All statisticians in the Pharmaceutical Industry will be impacted by the new EU HTA regulation. Activities around HTA submissions will happen earlier than currently, in parallel to the regulatory approval process for marketing authorization. There will be an increased scope of evidence for the joint clinical assessment to fulfill the needs of all EU member states, and so there will be a large package of statistical analyses that need to be provided in addition to the submission to the regulatory bodies.
This will redefine how you, as a statistician, work, and with whom you need to collaborate. Both HTA and clinical development statisticians will need to join forces to define the value story for the complete lifecycle of the drug. HTA specific analyses will need to be planned in parallel with clinical development.
In this episode, we will discuss the future skill sets that statisticians in the pharmaceutical industry need to adopt with the new EU HTA regulation being applied in 2025 already.
He is currently HTA and medical affairs statistician at Daiichi Sankyo and is actively involved in the PSI/EFSPI HTA Special Interest Group. Before joining Daiichi Sankyio, he worked as Real-World Data Scientist at Merck Healthcare KGaA and in academia, specializing in observational studies and complex event-history analysis. Arthur Allignol obtained his PhD in biostatistics at the university of Freiburg in 2013.
She is currently leading the Global MarketAccess&HEOR Biostatistics team at Bristol-Myers Squibb, and is actively involved in the PSI/EFSPI HTA Special Interest Group as well as the EFPIA working group on the EU HTA. Before joining BMS, she worked as a statistician in clinical development as well as in Medical Affairs. Katrin holds a diploma in biomathematics and obtained her PhD in data sciences at the university of Marburg in 2006.
Future implications of EU HTA and how Next Gen get involved
[00:02:07] Alexander: Welcome to another episode of The Effective Statistician. Today, I’m really happy to speak with Arthur and Katrin about HTA again. And today, we’ll talk about the future implications of HTA and also about the next generation of statisticians and what they need to be successful in this area. Probably also what people currently need to be successful in this area.
[00:02:38] So even if you’re not, next chance and this is for sure something that is of interest for you. But let’s start first with a short introduction of the guests today. Katrin, do you wanna go first?
[00:02:50] Katrin: Yeah, happy to go first. Hello, Alexander. Good morning. So I’m Katrin. I’m leading the group of statisticians responsible for HTA Market Access and HEOR within Bristol Myers Squibb.
[00:03:02] And my team is dealing with all additional analysis, which are not part of the [00:03:06] CSR, which are needed for market access purposes, for HTA submissions, for economic modeling. So all of, all those kinds of analysis on the clinical data. I studied Biomathematics and I have a PhD in Data Science, so a different field.
[00:03:22] And then I worked for a CRO. I was self-employed and I’ve been with BMS since 2014 working in the HTA space. And happy to have the discussion today with you, Alexander, and with Arthur.
[00:03:22] Alexander: Okay, very good. Arthur, what about you?
[00:03:35] Arthur: Yeah. Good morning and thank you for having me. I’m Arthur. I’m a statistician working at Daiichi Sankyo Europe.
[00:03:43] I am also responsible for HTA and Medical Affairs Statistics. And for what concerns us today, my main role is also to deal with all the supplementary analysis that are necessary for HTA dossiers and make sure our deliverables that are not part of the main clinical trial [00:04:04] analysis. So I’m a statistician.
[00:04:06] My training, I did my PhD in Freiburg, Complex Event History Analysis, then moved to Ulm for post doc. And then I joined Merck in Darmstadt as a data scientist in the Global Epidemiology Group, working on all things real world data. And I joined Daiichi Sankyo in August 2020.
[00:04:30] Alexander: Okay, so in the middle of the pandemic . Yup. So let’s talk about future implications for HTA and Europe.
[00:04:41] And I think that is really important even if you are not located in Europe. This topic is important if you’re working in phase two, phase three, absolutely. And you’ll see from our discussions that will happen in the next minutes, that this is relevant for you also, if your main focus is maybe [00:05:04] FDA approval or EMA approval. So let’s start with that.
[00:05:09] How’s the landscape developing in that area? Why is there an upcoming more link between HTA and the clinical part?
[00:05:17] Katrin: So maybe I can start with that. So first of all, there’s a close link with respect to the timelines now which will be changing. So at the moment, each country has its own HTA process and timelines deviate a bit.
[00:05:30] But when you have to submit your dossier, you have to start the discussions on reimbursement of price negotiations. But normally, this all starts after at least CHMP opinion is granted. And the European HTA now foresees that the timelines are in parallel to the regulatory approval process. So the companies have to submit the dossier 45 days before CHMP opinion.
[00:05:55] So there’s a lot of work done really in parallel. There might be some uncertainty about the labels still. So you will have a lot of back and forth with [00:06:04] regulatory people with the development statistics, how to analyze the data. So there’s a close interlink with respect to the timelines, which will be new and a lot more alignment will be needed on that one.
[00:06:14] Alexander: Yup. And of course, there’s also the UK, which is now outside of the European HTA area that already has very early timelines. And other countries like Canada are also maybe earlier than that. Yes. That’s the timeline that’s very important.
[00:06:30] What else is important?
[00:06:31] Arthur: Well, so because the work we need to start earlier, what might happen is that we need to start and develop a strategy for the dossier with data potentially not fully read out. I think that potentially data strategy would need to be defined in a principle manner, this kind of decision path. What do I do if this endpoint does not read out as we wish?
[00:06:59] What do I do if the results are even better than what we expected? So I think that will also be [00:07:04] a big implication in the way we work and in the way we need to work with the other functions as well.
[00:07:10] Alexander: Yeah. And also in terms of resourcing. You need to start putting people on it much earlier than before.
[00:07:17] Katrin: Yeah, and maybe also collaboration starts even earlier because there’s also a strong focus on, in the guidelines now, on pre-specification. So you have to start thinking around the HTA process when you plan your trial already and how to deal with those kinds of things. So it will be a process different from what we, or what many companies or we do normally now.
[00:07:37] So we start thinking about, okay, what do we need for our dossier once the CSR is available, once the data has read out, but there will be much more earlier collaboration on pre-specification too.
[00:07:48] Alexander: That is absolutely key. I think if your phase three and in some cases even your phase two studies are not built for launching the product effectively, you may end up in a situation where you have regulatory approval, [00:08:04] but you can’t get reimbursement. And we all don’t want to end up in this situation.
[00:08:09] Can you give a couple of areas where this collaboration will be specifically important? Katrin, maybe you wanna go with that one.
[00:08:21] Katrin: Yeah, so of course I can do so. I think it’s really important when it comes to the definition of endpoints and testing hierarchies. So we now face the challenges or see that the different HTA bodies have different kinds of views on how to deal with pre-specification and hypothesis testing and multiplicity. But it looks like from the guidelines we see that there will be a closer link to what is pre-specified there.
[00:08:48] So you have to make somehow sure that your endpoints, where you want to show your benefit with or your clinical benefit in the dossier somehow within this primary hypothesis testing, or at least somehow pre-specified in a way that HTA bodies can [00:09:04] accept it. For example, if you think of Germany, surrogate endpoints are always an issue there.
[00:09:09] They are not accepted. So, it might make sense then to think about how to specify endpoints. But also, when it comes to subgroup analysis, we see that there’s a lot of focus on pre-specified subgroup analysis. Also, when we think of the different PICOs we might have, so the question we want to answer in the dossier for the HTA bodies when it comes to population, intervention, outcome, they might be based on subpopulations and there’s a lot of need for pre-specification to draw some valid conclusions on the data afterward.
[00:09:42] Alexander: Yeah, very good.
[00:09:43] Arthur: Yeah. I think also in the HTA process, some endpoints have more weight than what they have currently in the regulatory field. So thinking about the PROs, for instance, and the safety. So especially if you go to regulatory authorities with an early phase [00:10:03] trial, it could be at the moment that the PROs are not analyzed.
[00:10:08] Or they are analyzed more descriptively than the one you might want for an HTA dossier especially in oncology, some of the phase two trials might be single arm and the HTA bodies will be interested in comparison with their standard of care.
[00:10:25] Alexander: Yeah, so that’s a really important topic. I see that all the time at the moment in oncology, but also in rare diseases.
[00:10:32] That’s an important topic. You have maybe small studies or single-arm studies and or randomized studies, but the comparator is very small from a sample size perspective. And what do you do with that? What are the kind of hot topics in this area at the moment?
[00:10:50] Arthur: What do you mean in terms of the methods or in terms?
[00:10:53] Alexander: Yeah. And how do you overcome this kind of problem with one-arm studies and then small sample sizes? Lack of comparisons?
[00:11:01] Arthur: So the HGA regulation works, I use the PICO frameworks so we have PICO stands for Patient Intervention Comparison and Outcome. So where all the member states will be able to define which population, which and I’m reading the rationale, but which comparator and which outcome they will be interested in.
[00:11:22] So all of these patients, patient population, or potentially subpopulation countries might be interested in as well as all the comparators won’t be available in the pivot or clinical trial. I think the hot topics would be indirect treatment comparison. So being able to compare your intervention indirectly with the standard of care in the different countries.
[00:11:50] What else? Potentially more use of real world data also to this aim.
[00:11:55] Alexander: Yeah, completely. I think this is the only way you can get to these points and fortunately, [00:12:00] some of the HTA bodies are much more, more open to that space already. I think the FDA is coming around a little bit at the moment, what I see and is more embracing this area.
[00:12:14] I’m not sure about the EMA, but for payers, for patients, for physicians, the end is very much about okay, I need to choose. Do I do the same as before or how much is this better? And yeah, single-arm studies, small studies might not be, it’s the only way to get to that. And it also reinforces a really good collaboration early on in the development process so that you can actually do all these kinds of different things that you work with someone that has access to registries, access to real world evidence.
[00:12:48] And you have the data set or the endpoints that it’s maybe collected there also in your studies. That is so important and absolutely speaks to the points that kind of like [00:13:00] when you think about early phase and late phase, how their statisticians collaborate. People working on the HTA and medical affairs part need to work more or less in parallel.
[00:13:12] These are phase three, phase two, phase three statisticians.
[00:13:15] Katrin: Exactly. And then we think of the methodological discussion about indirect comparisons. So we see at the moment a strong discussion of population adjusted methods versus methods using individual patient data on both sides. And we see that there is a favor for methods who use individual patient data, preferably score-based methods to pre-specification of potential confounders.
[00:13:40] There’s a systematic literature search. So there’s a lot of things to do and plan ahead. You need to make sure you have all the data access to your, if you compare that as real world evidence to your registry. So you have to have the IPT in-house to do that. So there’s just a lot of discussion ongoing at the moment on this kind of methodology and how to do that, and [00:14:00] what kind of indirect comparison in the end gives us valid results to draw conclusions on.
[00:14:05] Alexander: Yup. Absolutely. Now, another interesting area is who do we work with? So we talked a lot about the kind of collaboration within the statistics functions. How about the collaboration outside of statistics functions? How will that change and what are key stakeholders we need to be engaged more, much more with?
[00:14:26] Katrin: Very important point, Alexander. So there’s a lot of stakeholders statisticians have to deal with inside, but also outside of the company. So inside of the company, it’s internal colleagues working in market access, working in commercial, so dealing with the HTA submissions, working in health, economic outcome research.
[00:14:45] So a lot of people who are not trained in statistics, and so communication skills and translation of what they mean, what they need translated into a statistical question and then vice versa [00:14:57] is very important. But there are also a lot of stakeholders outside of the company like HTA bodies during scientific advice.
[00:15:04] So there’s a lot of different stakeholders, not just the statisticians talking to each other, speaking the same language.
[00:15:10] Alexander: Yeah. And then you have external key opinion leaders. You may have patient advocacy groups. There’s a whole kind of change of things. And what makes it even more complex is you’re not just dealing with a global organization here if you’re working in a pharma company.
[00:15:25] But you also have all the affiliate work or maybe a European region or an Asian region or whatsoever, there’s much more kind of people. If you’re thinking about working in phase two, phase three, you have this, the study teams and that’s about it.
[00:15:43] Yeah. Then when it’s in this space, it can, you easily get into the hundreds of people you need to basically work with. And depending on the company size, of course. Yeah, if you’re a small startup company and [00:15:57] you just launched your first product, it’s maybe not that kind of large. But if you’re working for one of the bigger ones, like BMS, GSK, Lilly and so on, that can easily go into a lot of people.
[00:16:11] What are the consequences for our statisticians in terms of that? And also, what does it mean for people that wanna join this part of the statistics area? What does it mean for them?
[00:16:23] Arthur: I think you should be able to, you have to discuss with a lot of different people, with a lot of different backgrounds and with a lot of different knowledge of statistics so be able to adapt your language.
[00:16:37] Alexander: What other areas of statistics are important beyond, let’s say, it’s typical clinical trial work?
[00:16:43] Arthur: I think potentially in the HTA EU, you would like to consider more evidence than just the clinical trials at hand, so that [00:16:55] some knowledge of epidemiology and non-interventional studies might be extremely useful besides the clinical trials.
[00:17:04] Potentially, real world evidence will take up more space as well. At the end of the day, it’s also linked to epidemiology, that’s this knowledge of non-interventional studies.
[00:17:14] Alexander: Katrin, you have a PhD in Data Science, if I remember correctly. So not clinical statistics, I would say. How has that helped you in that space?
[00:17:23] Katrin: I have to say it helped me a bit to really have a more focus on the broader data space. And then as a statistician, I have the feeling that people are focused on the primary hypothesis testing, the alpha adjustment, and they run the test.
[00:17:41] The trial is positive or negative. And they really just focus on this piece of statistics. But as an HTA statistician, you really want to draw conclusions on how huge your benefit is versus the comparator. [00:17:54] So I have a much broader picture on different endpoints. Not focused so much on hypothesis, but really looking for signals, interpreting, also secondary exploratory endpoints and draws.
[00:18:06] Draw a conclusion of all the science, the signals in the data there. And so this point of view from the data science perspective helped me a lot because they deal with the huge database and you also look for signals there. You want to draw conclusions and on what you see there. So this kind of you not focusing just on the classical statistics, hypothesis testing helped me a lot.
[00:18:26] Alexander: Yeah. You have a question and you look for what’s the best data to answer that question.
[00:18:31] Katrin: Exactly.
[00:18:32] Alexander: And what is a data scientist kind of approach and not so much like I have discretion what data do I need to collect to answer it?
[00:18:43] Katrin: But you have to be very careful. And I think it’s really helpful to work also then as a statistician, to have enough statistical rigor to do that because, of course, if you search long enough, you will always find a signal that shows what you want to find.
[00:18:56] I always say, if you always find Easter X, you will put yourself in the corner of the room. So yeah, and I think it really helps so also to have both sides of the view. And then try as much as feasible on pre-specification and pre-planning of the analysis, not doing really data-driven decisions, but having in mind that not everything is within a hypothesis testing framework.
[00:19:18] And that you have to take into account the whole amount of data you have.
[00:19:21] Alexander: Yup, that is really good. Awesome. I think that discussion clearly highlighted there’s much more collaboration needed between the clinical side to the FDA focus side and the HTA and reimbursement focus sides in the business.
[00:19:40] And both sides can benefit a lot from learning from each other, working closely together with each other because it starts already with the design of the studies. Phase two potentially, you said it becomes a pivotal study in the [00:19:55] end, like you know, oncology sometimes if you’re lucky or rare diseases, it’s so important.
[00:20:01] But also for all the other kinds of, let’s say, more common areas like where you maybe have more of the classical approach. It’s also important that you plan your study so that you can get the evidence you need for HTA submissions. And as we mentioned, pre-specification plays a much bigger role.
[00:20:20] So having a clear strategy on what you will do when also speaks to the points that you need to have HTA statisticians earlier involved in the process. And I think there can, there’s a lot of mutual benefit to each other. For example, if you think about network meta-analysis, if you have these in place early on, you can use that for planning your phase three study.
[00:20:45] Or when you have the readout, you can use another NMA to put your phase three study into perspective. Because what will happen if you have a phase three readout that is positive? [00:20:55] People will first say, “Oh, that’s great! What do we do against our competitors?” And of course, then having an NMA in place will be really key.
[00:21:06] So I think there’s a lot to work together. And last but not least, it will really strengthen the overall position of the stat function with the business because you are not just focusing on the R and D part, but you are focusing on the whole commercial company. Much more functions that you get involved in when, where you show how you can contribute value to the overall organization.
[00:21:34] Which in the end, I’m pretty sure will be very beneficial to statistics overall. And I personally believe the more influence these decisions have, the better, because that basically means that there will be more data-driven decisions. There will be more kinds of evidence instead of eminence-based decisions.
[00:21:56] And so I really hope we get there and in the end. Thanks so much. Any final things that you would like to give the listener?
[00:22:04] Katrin: So maybe I’ll give it a start. So my final thing I want to say, so it’s an interesting times, it’s a lot of possibilities now to shape this process, to be involved in this process. So I can only encourage statisticians to be curious to learn about HTA, to have a look at that field.
[00:22:24] So there’s a lot of things going on. And to work on communication skills because this will be really important in the future.
[00:22:30] Arthur: Yeah, these are all very good points. I can only just repeat what I just said. Something we do not mention is that I seem to be a good statistician and also understanding the needs of the market access teams. And one of these needs is the economic modeling that comes afterwards. So I think it’s also good as I see it, for a statistician to have some understanding of [00:22:56] queries and general economic modeling that comes afterwards for the pricing negotiation.
[00:23:02] Alexander: Yup, that’s a very good point. And stay tuned. There’s an upcoming episode with Min-Hua and again, Katrin. We will speak exactly as about this point. So, if you’re listening to this as it just comes out, you need to wait a little bit. You’re listening, this is maybe the other episodes are already live, and you can just scroll forward to it. Thanks so much again for this awesome discussion.
[00:23:29] And just one last thing, we haven’t talked about the HTA Special Interest Group. This is, of course, organized by the Special Interest Group and thanks so much for the overall kind of work on that. If you are interested in this area, then reach out. Everything you need can be found on the PSI firstname.lastname@example.org.
[00:23:52] Just look for the Special Interest Groups, SA the sixth [00:23:55] and there you’ll find the HTA SIG and all the work that they’re doing. Thanks so much. Have a great time.
[00:24:02] Arthur: Thank you very much.
[00:24:03] Katrin: Thank you.
Never miss an episode of The Effective Statistician
Join hundreds of your peers and subscribe to get our latest updates by email!