What does it take to develop good patient-reported outcomes?

Interview with Antoine Regnault

Why do companies want to develop new patient-reported outcomes (PRO)?
How does the process start when we want to develop a new questionnaire?
What are the key areas to look into for understanding the quality of a PRO?
What guidelines exist on developing new PROs?

In today’s episode, Antoine and I answer these questions and more. E.g. we talk about the importance of calibrating the relevance of the existing items in the questionnaire and developing new ones in order to reach more accurate measurements resulting in good PROs.

Stay tuned and listen while we also cover the following points:

  • How do capitalizing on existing instruments while improving them produce good patient-reported outcomes?
  • What is an “Item Bank”?
  • What is the data you need to gather first before developing new instruments and items?

Listen to this episode and share this with your friends and colleagues!

Antoine Regnault

Ph.D., Global Lead – Statistics; Modus Outcomes, A Division of THREAD

Antoine is a biostatistician with a strong interest in measurement science applied to patient-centered outcomes. As a consultant, he uses his combined skillset in psychometrics and statistics to support pharmaceutical companies with their analyses of patient-centered outcomes data from clinical trials and observational studies in a wide variety of disease areas (oncology, hematology, neurology and rare diseases, among others). Antoine has co-authored more than 40 peer-reviewed papers on analysis of patient-centered outcomes data and has been an active member of the International Society of Quality of Life (ISOQOL), where he currently co-chairs the Statistics Special Interest Group, and of several international research initiatives such as SISAQOL-IMI, SPIRIT-PRO and NeuroMET-2.

Cleanthous 2020 MG Symptoms PRO
Regnault 2021 MDS

Subscribe to our Newsletter!

Do you want to boost your career as a statistician in the health sector? Our podcast helps you to achieve this by teaching you relevant knowledge about all the different aspects of becoming a more effective statistician.


Alexander: You are listening to The Effective Statistician podcast, a weekly podcast with Alexander Schacht, and Benjamin Piske designed to help you reach your potential, lead great science, and serve patients without becoming overwhelmed by work.

Today I’m talking to Antoine about what it takes to develop good patient-reported outcomes. It’s a pretty cool episode so watch out for it. 

Patient-reported outcomes, the more you look into it the more interesting it gets, there are so many different aspects to it. So many things that you don’t learn about usually in University or when you just do the kind of standard high-level clinical trial design stuff. There’s a lot of really really cool things in it, and lots of work for us statisticians. So stay tuned for this really good discussion with Antoine. Speaking about PROs, PRO is always a topic for the PSI conference. There’s always something about the quality of life, there’s always something about patient-reported outcomes, and there is so much new stuff happening in this field. So check out the PSI conference at psiweb.org and see you in June in Gothenburg, Sweden at the conference. I’m really looking forward to meeting you face to face. 

Welcome to another episode of The Effective Statistician, and today I’m really happy to have Antoine here. How are you doing?

Antoine: I’m fine here.

Alexander: Very, very good, so today we are talking a little bit more about the patient side of things, and for our patients perceive to be, you know, good or the bad side of treatments and so patient-reported outcomes were always a really interesting topic for me when I started in the industry. I was working a lot in the Psychiatry area, and what the patients felt, though the experience was always a very important point to give because there are not so many kinds of lab values or other things that you can actually look for other than the safety side. So that was always a really interesting topic, I’m really happy to have you here. To introduce you to the listeners, maybe you can summarize a little bit how you arrived at where you are now? 

Antoine: Sure, so I was trained as a statistician and in fact, from the very beginning I would say I focused on patients and patient-centered outcomes and patient-reported outcomes. I did the Ph.D. where the topic was about patient-reported outcomes and the cross-cultural validation of bureau measures, this was 20 years ago. And from there I always worked in this field doing statistics and psychometrics on patient-reported outcome data, clinician-reported outcome data, anything that comes to scale. And I’ve been a consultant for industry all this time helping companies deal with their PRO data. 

Alexander: Yeah, and Psychometrics is actually a really interesting part. I only learned about this after University. Lots of different things I really didn’t know about, I knew about a couple of these things but actually more from linear algebra statistics. Yeah, that’s a really interesting field. But let’s start a little bit back to why actually do companies develop new patient-reported outcomes all the time. Why is that so much? Why don’t we have just kind of, okay indication X and you take these patient-reported outcomes and have a lot of generic ones that you can use all the time. Why do we need to develop new ones all the time? 

Antoine: Yes, it’s a good question. I mean, the question is, aren’t we reinventing the wheel? Well, I think if you come up in a broader sense, it’s about Innovation. I mean for imaging also we are always improving imaging MRI etcetera. And for PROs it’s the same, these are measuring instruments, we can always improve them. And, in fact, as you said in Psychiatry, I think we can realize the complexities to measure patient experience. 

It’s not an easy thing, and of course, we need to put in a lot of effort. So we have probably been developing instruments for more than 30 years. But there is still more to do, and in fact, if you think about it, this complexity is because there are two key concepts in our field which are concepts of interest and context of you. So what do you want to measure in what context? And the combination of these two elements makes a lot of different situations that require more precision or better. So for instance, the concept of Interest, you need to define as precisely as possible the concept you want to measure, and of course, some concepts are very widely measured in different areas, but still, they may not mean exactly the same thing in different areas. Physical functioning is measured in every condition, but maybe a physical function in one area is not exactly the same as in another. 

And then you also have a more complex concept: less static fatigue is a very important concept in many ways, more and more we see how important the concept is. But how do you define fatigue? What is it? For instance, we work on this and we published a paper last year about the conceptualization of fatigue and you realize it’s much more complex than that. It’s a very complex thing, but there are different forms of fatigue, physical fatigue, cognitive fatigue, elements like susceptibility of fatigue, how quickly you become fatigued like the battery of telephone, and to all these needs to be very carefully thought of and the more we learn the better we can measure and then we can develop new measures and the same for the context of use because you won’t use this necessarily the same instrument in different diseases, even if the same concept in different populations, depending on the severity of the disease. Like In early Parkinson’s, you wouldn’t necessarily need to apply the same instrument as of late Parkinson’s. Because what’s really important to patients at different stages of the disease is not the same. And you may not also apply the same instrument in the context of a clinical trial, and in clinical practice, or also depending on what the standard of care is, because if you want to compare it to another to certain other treatments, maybe the concept that you will want to differentiate with will not be the same or to all this together needs. Okay, we want to demonstrate our treatment effects. The first is to define what is a concept you want to assess? And what context? And with all this together, you can search whether there are existing instruments. But in fact, there is often not a good instrument for this specific context of use and concept of interest. So that’s why we’ll need to develop a new one. 

Alexander: Okay. Yeah, that makes a lot of sense. If I’m going to use something that was originally developed for Oncology, and now I’m going into an area where maybe the average person is 18 years old or something like this. It’s likely a completely different topic in terms of what a life of an 18 years old is probably very different and you know, stuff most of the oncology populations. 

Antoine: And, in fact, we see more and more new ways of approaching this question because I would say probably 10 years ago, we would not capitalize on the existing instruments and now more and more we capitalize on existing ones soon. Meaning that we look at the existing instrument maybe in other fields, or in the same place in the same area, but maybe for a slightly different population. We would try to start from there and maybe enhance it by adding new items or that’s the kind of thing we do more and more. We have a few examples like this where things in multiple Sclerosis. We looked at instruments that we use in fact in stroke first and then it was moved to multiple Sclerosis and it was about manual ability and we realized that we were lacking items to capture the very early signs and then you could go to add items to make the instruments better. So starting from what exists and improving it. And it’s also the idea of all the item banks that exist to the big item banks, in particular to Oncology. But also in other fields where you have existing items and you can fine-tune your instruments to make sure that you’re measuring exactly what you need for your specific context of use. 

Alexander: What actually is an Item Bank? 

Antoine: So an item bank is in fact the repository of items that have been developed in the field and had been calibrated so that you can create instruments that can be used in picking items that exist already and that had been put on a continuum. And in fact, it’s also the basis for computer adaptive testing. So basically, this is a more technological advance approach to measurement in which patients would not complete the full questionnaire or a fixed form. So busy, always all patients would complete the exact same items. They would complete items depending on the response they gave to the previous one so that you can reduce the burden of the patient because they just complete items that are fit for them. And in fact, there is an algorithm behind it so that it gives exactly what is the best question depending on their previous responses. 

Alexander: Okay, okay. So that is something that I actually haven’t heard about before because I’m still used to the paper forms, where it’s the same for every patient and you have these 10 or 20 items that you go through, by the way for the listeners, an item is a question usually with an answer that can be kind of scale from 1 to 5 or visual analog scale or whatever something like that.

Antoine: There’s no “A little, a lot, etcetera.” 

Alexander: So, yeah, that’s cool. When I want to develop a new questionnaire that’s because my drug has this specific advantage in this new concept. And that is not well captured or at least not kept shut into a specific context. How do you actually start with developing such a new questionnaire?

Antoine: Talking to patients. The first thing you need to do is talk to patients and I mean it’s kind of a bottom line in many cases but very important because what you want to measure is exactly capturing the very experience of patients. So that’s the main aspect when you develop an instrument, it’s talking to patients. So in fact what we really start doing and what’s really doing is to review the literature to make sure that nothing exists. But also review the literature, what could be the qualitative research literature? So really all the reports on studies that have in which patients have given all reports on their experience. So, you know, inpatient interviews, patient focus groups, anything that has been already published in this very area, but typically it wouldn’t be enough, you would have to set up specific qualitative research. Study where you would speak to patients to understand what is their experience, but what is exactly that perceived for this specific concept? So you would speak to anywhere between 21, probably. Yeah, depending if it’s a very rare disease you can go as low as 15 probably but the more patient you have, the more questions and feedback you have. And you want to also have feedback from patients with different perspectives or different severity from subgroups, maybe even for gender age, the group depending on the context of your disease and you get the most precise understanding of what it is that the patient has to go through and how you want to do this. So then once you have all this you create what we call a conceptual model. So basically we put the different types of all parts of the experience of patients into buckets, maybe symptoms. What kind of symptoms do they have gastrointestinal? Neurological? Etcetera. So what kind of impact do all these symptoms have on their life, to the called function, daily activities impact, social activities, etc. So we can get a full picture and then based on this, you can really focus on the development of the instrument to develop items, questions, and response scales that are good for all positions for this population. So that’s really what we will start with. 

Alexander: These focus groups have a discussion, a longer discussion with patients to learn about what they call these different things, what are these specific areas? And to see how these changes across the population of patients? Okay. 

Antoine: So, in fact, for the development of an instrument, we tend to prefer one-to-one interviews, but interviews can be for one hour with patients to get a sense of what they went through. And as you said, we also understand how they talk about it because then you can reuse the words they used to make sure that when you write an item because then you have to write this item. You have to write this question. You want to make sure that this is simply understood by patients. So you’re reusing the words that you have heard during the interviews is really, really important. 

Alexander: And if the patient says, this pain is still itchy or something like this and then you include this kind of feedback. 

Antoine: Yeah. And in fact what we do more and more now is to start collecting quantitative data at this stage, very early on. So with small samples that interview to get the first sense of how well these items would work, can get a third sense of this. So that’s what we call mixed methods research to really combine the data from the qualitative research and data from early qualitative research and data from the early quantitative research we can get that somewhere.

Alexander: So basically as a Statistician you’re involved directly from the start?

Antoine: Yes. Well, now we would be very much involved from the start. I would say that again, a few years back, it was really sequential. So qualitative research started really important. They probably already before but in early 2000, there was a lot of push for more qualitative research, but what would happen is that qualitative researchers would do their research, talk to patients, create the items, and then you would develop a big validation study in which the questionnaire would have been given to a few hundreds of patients. And then the Psychometricians would receive this and then off you go and start trying to understand how you would score instruments. So basically the score is when you aggregate responses to different item sets. To create the measure to the score, so it could be anywhere from simple to transform some, or even more sophisticated scoring. But it would be the Psychometrician who would do this. And then to reduce the number of items, etc. And create the score and develop and evaluate the measurement properties of the instruments. So that would have been really no sequential. Then slowly this tradition where the Psychometrician was involved more and more, earlier and earlier, so in the beginning, I was involved for instance, in the item generation process. To give us an eye in my opinion on, okay, you should avoid doing this kind of question, because when I will score it will create issues. So that was involved more and more. 

Alexander: What kind of issues were you thinking about then? 

Antoine: Like, for an instance, if you have this kind of question, one of the important questions was, do we want to add or not? None applicable response options because it could create issues depending on the scoring you want to do at the end. So this kind of item design question, also you don’t want to have items going into one direction in a positive up or negative, because then maybe you can have a method effect. You know this kind of thing that you know as a Psychometrician with some experience that will come up when you create through the score. You can tell from the beginning. Okay, this kind of item will not make it. So we started like this and now we have mixed-method research. As I said, it was more and more used and we have created some mixed methods. It’s really one of our areas that I really was really involved in mixed methods. And now we really push to have this from the very beginning to get to some data early on and so that we can also give inside feedback. So, for instance, you can tell beside them indoors, in fact, that very few patients will not make it, or you have an issue with your response scale. You don’t want to have 4 or 5 response categories, 4 will be really enough because this patient doesn’t really use all of them. So all these kinds of things you can tell from the beginning which kind of speed up the process because you have data very early on. So it’s not just qualitative, it’s also a first sense of how well it will work when you create your score and then measure your concept to do a lot of hosting.

Alexander: How do you decide on the number? I think I know two kinds of typical things, one is this kind of ordered categorical thing. Yeah, like my moderate-severe, very severe, things like that. And the other thing is the visual analog scale. Like if you have kind of 10 cm and, from no pain at all to the worst pain you can imagine and mark where you think the pain is. How do you actually decide at the beginning whether you use this visual analog scale or something like this order categorical scale? 

Antoine: I would say that we tend to see through and have fewer visual analog scales. But we see numerical rating scales which are, in fact, probably a little easier for the scoring algorithm in particular because it’s like a numerical rating scale and rs, where you say between 0 and 10 and you have to tell you to have 11 response categories, which come back. Which is much more a typical scale. So I think the choice between this is mostly related to the concept that you want to measure. Since the 10 the 11-point, an arrest for pain is a classic and I think it works pretty well. And I was always impressed by how well it worked because you can tell how a patient can distinguish between 11 levels of pain and I can tell you they can, that it’s possible. But when we try to apply this kind of response scale to other concepts, it doesn’t necessarily work so well. So, it’s really a question of the concept. And again, testing the question and seeing whether patients can discriminate between different levels and the methods that are more and more used now, which are more, where our modern Psychometrics have been doing, they are 15 years old now. Like item response theory or rash judgment theory. Really provide tools to understand response scales and really understand how the patients can respond to different levels of the categories. We can really give you a very good insight. And regarding the number of categories, there is a kind of belief that the more item categories the more responsive your scale is. And the fact it’s not always the case because sometimes the patient can’t really distinguish between the two adjacent response categories. In fact, it gives you more noise than anything. So it’s not necessarily a good idea. So we tend now to have fewer response criteria, which may be items that capture more subtle differences, so it’s not in the response category but in the question itself that you have the gradient of severity, it could lead to better tools and in fact also easier for the patient because they don’t have to think, I’m a moderate. Well, in fact, maybe it’s the same or your response option, like a little and quite bit for instance. Okay. What is the difference between a little and quite a bit? Well, that’s not easy. 

Alexander: Yeah, but if you have your more specific in terms of the item, is it burning pain? Or is it a different type of pain? And I think it is much easier to say. Okay. This is moderate or this is severe or this doesn’t have burning pain.

Antoine: There is a way to add a qualifier. So basically, for example, shortness of breath, you can have shown us abreast at rest. Can I be shown as abreast with mild exertion to link the working more to try to suppress when climbing stairs? And then the question itself gives you an idea of the severity much more than the response which would be okay. Do you experience shortness of breath? Well, depending on the situation and you should put the different situation, then you can have more breath of curvature of the concept which could be very helpful, and it is also very good because it’s easier for the patient and it’s also easier for interpretation because then you can tell, okay? In this situation there is a problem but not in this population. For instance, in UPD, patients can work, but they can tell this is very severe because they can’t even work, they have to stay home, and even worse there is shortness of breath even when they are resting is a very severe case.

Alexander: Yeah, I think that’s actually a very interesting point. I’ve seen that in, for example, Alzheimer’s. Yeah, if you look into some of these Alzheimer’s cases, you see that certain items only actually make sense. In certain severity of the population. Yeah, so I did some research on it, kind of how different, how much 2 kinds of items differentiate based on the total score. Yeah, and you can really see if you go from with a total score from no symptoms to little symptoms. Yeah, certain items actually really change where certain other items, they just are still at zero. There is still no change and speed like this. Whereas if you go to the more severe ones, yeah, all the earlier ones have already gone to the most severe category. And now the other items start to change. 

Antoine: Exactly. 

Alexander: And so I think that’s a really interesting concept that different items are applicable to different severities of the disease.

Antoine: Yeah, and it’s actually a very topical question in a very important way, because the more we advance, the more Innovation there is in drug development, the earlier we treat patients. In fact, the particle in origin is Parkinson’s Alzheimer’s. And the question is, can we use the instruments that were developed originally more for more severe conditions to capture change, or an early progression of the disease is less severe. And there you can see that this measure may lack sensitivity at least for these many patients. There was quite a lot of work on Alzheimer’s for existing instruments. We are actually part of the consortium where we are working with the Metrology institutes. So Metrology, the science of measurement. So it’s another Innovation, we try to bring in the field to work with people who are working with measurements. So people would define what a kilogram is etc. And to see how they do this, and whether we can do it also for our kind of measurement. And so we have this European initiative, which is called a neural net and where we look at the measurement of memory for Alzheimer’s and try to develop tools that would fit this kind of quality assurance that you can come with for Metrology. And one of the key questions is, okay. Can we separate between normal cognitive impairment and Alzheimer’s and have the tools to do this? This is exactly what you said. We’re seeing Alzheimer’s.

Alexander: Yeah, so being able to separate between different groups of patients is one thing. I think the other thing that directly comes to mind is that you can easily detect changes over time within a patient? 

Antoine: Yes. 

Alexander: What are those quality aspects you would look into when you have a new PRO that shows that it’s a good measurement?

Antoine: Well, it’s an art, which takes a lot of experience while I think in fact the first thing I always recommend is to read the questions. If it’s a patient-reported outcome, read the question. You can’t imagine the number of people who are using instruments and haven’t read the actual questionnaire. Yeah, and Even for a Statistician, analyzing data from control with those always recommended before you do anything, read the questionnaire because if you read the questionnaire you understand what it is about and you understand how you can use it and how we can analyze it. So I think it’s very important for us. Also in general because it can give you an idea of, if you can’t make sense of the question, you are sure that the patient won’t be able to make sense of it. 

The first thing is to try to put yourself in the patient’s shoes and read it and see whether you would be able to respond to this question. And sometimes, even for questions that are very widely used, I can tell you they are not so good because whenever you read them, they don’t make sense. So that’s true, then what I like to do is to try to talk with clinicians to get a sense of what are the clinical aspects and to see whether they are reflected from the patient perspective in the questionnaire. I think it’s very important to have both aspects of the question that needs to be important to patients. So it’s important that the questions are being developed involving patients. When you read the paper you need to make sure that patients have been interviewed, they have been really involved so that the questionnaire reflects patient experience. But also it’s important that it’s also in line with some clinical hypothesis or and that’s very important, so that’s that would be it. And then looking at the development of the validation papers to the segmented papers and I mean, validity, reliability has been analyzed for years, it’s very important. But I think that one of the key elements is targeting. So that’s what we have been talking about all along, to how well this set of items fit the population in the sample that I want to measure. So again, we’re modern Psychometrics, Rasch Measurement Theory, when you can’t compare where the items sit on a continuum compared to the sample because it’s a model that puts the patients while the opposite of the responders and the items on the symmetric. So you can actually see these items mid-match by population or not. And that’s in fact, this property drives a lot of things because if they don’t match, you are floor effect, so you won’t be able to have to make a change. You will not be able to discriminate your sample to question in terms of reliability. So really this question of how well the items match, the sample is a critical aspect because it’s in a clinical trial in particular. If you don’t have this, you won’t be able to detect the treatment effect. There are a lot of cases where it’s okay, where it happens. 

Alexander: So floor and ceiling effects are when nearly all your patients score the lowest or highest. Yeah, and then well if there’s no variability then well you can’t see anything.

Antoine: If your patients are at the maximum already, you won’t be able to improve them. So you need to have more items that can often extend the ruler. You have to go further. 

Alexander: Yeah, although if all on the lowest and you need to look closer and then see all the said items actually makes sense. Anyway, this is like in the Alzheimer’s area, if you look into a mild patient, certain symptoms and Alzheimer’s just don’t occur in these patient populations. So adding these is just adding noise and doesn’t help you and burn to the patient, by the way. 

Antoine: Yeah. It’s another big question, in fact, we already have this question. How many questions we don’t want to burden overburden the patients, which is true. We don’t want to have the patient complete a lot of questions that I would say if the question is not useful, because if the question and I think that’s a big issue in clinical trials, that we ask the patient a lot of questions, but maybe we don’t use all of them because, for instance, the full questionnaire that had been developed 30 years ago and has a lot of questions, but half of them are completely irrelevant to the patient population. So maybe instead of asking these 15 questions, let’s ask another 15 that are really relevant, and then this for me, is really a patient burden. 

Alexander: Yeah. I’ve seen a few protocols where it took the patient maybe half an hour to fill out the questionnaire. Yeah, and some of these are even kind of overlapping and you kind of go through and think, like, didn’t I answer that question already or something like in another fashion earlier? So yes, these can be quite burdensome, but you mentioned this new development you have targeted questionnaires by having kinds of questionnaires that change depending on how you respond to it. Like, do you have the symptom and then get out more questions about the suitable come up, things like that, I guess. So how do you actually develop that? That sounds really sophisticated. 

Antoine: Well, yeah, so this is inspired by other fields because you know the field you have this kind of system where you don’t have access to all the questions you can. Imagine a patient, you have to select health. It’s starting to be more and more used, to be honest. There are still some regulatory questions about it because if you don’t ask the same question to a patient, how can you make sure, I mean, the question is how you can you make sure that you can pool and compare that’s obviously an important question, but in terms of technique, it’s really based on this notion of Modern Psychometrics and again, the Rasch Measurement Theory is really important here because you would have a continuum and all your items are calibrated in that’s the keyword. They have to be calibrated. 

Alexander: What does that mean, calibrated?

Antoine: It’s exactly the same as any measurement instrument that you have when you have a scale in your home, they are calibrated. So that if you use this scale, you have one kilogram on this scale, it will be the same with another scale. So you have to have this process that makes sure that you what when you are using a new measurement instrument. You can trust the result and you can compare afterward. If you weigh two different objects, you can still compare the weight, the mass inside. And that is very important for any measurement instrument. And for PROs, we realize that it’s very important too, and this calibration means the item needs to understand what level of the concept is measured by each item, very precisely. So that depends on the response of the patient to this item. It gives you a precise idea of where they sit on this continuum on this metric. 

So, that’s where you have to run an analysis upfront to estimate the parameters of the items on a sample large enough to have a reliable estimate and with as little uncertainty as possible, and then you get this calibration. And then once you have this, as I said you can add an algorithm that can be applied depending on the response of one item of the patient when it basically says they are above a certain look threshold. You will ask this question and if they are below, you will ask the other one and etcetera. And so the more you go, the more precise your estimation of the person parameters to basically, where they sit on the parameters that say in the continuum, so basically, the score that you want to see, it kind of makes the score more and more precise and then at some point you say, okay is precise enough I can stop and don’t need to do to ask more questions. 

Alexander: So it’s basically like if I start with a very kind of wide question. Yes, it helps me to understand where on the Continuum the patient is, and then I see if he’s in some more mild area. Then I look into items that are more relevant. They are and more and more until these different things. That’s pretty cool. 

Antoine: Yes, it’s cool. Again, I can’t wait for the time when we will be able to implement this into your child because it will change things dramatically.

Alexander: Yeah, that sounds really cool. Yeah, and also makes it much more applicable to the patients. Because I think, you know, answering lots of questions that are not applicable, is kind of a little bit pointless. So, when we want to develop your guidelines and want to use them with the FDA or the EMA, not new guidelines, we want to develop new PROs, what kind of guidelines are out there?

Antoine: Well, the standard one is the FDA one which was published in 2009. The patient-reported outcome guidelines guidance, but more recently the FDA started, well still a few still a few years now, they have the Patient Focused Drug Development Program though they have put together a lot of effort to update this guidance. So now there are four work streams for prospects. One guidance is final, and there are three others that are ongoing, which will bring more information to print. For instance, introduce the notion of mixed methods. There is a lot of discussion about meaningful nests of how you interpret this course. That’s a key aspect for PROs because there is no metric, so how can you tell if this change is meaningful or not? There’s a lot of work around this, and there is a large part of one guidance, which is about this. It’s about data collection, it’s about developing instruments, it’s about interpreting instruments or implementing instruments, in clinical trials. And this is all going up in the PFDDs or Patient Focused Drug Development Program of ideas. So it can be found on the website. 

So it’s the most comprehensive guide. And EMA, there are a few aspects, but much less for development than using clinical trials. For instance, in Oncology, there is an appendix to the Oncology guidance, which is about quality of life in PROs that you can use in this specific field. So ICH also added some development of our PROs with patient-centered outcome research. So there are a few and it’s always evolving. And what we have now is documents, for instance, for protocols, there is SPIRIT-PRO. So basically, when you have a pure one point in your clinical trial protocol, what do you need to look at? So its SPIRIT-PRO papers. There is a Consort Pro for reporting, and there is currently an initiative, which is called the C-Circle, which is an international initiative that is looking at standardizing the way PRO endpoints are analyzed in concept trials in oncology. So, you see a lot of ongoing things. 

Alexander: That is brilliant, so thanks so much for this really great discussion. We started with where and why we actually need Patient-Reported Outcomes. And why this will never stop potentially and you know what is context and what a concept and what Statisticians and Psychometricians can bring to the table to ensure we have high-quality Patient-Reported Outcomes. And what is Quantitative and Qualitative research in this area and finally, what are the areas that you need to look into to make sure that you actually have high-quality PROs, and also, what are the regulatory aspects to it? So it’s just kind of thinking about all the different things we talked about. We probably can have a couple of other episodes about these topics.

Antoine: We can have series. But I think for me, the last point I want to make is, which is very much in line with what you’re always saying. This is key because this is how to bring the patient’s voice into clinical development. And I think that’s what we try to do and that we want to make sure that the patient is heard. The patient voice and we can really make clinical development for patients and that at the end their life is improved and we can demonstrate that the treatment improves lives of patients. And that’s why I’m so passionate about it because I think it’s really making a difference. 

Alexander: I completely agree. I was just yesterday with a Physician. And he asked me what kind of doctor you are. And I said, well, I’m a Statistician and I work in this field and we talked about how data is conveyed to Physicians and to Patients. And he said it’s so difficult to understand what actually all these different things mean. Yeah. And if we can’t explain to a physician, or a patient, how they will benefit, what they will expect from the treatment. That’s really bad because then they can’t make trade-off decisions. I can say, will I accept this kind of side effects? Or things like that because they can’t understand what would be the benefits. And so I think this is a really, really important area of research. And there’s also patients who are more and more taking actions and are much more kind of knowledgeable about the disease, and I think the time where you just go to your Physician and the Physician decides and you kind of adhere to it are largely gone, isn’t it? 

Antoine: Yes, that time is gone. Yeah, but it’s good, it’s very good. It’s a good adaptation and I’m looking forward to it. 

Alexander: Thanks so much.

Antoine: Thank you very much.

Alexander: That was an awesome discussion, and all the best with your career in that area keep on pushing these things because it’s really, really important. 

Antoine: Thank you, Alexander. 
Alexander: This show was created in association with PSI and the conference happens in June, don’t forget to register. Thanks to Reine who helps us with the show in the background, and thank you for listening. Reach your potential, lead great science, and serve patients. Just be an effective Statistician.

Never miss an episode!

Join thousends of your peers and subscribe to get our latest updates by email!

Get the shownotes of our podcast episodes plus tips and tricks to increase your impact at work to boost your career!

We won't send you spam. Unsubscribe at any time. Powered by ConvertKit