Why are clinicians leaving medical practice?

[00:00:00] Dear listeners, welcome to Faces of Digital Health, a podcast about digital health and how healthcare systems around the world adopt technology with me, Tjasa Zajc. By 2030, we are about to reach the workforce shortage of 10 million doctors globally, according to WHO.

[00:00:24] So one of the things that I've been wondering lately is not just how we can train more doctors, but how can we improve the retention of the existing workforce? So in this discussion, I spoke with Dr. Amanda Herbrand, a clinical data specialist at the University Hospital Basel and a former oncologist, about her insights on the transformation and the shift that we see in the healthcare workforce,

[00:00:54] the changing expectations of younger generations around work-life balance, and Amanda's vision on the future of health tech development that will be shaped by the collaboration between clinical experts, healthcare IT experts and individuals such as her, experts in healthcare IT who also have a medical background.

[00:01:22] Enjoy the show and if you haven't yet, make sure to subscribe to the podcast and also jump to our newsletter page, which you can find at fodh.substack.com. All the interviews are also published in a video form. So go to YouTube and check out our YouTube page if that's your preferred way of following content. Now let's dive in.

[00:02:01] Amanda, hi, and thank you so much for joining me on Faces of Digital Health for the discussion on clinical data modeling, physician careers and the future of workforce in healthcare. You're an oncologist by background, but you're actually working as a clinical data specialist at the University Hospital in Basel. So let's start with that.

[00:02:27] How did you transition from oncology to clinical data specialization? And maybe we can just clarify before that, what does it mean to be a clinical data specialist? Yes, I'd love to talk about that. So the transition was gradual and it happened by accident, I have to say. I was very interested in the way healthcare IT is working because I encountered that in my daily life, like how difficult it is.

[00:02:53] And everyone that has a smartphone knows like the state-of-the-art technology and it's still not just not there in the clinics. That frustrated me quite a bit. And just by chance, I got the opportunity to work on a software development project. And that was still part-time, like 50-50, two and a half days a week. I was in IT and two and a half days a week I was in the clinics. And then after one year, I realized, okay, it's quite hard to have this split and I have to decide what I'm going to do.

[00:03:22] And I felt like the point that we were at with the project at the University Hospital was just so interesting. We had just touched upon the topic of open air and I really wanted to continue doing that. So I made a decision and I said, okay, I'm going to go away from the clinic for now and then transfer it to IT full-time. What does that mean for your clinical specialization? Can you go back? What are the conditions?

[00:03:50] What's your personal interest into doing that? I could potentially still go back. I have quite a bit of time to do to finish my specialization because I've been doing research. I've been doing some full-time work, but also then moving to part-time. And so in the field of oncology where I was working, it would be six years of full-time clinical work to reach the specializations. And I think I would still have two or three years left of this full-time work.

[00:04:20] And for me, I realized where I am now. It just makes more sense for me personally. And I will probably not finish my specialization. So for me, it's like an exit. I chose to go to hospital IT and will probably stay here. It's definitely much needed for healthcare IT to have more people with the profile like yours.

[00:04:42] So people that understand the clinical demands and the clinical complexity and are trying to then transfer that to IT. But the reasons I asked you all these questions is because I'm really trying to research a little bit more what the causes are for clinicians to either leave the public healthcare system or to leave clinical practice altogether.

[00:05:05] Because we are approaching 2030 where we are supposed to have a shortage of clinicians of 10 million globally, according to WHO estimates.

[00:05:18] So just staying with the topic for a little bit longer, what do you observe among your peers or in the clinical practice in terms of the fluctuation of workforce or the expectations that clinicians have in terms of how their work should look like and how is technology basically supporting it? It's quite interesting at the moment.

[00:05:42] So I think my generation, so let's say 30 plus, you can see the shift starting from being really dedicated to work, to a place where you would like more of a work-life balance. So this, I think, started with my generation and it's even stronger in the younger generation. So the people that are becoming junior doctors right now, so they just have graduated medical school, are in their mid to late 20s. It's a really different requirement towards their work-life.

[00:06:12] So I have some reports from my peer group that are now consultant doctors and have to take care of these junior doctors where there really is a shift towards a, hey, I'm going to leave at 4 p.m., have fun with the rest of the work. So it's really a shift. So I think this is the first thing that we can observe in general. And then, of course, it's not just the wish for a better work-life balance, but also the tools that are currently,

[00:06:37] that doctors have to work with that are not sufficient in supporting an efficient, yeah, efficient work progress or work-life. So I think this is something that we will really have to deal with that healthcare is facing. And currently, I don't have a really good solution for it. We often hear that one of the reasons in the last years for burnout among physicians, especially in systems that are more digitized,

[00:07:04] is the workload with bureaucracy and the need to input data into IT systems. So my question isn't, is that a problem or not? It's more, to which extent do you think that this could be mitigated? And to which extent is this really the cause that drives people away completely from clinical practice?

[00:07:31] And where do you see that basically clinicians that decide that they want to find a work-life balance somewhere else, where are they going, if there's any observations that you have? So the mitigation, of course, could be systems that are more intuitive to use. So something that comes closer to the everyday apps that we use in our smartphones or that we handle, or we used to handle our life, be it banking or, I don't know, any kind of subscription contracts.

[00:07:59] Like the everyday things we by now really tackle a lot with digitalized tools. I think in medicine, the same thing has to happen, like the same level of intuitive mass, of integration, of user-friendliness. So I think there's still a lot of room to reach that. And where are the people going? It's a good question. I think a lot of them or some of them actually are going to the IT sector.

[00:08:23] So I've heard of some like either personal acquaintances or also through others where they really go to IT to either do a startup directly just from med school, never even really entering the clinics or then leaving after a few years because of this frustration with it. You're still in a hospital setting. So you're still in the clinical environment.

[00:08:50] So one thing that I'm also wondering when I'm listening to all these things that you're saying, I'm also between 30 and 40, so it's absolutely logical for everything that you said to me. But I'm wondering what kind of generational gap do you see among older physicians when they are observing these new demands and expectations of younger generations? How do you think, like what are the observations that you have on that note?

[00:09:21] I think it will become even more drastic the further the year, like the more, let's say, the Gen Z pushes into the workforce market. I think the stronger this will appear because we could feel the differences already between, let's say, us millennials and the older generation that is now currently having positions like the department heads or like leading positions in the departments. And I think this will become even more drastic.

[00:09:47] And I feel like we, like my generation should probably have a mediating role between those two because within the middle, we still understand the pressure and the kind of ambition and dedication to the subject. But also we would like to have a work-life balance. So we are in the middle. And I think this will be, it will be important that we soon have positions as well where we can live these role and make, yeah,

[00:10:14] a better communicator between the different generations and requirements. It's similar as you need to be probably the translator between clinicians and the IT department, which is likely very useful. Yes, and difficult. We have to get to an eye-level understanding where it's about no longer.

[00:10:39] So I think doctors are, and I can say that, I hope because I'm one of them, but I feel like doctors are a special breed because we have so much responsibility and so much knowledge, which I think sometimes can get to our heads where we feel like, okay, we know so much in our area of expertise that we can decide in other areas as well. But actually we cannot because we don't have this expertise in IT, for example.

[00:11:04] So I think starting to work with this attitude of saying, okay, you have your area of expertise, which is the medical world, and I have my area of expertise, which is the IT world. And I'm not going to tell you how to do your surgery and you're not going to tell me how to build a solution. I think this is the level of understanding that we will have to reach. And it will be difficult because this attitude has sometimes been lived for sometimes even decades, but I think it's a necessary step.

[00:11:34] One necessary step that we address as an important topic is also the increase in digital health literacy. So how are you approaching that at the University Hospital Basel? Do you have any additional ideas on how that could be approached? The challenges that we often hear about are that clinicians need to have the time to learn new tools, so that should be embedded in their working hours.

[00:12:03] Yeah, but it's also sometimes just the basic digital literacy of being used to using new technologies. What kind of observations do you have there? I think here is also a generational thing. So I think that the younger generations, they kind of grew up with technology, so very used to getting to use new tools as long as they're intuitive.

[00:12:26] I feel like most applications nowadays, they follow the same logic and are somewhat similar to you, so you understand them quite quickly. I think here, let's say older generations might be a different, like a difficult topic as well, because it's this mindset of we've done it always on paper, so why not continue on paper even sometimes, which might be more difficult.

[00:12:50] And another thing is digital literacy is one thing, but also making doctors aware of the intricacies of digitalization is sometimes not so easy. For example, explaining to someone that something like SnowMed City exists, which is this incredibly comprehensive terminology for medical terms. It would be great to get into med school, to have a subject that really explains about this the same way it explains about evidence-based medicine, for example, or statistics.

[00:13:20] That might be dry to some people, but really necessary to understand sometimes the literature that you're working with, and it would be the equivalent to that, but just for the tools that you're working with. But I think this might still be quite challenging because of the way the academic system is structured right now.

[00:13:37] So if, say, education about coding and SnowMed City and other data standards would be embedded in the education of clinicians, what do you think would be different for you, for example, as somebody who's doing research and analytics in a hospital setting?

[00:13:58] So I think if more doctors understood that it really impacts the whole downstream, if data being captured in good quality at the very beginning of the care chain, let's say.

[00:14:13] I think if this knowledge would be more widespread and people were more aware of this, I think we would have it sometimes easier in making suggestions for certain type of solutions, like entering structured data or dealing with terminologies. At least that's what I would hope is that the people would accept it more easily if they understood what the implications are. You are a clinical data modeler.

[00:14:42] Can we explain what that actually means? Because I think it's going to help us understand the importance of what you just mentioned. Clinical data modeling, to me, is translating the medical knowledge and also sometimes the medical process into a data structure. So making what's happening in the medical world, packing it into neat little blocks that you can reuse and then build a system with.

[00:15:10] So that every, let's say, every piece of information ends up in the right box inside the final database, let's say. And how is that in your specific example of the University Hospital Basel, what's the long-term vision of building out these clinical data models? You mentioned that you are working with OpenAir.

[00:15:33] So are you basically doing the clinical data modeling just to contribute to the broader database of clinical models in OpenAir? Or what's the direction that the whole institution that you work for is going into? So we would like to really stay close to the international standard. That means that wherever we can, we would like to reuse the available OpenAir models. So the archetypes, as far as possible.

[00:16:01] And then wherever we see there is a gap, so something does not exist yet, something has not been modeled yet, we model this and we lay it back to the community. So publishing it then to the International Clinical Knowledge Manager. And so we hope to just help expanding the library of available OpenAir archetypes.

[00:16:24] Can you explain from your clinical perspective, what is the most challenging thing for you when you're designing these clinical models? Who do you discuss? How good is the model or what needs to be changed? Who validates the structure that you designed? Just the whole process. And also maybe you can add to that, what kind of fascinated you about all these data structures that you decided to dedicate yourself fully to this?

[00:16:55] So wherever we can, we consult specialists about the validity of the model. So for example, we modeled some neurological tests. So our first step was to look at the literature. So we ourselves looked at how should this neurological test performed and then transforming it into an open air compatible structure. So finding the right data types to fit the steps of the examination, for example.

[00:17:22] And then in the end, having this nice mind map that OpenAir comes with, which makes it so tangible and approachable for clinicians. And then showing that to the specialist and saying, okay, so this is what we came up with. Explaining it and then asking, do you see something that's missing? Does it make sense? And sometimes even like building a very raw form with it so they can enter some data and see how it would play out. And if this has been validated, then we'll publish it to the CKM.

[00:17:53] And the second part of the question, why I was so interested with this. I personally really like structuring information. So this was also the way I was studying in med school. So I was drawing systems of how I would categorize, for example, the artery system or any kind of anatomical structures that I had a hard time studying. Otherwise, learning, memorizing by heart. So this, I don't know, it's fun to me somehow just making something easier.

[00:18:21] And I think it's about making things easier and making them more understandable, which I really like doing. If you try to put this into the context of the hospital, so how does all this data modeling go into the current healthcare IT infrastructure of the hospital? Are you only beginning with the structuring of data models? Are you already capturing any data with it?

[00:18:48] Can you talk a little bit about the practical use of this approach? So we have purchased now an open-air platform. So this is where we are organizational-wise. And the next step will be to implement it and then find functionalities to build on top. Currently, we have a very fragmented application landscape.

[00:19:10] So we have not yet a monolithic hospital information system or EHR, but we have smaller pieces that cover different areas but are not integrated well. So it's quite the Frankenstein's monster at the moment. And what we'll try to do is go process by process. So look at processes that at the moment are very fragmented in application, like application-wise, and then try to move those to an open-air native solution.

[00:19:39] So, for example, at the moment, our emergency room, they are using hardly one system, then printing something out on paper, continuing to work on paper, then scanning that later or manually putting it into another system. So this would be something where we would start and take this, because it's fragmented anyway, and try to make a larger piece that is open-air native out of smaller fragmented pieces.

[00:20:08] What will that bring in terms of the benefits for clinicians? So how do you explain to clinicians what's the long-term benefit for them? I'm assuming you're introducing new systems in order to ensure this new data capture, so it's good for doctors to know what's in it for them at the end of the day. We hope that we can achieve that by starting what their pain points were.

[00:20:34] So at the moment, these processes that we would like to replace first, they are driven by the urgency and the pain of the clinicians that have to use it at the moment.

[00:20:44] So we hope that we can really address what they came to us with and then say it will be a gradual process and not everything will be solved at once, but saying piece by piece, okay, by doing this, we can save you this amount of time or you don't have to manually transport the data from one system to the other anymore because now everything is here. And this is all done by my colleagues who are doctors as well. So they understand the pain points.

[00:21:12] They have worked in a very similar setting. So having this kind of eye-to-eye conversation, I think hopefully helps in getting further in making a solution. What's the timeline under which you are operating? So the roadmap, where do you want to be in five years' time?

[00:21:32] And also maybe it would be useful for us to mention what's the size of the hospital so we understand a little bit better how complex the project is and how many people work on it. Yeah. So maybe starting with the hospital first. So we are, I would say, a mid-sized hospital.

[00:21:49] So we are a university hospital, which means we have the full spectrum of treatment from emergency care through intensive care units, through highly complex therapies such as, I don't know, CAR T cell therapy or stem cell transplantations. We are about 800 beds big. We have about, I think, 8,000 employees.

[00:22:12] And we are located in the northwest of Switzerland, so very close to the border to France and also to the border to Germany. So we also have a lot of international patients. Timeline-wise, we have just procured the platform, which will be deployed until June, which is the roadmap. So in summer, we hope to have a fully functioning open-air platform. And then, of course, the next step will be to bring functionality on top of it for our users.

[00:22:42] And by summer, also, we would like to have a concept of how we would like to do that, so a roadmap. And I personally hope that by the end of the year, we have the first small working use case on the platform. And understanding the healthcare IT landscape and the importance of data models, what do you miss most?

[00:23:06] Or what are your thoughts about the changes that you currently see in Europe, the development of the European health data space, discussions about how the European health data space should be designed from the technical perspective? I thought it was very interesting when you mentioned that you're basically on the border between three countries, which also means that, like, how does that impact the language that you use when you're doing the data modeling?

[00:23:34] And how you work with patients at the end of the day if they speak different languages? Yeah. I mean, we are in Switzerland anyway, so technically we have, I think, three or four official languages. We are in the German-speaking part, but we have a lot of French or Italian-speaking patients as well. And Basel is quite the international city, so we also have a lot of English-speaking patients just being expats that work in the pharma industry. The solutions that we build will have also to be multi-language in a way.

[00:24:03] But I think OpenAvenue nicely accommodates that because it has these, like, inbuilt translation qualities. And I think the way we already are working also will be highly compatible with that because everything we do, or not everything, but a lot of what we do, we already do in a European-wide collaboration. So, for example, we currently work on echocardiography archetypes together with Sweden. We have worked on the neurological archetypes together with Luxembourg.

[00:24:33] And we have a collaboration for modeling with the region of Catalonia and Spain. So, I think we are very well equipped to whatever kind of requirements will come from the European health data space, because that's basically already what we're doing and what OpenAvenue really is also meant to provide. You just reminded me of a speaker that I spoke with in the past from Switzerland,

[00:24:57] who said that basically, if you can make it in Switzerland, you can make it anywhere just because of this issue with multilingual requirements. So, I think you have to be in multiple languages. I think so. I think at least, like, when it comes to, what are they called, the patient, like the informative letters that patients have to sign, for example, before surgeries, we are required to have them in multiple languages. So, we are already quite prepared for that.

[00:25:25] And, yeah, not just the languages, also the fragmentation of the healthcare system here in Switzerland. If you can make it here, it's a good statement. And you can probably make it anywhere. You mentioned collaborations earlier. So, with Luxembourg with Catalonia, how do you forge those collaborations? And what are the key benefits of them? And just to explain where I'm coming from, I'm curious to hear from you how much unification is possible,

[00:25:53] given that each hospital has different ways of doing things or follows different clinical guidelines, which can sometimes be a challenge when it comes to data capture and what you think is important. So, the collaborations, this is a nice thing about the open air community because it's really open. And we usually just reach out through forums. So, whenever we work on something, we post it on the open air forum and say,

[00:26:22] hey, this is what we're currently working on. And then usually we get reactions to that or just on the conferences and events. And it's a very nice community that, yeah, we feel is very open to collaboration because I feel like this is the challenge that everyone is facing, that there are not too many specialists doing that. So, whoever is helping is really, yeah, there's really a lot of synergies being found in sharing the work.

[00:26:49] And I think when it comes to, let's say, exchangeability of the models, I think medicine is pretty universal. Yes, there's different processes and there might be different workflows, but essentially what's being done or the data that is being captured, I think are more or less the same. Maybe sometimes in a different granularity or on another process, as you said.

[00:27:16] But I think that's something that happens usually on the application layer, like the logic and not really the data. And whatever then really is like special administrative, I don't know, factors, attributes that you need, especially in your country, you can do that as well. So, maybe a bit of a philosophical question.

[00:27:41] Where do you hope to see healthcare data in 2030? And you can think of it from your institution perspective, imagining that you're going to be building up models and applications by then, but also on the broader level. What's your hope where we could get in Europe, especially? So, I think I am coming from two angles and one really is the healthcare professional angle.

[00:28:10] So, I would hope to really have at some point really good overviews over my patients. So, that basically I can see aggregations of relevant information about that patient, how I need it. So, not this kind of siloed thinking that we have at the moment, but really more aggregated views that also follows more the thinking process of a doctor.

[00:28:35] So, I can really see the relevant parameters that might lead to the next decision or a new idea that I would need for assessments or evaluations. That would be the healthcare professional view.

[00:28:50] And then, for me, the other aspect would be real-world evidence so that we collect data that is good enough, a quality, that we can actually understand that if the treatments that we are doing are having the considered or the desired effect. So, this is something that I think at the moment where still there is a gap.

[00:29:11] So, we have the trials in clinical studies, but sometimes there's a gap in understanding, okay, does this insight that we gain from a clinical trial actually replay in the real world?

[00:29:25] And being able to learn from that, from the real world information and really steer towards a more value-based healthcare where we can really reach value for our patients by what we do and not just make money with it or keeping them in a mill of treatment that they might actually not profit from.

[00:29:46] Yeah, it's quite fascinating to see how it's hard to translate statistical findings to the individual level because you never know if the specific patient is in the 20% of patients that didn't respond to something. Not to mention, yeah, the complexity of the human biology that we still don't understand. And there's so much that we don't understand.

[00:30:12] Actually, the previous episode was about the use of AI in the design of drugs in oncology. And I always find it so interesting when it comes to biomarkers to know that maybe 30% or a small percentage of patients might have a biomarker that's already discovered. And even if they have it, maybe one-third of those patients are going to respond to a targeted treatment for that specific biomarker.

[00:30:39] So maybe as a data scientist or a researcher and a data modeler and an oncologist, how do you see this whole issue around precision medicine and the data that's still missing to get there? And how does basically what you're doing impacting these issues? That's a tough one.

[00:31:03] Coming from oncology where there already is a lot of what they would like to call precision medicine, I think it's still very early stages of understanding like what's a prognostic marker, what's a predictive marker and how it all plays together. And sometimes we know a drug is doing something, but we really haven't fully understood what it is actually doing. And I think there is probably more that we don't understand yet than we have understood.

[00:31:34] So I'm not too hopeful about precision medicine, I have to admit. Why not? Why not? Because of... Because I think biology is so complex and I really think we only scratch the surface of it. And if we look, it's a whole topic of itself. But if we look at how little we have actually researched and understood about the very basic topic between what impact does a drug have on a male patient versus a female patient?

[00:32:02] And it's like very basic biology that we haven't even really touched yet. This is, I think, where we might start with precision medicine, right? Because it's 50-50 of the population more or less. So I think it's quite... Yeah. How can I say? It's... We have a lot more to uncover. We have a lot to cover. And just by finding one marker and saying, hey, I can predict something with it, I wouldn't trust. We can do some things, but there's still so much more.

[00:32:33] Do you have any last thoughts in terms of data science, data modeling, healthcare IT, anything that you would say to anyone that's interested in the field or clinicians that hate everything that's related to healthcare IT? I think we are living in very interesting times. I think this might be the brink of the next generation of medicine. So there's a couple of things coming together.

[00:32:58] One is the, let's say, just basic IT finally being ready to deliver a structure that might live up to the requirements of all these different aspects. But also, of course, AI coming into play. I think now is the time to bring these pieces together and use them where they make sense. And the way it's evolving at the moment, it's also hard to say where we will be in 20 years.

[00:33:27] You've been listening to Faces of Digital Health, a proud member of the Health Podcast Network. If you enjoyed the show, do leave a rating or a review wherever you get your podcasts, subscribe to the show or follow us on LinkedIn. Additionally, check out our newsletter. You can find it at fodh.substack.com. That's fodh.substack.com. Stay tuned. Thank you.