needing emergency care and what are the best kind of pain medications or like distress management or other kind of infectious disease management in the emergency department. And then we also have a big project in critical care. So patients who’ve been admitted to the ICU and they need ventilation. So they’re having respiratory difficulties. And what are the different ways that we can investigate those? So mostly those, but we have projects across a huge range of different areas. And mostly now I do a lot of management, but I work with the people on the team to kind of implement, particularly to the design. So to work out what the best analysis method for the trial is, the best, sorry, the best like outcomes and what are the kind of key quantities that people need to do. And I also work in the health economic evaluation of trials. So trying to investigate whether trials are worth the money that they cost. And in particular, that kind of takes a decision-making perspective on clinical trials. So rather than saying that clinical trials are looking to like determine just efficacy based on a single outcome of interest, we try to look at the kind of whole systems perspective, including costs, impact on other services and effectiveness often in the long term.

Jocelyn: I see. Would you say the health economics metrics or measurements are widely applied in decision-makings when it comes to clinical trials?

Anna: So it’s definitely not applied in clinical trials almost at all. It is applied in policymaking in Canada. So any drug that needs to be or new device that needs to be used by patients in Canada needs to be assessed for economic benefit as well as efficacy. So it’s what the work that we’re trying to do is to kind of bring those two worlds together. So often what happens at the moment is that the clinical, like the clinical trial is done and then the health economics team comes in like right at the end and says, oh, you should have collected all of this extra data. We’re trying to bring that health economic perspective and that decision makers perspective into the design of clinical trials. Interesting. I’m wondering why specifically in Canada. Primarily it’s anywhere with a public health care system. So where the public or where taxes pay for health care. So particularly I come from the UK and that’s where it’s really was developed initially because we have a national health service that provides funding for all I say all health care. But the Canadian system is very similar. And so they also have a kind of public payor perspective in which they try to assess the value of interventions rather than just the efficacy.

Jocelyn: I see. That’s very interesting because I don’t know a lot about health economics. So that’s definitely very refreshing. I’m also wondering since you were in UK before and you probably work with a lot of European people and now we’re in Canada and then primarily we’re working with people from North America. How would you say the environment is different between these two continents in general?

Anna: That’s a really interesting question. I actually still have a lot of collaborators in the UK, so I haven’t entirely left behind my European collaborative relationships. The challenge I have is that now that I’m in Canada, I work in a hospital environment which is much more collaborative with people outside of the field of statistics, whereas in the UK I was a PhD student in a statistics department. So I do have kind of a different perspective anyway. I haven’t found that big a difference between the two worlds. Yeah, I think I really enjoy working in Canada just because I feel there’s a lot of potential to kind of grow these methods and it’s been much easier for me to make these kind of networks of people who are really interested in innovation in the design of clinical trials. But I can’t say that that might not have been the case if I stayed in the UK. I think the UK has a much bigger trials infrastructure, which can mean that it’s challenging to break into that space just because they have kind of a much more, they have these clinical trials units that do a lot of the clinical trials. And so you’d have to kind of be in that space in order to be able to make the impact, whereas Canada seems a little bit more dynamic.

Jocelyn: Yeah, well, that makes sense. So with that being said, how do you think biostatistics research will evolve in the coming years and what new developments or innovation do you see?

Anna: That’s a great question. It’s hard to speak about the entirety of biostatistics because I do a lot of my work primarily in trials. But I think probably one of the biggest changes that we’re going to see is just, again, like as in many cases, that the kind of evolution of data science and machine learning and as it’s used in statistics, I think that’s going to be a big, the fields are probably going to learn from each other. And I certainly know that as these new tools come on board, as we have more granular data, as we’re looking for these kind of what we call like sub-phenotypes, so where you sort of get more precision in the disease, we’re looking to kind of use that in trials to develop interventions that are targeted to specific clinical characteristics of patients. And that’s definitely going to be needing these more complex methods from machine learning and obviously having more data. So I would say just more data. And I think potentially in trials, we also have the potential to be thinking more carefully about more continuously updating data. So you might have something where you get somebody to wear something like a physical activity tracker and then you’re using that data rather than like a single outcome. We have much more complex longitudinal data. So I think the types of data are probably going to be the big change. And then something that I think’s changed as we go forward, it’s just, again, like even as we get more and more computing power, we’re able to design much more complex trials and think really carefully. I think that will change what we’re able to do in trials as we go forward.

Jocelyn: I see. And about the machine learning aspect, you’re saying that will change things. Are you talking in the context of statistics in general or primarily in clinical trials?

Anna: I think these kind of concepts are changing statistics. Like I think it changes how we think about what we do in statistics and what we’re trying to achieve. But I also think they will become something that we can use in trials, which at the moment trials is a very conservative and controlled environment. And I think the innovation of machine learning will probably take longer to translate into trials, but we’ll start to see over the next few years. I see. I guess in my head right now, I can’t really think of how it will directly change the design phase, but it might.

Jocelyn: I mean, I’ve seen talks where they’re like, oh, we use AI to aid the data collection or for better exclusion inclusion criteria. So yeah, that’s definitely interesting. And as you said yourself, Biostatistics is a very collaborative field, and you work with a lot of people from other disciplines as well. So what are some skills and traits of an individual that makes you think that’s the kind of person you want to work with or hire?

Anna: Great question. I think in terms of a collaborative relationship, I would say I’m looking for people who are interested in innovation. So depending on who comes to the relationship, some will see biostatistics as more of a service role, so something that they need to do their trial. And some will come to the relationship as something where we can get a mutually beneficial development. So we can, and the team can design some kind of innovation that will help the clinical team get to their answer faster or more efficiently or with the correct answer or something like that. And I’m always looking for that sort of curiosity or interest in what we do as a team and someone who doesn’t just see this as like a service role to kind of further their research. They see this as kind of a mutually beneficial setting. I think that’s number one. In terms of somebody that I work with or that I would like to employ, I think being a good collaborative biostatistician is really about communication. That’s the primary goal I think of a biostatistician is to communicate, to translate between the world of statistics and the theory and all of that into something that’s applied. And the only way to do that is by communicating with clinical teams. So that’s what I really look for is somebody who is able to communicate ideas clearly and effectively and someone who has an appreciation of speaking to different audiences. The other thing I think for my team that I really need is someone who’s curious, who’s interested in what they’re doing and interested in finding the best solution and not just the fastest solution. I really think that. And then generally, I guess it’s an easy thing to say, but statistical skills. So do you have good coding skills? Do you have a good understanding of key statistical ideas? And that’s often fairly evident. In an interview, I’ll always ask somebody to describe the most recent project that they’ve worked on. And that’s usually what I’m trying to assess is that you have a clear understanding of the differences and the challenges of the statistical methods that you’re using.

Jocelyn: I see. That’s good to know. Hope I’m living up to that expectation. So not on the employing side, for students who wish to pursue a degree in biostatistics, what advice do you want to give them?

Anna: Great question. I think, I guess it’s similar to what I said. I think if you want to pursue a degree in biostatistics, firstly, I would think about the curiosity. Like, is that something that you want to pass the exam? Or do you want to understand? Do you want to like, go get to the root of the problem? And I think that’s really critical, because if you want to go forward and work in biostatistics, it’s not going to work like the textbook tells you. So passing the exams is not enough. You really have to have like a good understanding of the map of the different methods that you could use and where to apply them in practice. So that would be my first thing. Like, are you curious? And curiosity also, I think, has to extend to the clinical area that you’re working in. You have to want to understand that clinical area enough that you can make intelligent conversation about what you put in there. So that’s number one. Again, I do think there’s a set of technical skills that you need. So certainly, most programs will have a sort of prerequisite of some mathematical skills, like calculus, linear algebra. I think it’s really important that you look at those and make sure that you feel confident that you have those, because that is going to be something that you build the foundation on. And if you don’t have that, you’re going to really struggle. And I think, again, like biostatistics itself is a collaborative field that requires communication. So if you want to pursue a career, if you want to be a student in this area, you have to be thinking about how your mathematical skills can apply to understanding real world problems. And so you have to have a drive to understand those real world problems.

Jocelyn: Those are great advice. Thank you. When you’re talking about clinical knowledge, I’m wondering, in your work, how much of clinical knowledge do you need to know to be able to communicate intelligently?

Anna: So I have actually very little clinical knowledge. I actually stopped doing biology, so anything clinical, when I was 16. So I really have not got that foundation of clinical understanding. I think you have to want to learn some of the clinical. So I don’t necessarily know every piece of information, but I have to have a general idea of, like, what are the interventions that are in the trial? What’s the disease area that we’re studying? Where in the hospital are they enrolling the patients? And then I have wonderful collaborators who are patient often with me and explain the clinical context. And I also think you have to be not scared to ask questions. They are the experts. You are not the expert, but you have to know enough to be able to suggest the appropriate analysis or the appropriate design of the study. And so that’s really where you have to know enough. And that will change from context to context. And you can work with the clinical team to get the knowledge that you need.

Jocelyn: I see. That’s very insightful. Thank you very much. And I guess you sort of, I wouldn’t say had a path changing, but you did switch from math to a biostat side. So what do you wish that you knew when you first started your biostatistics career or study?

Anna: I think, I guess I just said I stopped doing biology at 16. I think if I’d known the path that I was going to take, I would have liked to have done a little bit more biology. So I had a little bit more understanding maybe of physiology and like how, you know, the lungs and the heart work and things like that. In terms of what I wish I knew, in addition to that kind of technical knowledge, I think I wish I knew the importance of communication and the communication skills going into the role. I think I kind of underestimated how much communication there was going to be and how much was about taking my expertise and translating it to a clinical or to a non-specialist audience. That took me a while to learn. And I think I made a lot of mistakes early on where I went in too high and I lost people in the room. So that I think would be my main wish if I could go back.

Jocelyn: I see. I do know that you teach a lot of courses, a lot of seminars. So what are some upcoming interesting seminars that you’re giving?

Anna: Oh, I have a couple of conference presentations that will be about designing trials in an efficient manner using different computational techniques. And then I’ve also been invited to give a presentation at the University of Pennsylvania. Again, it will be related to trials and how we kind of think about what I’m interested in is trying to use these health economic ideas to design trials. So I still am working in that trial kind of context when I do a health economic analysis. So I have, again, those in different contexts, some in the pediatric emergency department, a big project in oncology that was funded and that we haven’t managed to start yet because we’re awaiting the postdocs arrival. And then I have lots of methodology in this health economic concept, trying to improve the computational efficiency of the methods so that they can be used in practice.

Jocelyn: I see. That’s very interesting. I guess another question would be, you mentioned before, mostly health economics concepts are applied on the policymaking side. So when you talk about them they will help with the trial design as well. What aspects do you expect them to help?

Anna: Yeah, so typically we use them for research prioritization and design. So the first thing you can do is you can ask yourself, what are the key outcomes of interest if we’re targeting a policymaker? And sometimes you’ll find surprisingly that the key kind of clinical outcome of interest, so maybe 60 day mortality or like trying to see whether you save lives, actually that can be augmented with maybe an understanding of the long term consequences of. Curing this disease or something so you can think about things where you might have. A trade off between quality of life and length of life. And you may find often that quality of life is actually a really important component for the health economic analysis, whereas quantity or length of life is really important for the clinical people. And that’s what we typically can find in our economic analysis is that we need to look at other things in addition to the kind of key clinical outcome. So that’s what we find. I had a colleague who did a really amazing study that I was just reading about the other day where they were planning to do a clinical trial of ultrasound. So basically, if you have someone who’s pregnant in the UK, you get two ultrasound scans. And then if there’s a potential danger later on, they’ll have a third ultrasound just to check that everything’s OK. And they were asking to see whether you should always do a third ultrasound because that would have the potential to save babies lives because you would notice some of these issues earlier on. And they discovered that a traditional clinical trial in this area would require about 100,000 patients because of what they were trying to look for. But they did an analysis to look and see, OK, but what’s actually driving our uncertainty? Like how would we make a better policy decision? And it turned out that actually the efficacy of the ultrasound was not very important. What was important was like the costs associated with inducing labor because that was what was really driving the uncertainty in the analysis. So really that changed the way that they were going to do their study. Instead of doing an almost impractical clinical trial, they’re now focusing on a costing study instead. It definitely sounds very important to incorporate this information.

Jocelyn: Thank you for sharing that. I guess that brings us to the last question. What is one question that you wish I would have asked and how would you have answered it?

Anna: That’s a great question. I think I would say the question would be kind of what are the challenges of doing biostatistics? I think from my perspective, the challenge is really having to do lots of different tasks. Like you’re never doing the same day twice. Some days you’ll be having lots of communication. Some days you’re being really deep in the data. Some days you’ll be designing a study. I think that’s interesting, but it also leads to challenges because you have to master a lot of different skills in order to succeed. I also think another challenge is really finding the people that you want to work with. So I think in any role, relationships with people is often the hardest part of your job. I think that’s no different in biostatistics. You need to have great collaborators and great people that you work with in order to have a productive and an interesting job. But sometimes finding those people and finding the people that can support you and your goals, it can be a challenge. But overall, I really enjoy what I do. I think it’s a really great field to be exposed to a lot of different projects. I think I’ve heard people say before, the best thing about being a statistician is that you get to play in everybody else’s sandbox. I think that’s really true. I was never going to be able to… I had not very much interest in becoming a clinician or doing research like that, but I love to be able to help other people to get to that point. And I think you can always get involved in the fun bits of the project, the data, you understand, you get the result, and you can help people make understandings of the world, which I think is really a great thing to do. Thank you for sharing. That’s definitely one question I should have asked.

Jocelyn: Well, thank you, Anna, for being with us in the Biostatistics Podcast. It’s great talking to you.

Anna: Thank you for having me.