Jo Carlowe Hello, welcome to the Papers Podcast series for the Association for Child and Adolescent Mental Health, or ACAMH for short. I’m Jo Carlowe, a Freelance Journalist with a specialism in psychology. In this series, we speak to authors of papers published in one of ACAMH’s three journals. These are the Journal of Child Psychology and Psychiatry, commonly known as JCPP, the Child and Adolescent Mental Health, known as CAMH, and JCPP Advances. Today, I’m interviewing Postdoctoral Early Career Researcher, Lizél-Antoinette Bertie and Professor Jennifer Hudson, both of the School of Psychology, University of New South Wales, Sydney, Australia and the Black Dog Institute at the University of New South Wales. The Black Dog Institute investigates mental health across the lifespan. Lizél and Jen are authors of the paper “Editorial Perspective Extending IPDMA Methodology to Drive Treatment Personalisation in Child Mental Health,” recently published in the JCPP. This paper will be the focus of today’s podcast. If you’re a fan of our Papers Podcast series, please subscribe on your preferred streaming platform, let us know how we did, with a rating or review, and do share with friends and colleagues. All listeners to this, and indeed any of the ACAMH podcasts, are eligible for a free CPD certificate. Do please visit acamhlearn.org for details of this, together with information on how you can access hundreds of hours of free talks, lectures, interviews, all of which you can also get free CPD certificates for. The URL is www.a-c-a-m-h-l-e-a-r-n.org. Lizél and Jen, thank you for joining me. Can you start with an introduction about who you are and what you do? Lizél-Antoinette Bertie Thank you, Jo. My name is Lizél, and I’m a Postdoctoral Fellow in Child Mental Health at Black Dog Institute and University of New South Wales, and my research focuses on improving personalised mental health, but in a data-driven way. Professor Jennifer Hudson I’m Jenny Hudson. Thanks so much for having us on, Jo. I’m the Head of Child Mental Health Research at Black Dog Institute, it’s in the Faculty of Medicine at UNSW and also part of the School of Psychology. So my research that I do alongside of Lizél really focusses on trying to improve mental health outcomes for kids and families in the community, and we conduct a lot of clinical trials, we develop new treatments, assessments, preventions and early interventions with really a focus on school-age children and adolescents with anxiety disorders. Jo Carlowe Thank you very much. So, today, we are focussing on your JCPP “Editorial Perspective Extending IPDMA Methodology to Drive Treatment Personalisation in Child Mental Health.” Before we go into the detail, can you define and briefly summarise how individual patient data meta-analysis IPDMA works? Professor Jennifer Hudson IPDMA is really a specific type of a literature review process, really, a specific type of meta-analysis. So, typically, a meta-analysis review provides Researchers, Clinicians, as well as the broader community, really an opportunity to synthesise a heap of research on a specific research question. And the idea is that it’s a systematic process, that Researchers scour the literature using strict inclusion/exclusion criteria, and the idea we want to arrive at a set of studies to synthesise all the findings together. So Researchers who conduct a meta-analysis, the idea is that you extract that data from that selected group of published papers, and what we extract from that data is means and standard deviations so that we can arrive at effect sizes. And then we pull that information together in a way that can help us really understand what should be closer to a true effect rather than being, you know, reliant on a particular sample or a study. But an IPDMA differs really at the point of data extraction. Instead of extracting study-level data, so means and standard deviation from published papers, the idea of an IPDMA is that you ask the study authors to provide de-identified data from each participant and then – so instead of, kind of, pooling the average of average effects, you actually are pooling the individual data. So it really allows for a much more nuanced approach to data, it kind of allows us to take more than just an average effect across all participants, and giving us a larger sample size. So we then have the capacity really to do a much deeper dive into the dataset and answer much more useful questions about specific groups and specific treatments. Jo Carlowe I’m just wondering though, can you talk through the limitations of randomised controlled trials, systematic reviews and conventional meta-analysis, in terms of answering research questions about what works for an individual? Lizél-Antoinette Bertie Typically, the interventional trials focus on the average outcomes of treated children and adolescents using indexes like remission or response, and that’s really important when trying to understand how effective the treatment is, and also, of course, which factors are associated then with better or poorer outcomes. A systematic review will then combine the information from randomised controlled trials and draw conclusions based on their findings, and a conventional meta-analysis, as Jen was saying, will then synthesise those study-level effects. But none of these methodologies give us more information or explain why children respond differently to treatment. Professor Jennifer Hudson So I suppose individual clinical trials, they’re typically, you know, designed to have the power to detect group differences over time. So, clinical trials are really expensive, they take a lot of time, they’re really labour-intensive, and it’s really hard work, I suppose, achieving the sample sizes that are required, but that clinical trials really help us to answer questions like, “Does Treatment A work better than Treatment B?” but it’s for all children, so it averages without the effect. So it’s very much a one-size-fits-all approach, and of course, as Researchers, you know, after we’ve conducted these clinical trials that help us to answer these specific questions about, you know, treatment efficacy, we often want to do more of a deeper dive into answering other questions. But we do post-hoc explorations really, because the studies are really only ever designed to detect that main effect of, you know, “Does Treatment A work better than Treatment B?” We will go about asking questions like, “Well, what about the effects of this treatment for younger kids in the sample?” or “What about the effects of the intervention for families from non-English speaking backgrounds?” But these questions really – the useful questions that we want to be able to know the answer to as Clinicians, because we do see, you know, a very diverse group of children and families. But the way that we approach that typically in clinical trials, it’s always exploratory, because we never really have the power to answer that question in detail. Particularly if you’ve got, you know, in one clinical trial you only had a handful of, say, six-to-eight-year-old children or, you know, a handful of people from non-English speaking backgrounds. So, that approach really is underpowered and has produced really a mess in trying to answer what works for whom. You know, each study finds something different. When one subgroup effect is found it’s hardly ever replicated in the next study and so we just, kind of, have this, you know, very difficult set of findings to understand. Jo Carlowe Let’s go into the detail of the paper itself. So can you give us a brief overview, telling us what you looked at and why? Professor Jennifer Hudson Well, we’ve been on this quest, like a lot of mental health Researchers, to really find the holy grail of personalised treatment, you know, what treatment by whom is most effective for an individual with a specific problem and under what sort of circumstances. This is, kind of, the question that we’ve been trying to answer for a long time in psychotherapy research and mental health research, and this kind of personalised approach has been really successful in cancer treatment, but really in mental health research we have not yet found this elusive holy grail. So this, I suppose, could be for two reasons. It could be, one, that this holy grail that we’re in search of doesn’t exist, despite our desire for trying to personalise mental health care, and it could be actually that modifying treatment components based on different aspects of a child’s presentation will in fact not lead to better outcomes, where an answer to, you know, the “What works for whom?” question is that, you know, most things work for everyone. You know, many of the meta-analyses of psychotherapy have pointed to this quite unpopular answer, and I’m not really ready to swallow that bitter pill, because, you know, the second reason why we might not have clear information about what works best for whom is that we haven’t really tested this question with rigorous methods and data. And to design a study to answer these questions rigorously would really be a lifetime, a couple of lifetimes’ of work, if we were to, you know, embark on that, you know, study as fully funded – you know, if we started today. So instead of doing that, what we have done in this project that we’ve published on is actually pooling data from existing datasets. So we already have – we’ve invested a lot of time in a lot of clinical trials in child mental health, and a lot of families have provided data contributing to science. So we want to maximise the use of these data. So we conducted a systematic review of the literature and identified studies for us to include using that systematic approach that we were talking about. We’ve established a procedure, we’ve got data-sharing agreements from our authors and requested that data, and we’re trying to synthesise it. We’re at that point, and I’ll get Lizél to talk a little bit more about that in terms of, you know, where we’re up to. But that’s what we’ve been talking about in this particular paper and that process of bringing together this living systematic review, or a repository, a topic-based repository. Jo Carlowe Lizél, do you want to add anything to that before we go into the details of the findings? Lizél-Antoinette Bertie Just a little update on where we are at. So we’ve conducted the review, and to date we’ve identified 350 studies that meet our criteria, and of course we know not all of those will be accessible. They range from, I guess, about 20 years’ spectrum, so, you know, some of them might not be available to us. There might be some limitations to the sharing of data from some of the authors, but we have collected to date 91 studies and we’re in the process of harmonising the data for that. It equates to around 7,000 participants in this pooled dataset, so that’s a huge number, making this one of the largest combined datasets in the world. And we’re very excited to get to this point, you know, where we have something to show for the hard work that we’ve put in over the years. Jo Carlowe And my understanding is you chose to focus on anxiety disorders in the context of youth. Why specifically anxiety disorders? Professor Jennifer Hudson Anxiety disorders in kids and teens are still overlooked a great deal. Despite the significant contribution they make to the global burden of disease, it’s still a disorder that is not really focussed on a great deal in research, and there’s popular belief really that mental health problems don’t really start until adolescence or adulthood. So this, kind of, childhood/early adolescence period is often, you know, really overlooked. But we know that when you look at the development of mental health problems, that particularly for anxiety disorders, they start in childhood, not in adolescence or adulthood. So we really want to be investing in improving interventions for children who present with common mental health disorders like anxiety and delivering interventions when they first need it, not, you know, 20/30 years later. So, we think this area of anxiety disorders, because it’s overlooked, deserves more attention, and also, you know, we’ve been working with this team of Researchers for a really long time now. We’ve been meeting at conferences, international conferences, and it’s a really nice group of people. We work with kids, we’re fun and friendly, and mostly, on the whole, we are quite collaborative as a cohort. You know, a strong cohort of female Researchers as well, I think, who really understand that it takes a village to get this kind of work done, and you know, as we work together we can answer more meaningful questions together rather than in our own silos. So there is quite a really nice collaborative spirit that we’ve, kind of, built together in the child anxiety field, and this has really helped to set the platform for this work to be done and for it to be done well, because people are willing to provide their data, you know, we’ve got a process and we’ve got that trust built up with our collaborators as well. And, you know, there’s that – this goal that we want to be able to deliver effective personalised care for kids, when their anxiety disorders commence, and offer something that’s tailored to their needs. Jo Carlowe Thank you. What key points would you like to highlight from your editorial? Lizél-Antoinette Bertie I think one of the first points – although this is not the only way to establish a data repository, we wanted to follow a rigorous methodology, especially if you think about what Jen said earlier. If indeed personalisation rests on the methods we use and the approaches we take, we wanted it to be as rigorous as possible, so we followed the specific approach that we mention in the article. It also ensures clear and transparent reporting, which is something that can be followed through the whole project, which is also really important in the field. It also facilitates a specific data structure, which is important, not only retrospectively, you know, pooling the data in a certain way and ensuring that there’s a standard structure for it, but also prospectively. We’re hoping that Researchers in the future will design their studies based on this structure that we’ve created, which will just make it that much easier for us to accommodate into this platform, going forward. Professor Jennifer Hudson And I think the point that I wanted to highlight was how much of a team effort this has been. So, I think, you know, so far, we think it’s been about seven years that we’ve been working together on this project. Professor Maaike Nauta from the University of Groningen and I started working together seven years ago, unfunded, to try and establish this piece of work and we both now have preliminary funding, and this has really helped us to bring along more team members and to really fast-track the process, and we’ve got an incredible team. Lizél’s been working with us for almost the whole time, I think, and we’ve got other team members who are authors on the paper, Wenting Chen and Bas Kooiman, and that’s really helped to establish this team that’s enabled this work to be done. Lizél joined the team four years ago, and she’s just submitted a PhD thesis on this PADDY project, and Lizél’s really been the passion and drive behind this piece of work. She’s taken the project to a whole new level, and I’m so grateful to have had the opportunity to work with her on this, you know, first as a Grad Student and now as a Postdoc Researcher. So it’s been a really exciting, fun team effort to work together on. Lizél-Antoinette Bertie Thanks, Jen. It has been a real joy to be part of the team. Jo Carlowe Lizél, you touched on the point that you would like other Researchers to work in this way, to use this design. Is there any further message that you have for Researchers, that they should – could take from your editorial perspective? Lizél-Antoinette Bertie Yes, definitely in the pipeline we have a number of interesting ideas to facilitate this transformation in our study design, and we are working on specific items that belong to this project codebook, how to code our variables, or specific methodologies that we will incorporate. They will be available to Researchers to read as a resource and to use in their planning and execution of their studies. Jo Carlowe Excellent. Just going into some of the details in the paper, you state in the paper that although treatments are frequently personalised in clinical practice, there is a lack of science to guide Clinicians in their adaptation of treatments for clients with complex or varied presentations. How do you envisage the approach outlined in your paper moving the field towards evidence-based personalised mental health care? So in other words, how might your viewpoint be translated into practice? Professor Jennifer Hudson Yeah, that’s a really good question, ‘cause we really desperately need science to drive personalised care. Some of our efforts really can be translated into practice already. We can now predict which children are less likely to respond to standard treatments based on the, kind of, preliminary work we’ve done together building this dataset and other collaborative projects. So this is helpful for Clinicians to be able to say, you know, how likely a particular approach will work with their client population. It will also now, with the development and establishment of PADDY, be able to test out some of these early predictions that we’ve been able to make from the data so far, and potentially discover new predictors when PADDY is fully operational. But at the moment, this is the kind of point that we’re stuck. We can’t say what else a Clinician could try. So we have evidence about which children are less likely to respond to treatments, but we can’t say what else a Clinician should test out. We can’t say what would work better for a child who’s at risk of having suboptimal outcomes following standard treatments. So despite our best efforts, we really haven’t moved very far as a field in psychiatry and psychology in this regard. We’ve spent years developing new, innovative treatments that we think might work better, but we still roughly get the same average effects. Jo Carlowe From the research, which children are less likely to respond to treatment? Professor Jennifer Hudson So we know that children who present with more severe anxiety, higher levels of social anxiety and low mood, and also who have parents with psychopathology themselves, are less likely to respond to standard CBT. You know, that’s the bit that we can say, “Yes, we can tell you which children are going to respond sub-optimally.” But we can’t say at this point what else to do. It would be great for the science to catch up, and this is what we’re trying to do with PADDY, is to build this dataset where we’re bringing together, harmonising all of these datapoints, so that we can answer those questions in a better way, with larger datasets. So at the moment, you know, Clinicians really do their own personalising in treatment. They develop a treatment plan that’s based on a formulation, and this might be based on their training or their clinical experience or their beliefs. And, you know, they might do a bit of this treatment, and then if that doesn’t work they’ll try something else, they’ll adjust it. But really these adjustments are theoretical and not empirical. So our goal is with PADDY, is to be able to keep adding to this sample to make it more diverse, building up the subgroups within the population so we can really use this to be in a position to advise a Clinician the best treatment options for their clients. So if you want to think about an example, you know, maybe a Clinician who’s been referred a ten-year-old girl, she’s got severe levels of social anxiety, comorbid general anxiety, and comorbid low mood. Perhaps both of her parents have high levels of anxiety and they come from a low socioeconomic urban demographic. Maybe mum has past trauma and that young girl might be refusing school, and not responding to previous treatment that she’s been getting or support at school. Now, if you kind of consider another referral, let’s just say a 16-year-old boy with panic disorder and specific phobia of heights, he’s had no previous treatment. At the moment, what our data – you know, what the guidelines recommend is that the treatment for both of these cases would be identical CBT, and then if that doesn’t work, try antidepressant medication. And that’s about as personalised as we can get based on the average of averages that clinical trials and meta-analyses have produced so far. much more nuanced information when a Clinician does have a referral for a child that has much more unique or diverse backgrounds, they can be more prescribed and more personalised in their care. And that that can be driven by data rather than being driven by, you know, trial and error. Jo Carlowe That’s really helpful. It really brings it home, really, the limitations there. You’ve – a few times now you’ve mentioned PADDY, which stands for the Platform for Anxiety Disorder Data and You. In the paper you discuss the need for the formation of a topic-based data repository. Can you explain why this is important and illustrate how this has worked with the development of PADDY? Lizél-Antoinette Bertie Yes, so, firstly, Researchers, we’ve known for a long time that our typical analyses have not been sufficiently powered to detect these moderators, which are the factors that Jen has just described so well. You know, these factors that tell us which treatment will be better for the ten-year-old girl versus the 16-year-old boy, will lay the foundation for us for personalising treatment. So we’ve known that this has been a problem for a long time, and we’ve had some individual IPDMA studies that have been conducted, resulting in really great information and useful clinical guidance, but it’s never been used subsequent to that one once-off exercise. And we just thought that’s such a waste of fantastic resources. So building PADDY has been really important for us to overcome that limitation. It also, typically, for an IPDMA to be conducted it takes a huge amount of effort. As Jen has also mentioned, you know, literally years in the making, communication, resources from a time and skill perspective, just so we can use that data again. So we began to build PADDY, collect the data on childhood anxiety exclusively, which is – makes it a topic-based data repository, so that we could have those adequate sample sizes we need, and of course the statistical power to answer existing and newer research questions. Professor Jennifer Hudson For popular topics, like having this – like a – well, a topic-based repository, it really helps us to answer. You get questions. You know, Researchers tend to have specific interests. You know, “I want to answer this particular question, and so these will be my inclusion/exclusion criteria, and we then maybe want to include certain studies.” But there’s a lot of overlap perhaps between what I’m interested in and in what other Researchers might be interested in and what Lizél might be interested in. You know, and as Lizél was saying, you know, so much effort goes into building each individual meta-analysis or each individual IPDMA, that if we can kind of share that load, it reduces the overall workload, and you can reduce burnout in Researchers as well. It makes each clinical trial funder dollar go further as well, and also provide opportunities for other Researchers, within a particular topic, to be able to answer other questions that they might be interested in. Jo Carlowe In the paper you state that “The formation of a data repository contains ethical risks and logistical challenges.” Can you elaborate on these and talk through how such challenges can be met? Lizél-Antoinette Bertie Yes, so what we have learned from the literature, as well as from a couple of lived experience situations, is that the greatest concern is anonymity and confidentiality when it comes to participant data. So what we’ve done in PADDY is we’ve of course made a blanket statement and rule for participation, is to only share de-identified data. We also, as Jen mentioned earlier, have a data share agreement in place, so that will govern all kind of communication between the Researcher and the author and our team. And then for Researchers on the other hand, you know, they might be faced with not knowing whether they have an ethical obligation to share or not share their clients’ data. And that might differ from country-to-country, but we have established a procedure around that as well in PADDY, that we can share with the Researchers to guide them in the sharing of the data. Professor Jennifer Hudson So yeah, there are a number of logistical challenges as well to co-ordinating this large amount of data, and we need to – once we’ve identified the studies, we need to ask authors to contribute their data. And for a lot of Researchers that’s, you know, a big ask. You know, perhaps Researchers are really busy, maybe they conducted the research 20 years ago and it’s sitting on a hard drive. We’ve had, you know, people saying, “I’ve moved on several universities, it’s now sitting in a hard drive in a computer in the bottom of a hospital,” or, you know, that kind of reluctance to want to actually share their dataset in that, you know, maybe everyone else is going to have a look at how they construct their datasets and it’s a lot of chasing of Researchers as well. So there’s, you know, a time factor that Researchers have in actually being able to get their data in a point that they can share it with us in amongst their already, you know, overburdened, busy schedules. We’ve got to do a lot of chasing as well, reminding people. I had the same experience being on the other end of it when other people are asking for my data for an IPDMA or a meta-analysis. It’s like, ah, every time I see their name in my email I have this sense of guilt, I haven’t got back to them yet, and, you know, I kind of get that sense that we’re building that within our collaborators that, you know, every time they see our names they have this fear and guilt they haven’t sent the data on time. But, you know, it has been a lot of time chasing and following up, yeah. So this is why it’s great to use it from – rather than just for this, for a one-off IPDMA, which is often the way it happens with IPDMAs, you do all this work and then it’s just a one-off, you know. There is this expectation that IPDMAs should be shared and that, you know, most IPDMAs, you know, people will be looking at PADDY, and saying, “Well, you know, isn’t that normal for IPDMAs that, you know, you’re meant to make the data available.” But I suppose what we’ve done differently is that we’ve set up the process right from the beginning that the data are going to be shared and going to be made as a resource for everybody who’s contributing. Also that they’ll be, you know, able to use it and that we have those procedures for authors for sharing, and the process of updating it as well, so it’s not just a one-off. Jo Carlowe Lizél and Jen, is there anything else that you would like to highlight from the paper, or any further recommendations that emerge from it? Professor Jennifer Hudson You know, we really need to be more collaborative to advance the field, not just in child anxiety, but across mental health space, to be able to advance the field of personalised mental healthcare, to really answer this question about whether or not we can build the science to personalise care, or is it just that most treatments work for everybody? And to be able to answer that question, we’ve all been trying to answer it for a really long time, but in silos, and, you know, I think the idea of bringing this together as a collaboration, I think is a really good example of being able to advance the field by, you know, being more than just the sum of our individual trials. Lizél-Antoinette Bertie Yeah, and I think, just building on that, another part that is a highlight for me has been – it’s been very satisfying to interact with all the international collaborators. You know, so that sense of collaboration has really grown over the last four/five years, and it’s a network that’s ever expanding. And that’s just really been such a satisfying part of the project as well, getting to know each other, connecting on a different level, and, you know, connecting data from 12/14 different countries has just been fantastic. And of course, we’re hoping to expand that. Jo Carlowe Brilliant. Are you planning any follow-up research or is there anything else in the pipeline that either of you would like to share with us? Lizél-Antoinette Bertie Well, I have given you an update on where we are at. So we’ve completed our third systematic review and we’ll continue doing that. So, as resources and time allow, for now we still have our work cut out for us. We’re just harmonising our first batch of studies, and I think we have in our team a number of studies that are waiting for the data to be harmonised and, you know, situated on our platform. We have a system in place to make the data available to Researchers, and there’ll be a procedure involved, of course, for them to access the data. But it’s really exciting just to think about, you know, what PADDY has made possible for Researchers across the globe. Professor Jennifer Hudson Yeah, and we’re about to present this work to an international conference in a couple of weeks, an international conference in the US, so – and also having some collaborative meetings with our contributors as well. People who’ve contributed the data, and giving them an update because it has been, you know, a long road and, you know, they’ve shared their data with us and trusted us with their data. So it’s good to be able to provide some updates to them, you know, where we’re at and what we’re hoping to achieve. Jo Carlowe Finally, Lizél and Jen, what are your take home messages for our listeners? Lizél-Antoinette Bertie We hope that our listeners are as excited as we are about the opportunities that PADDY is creating and for me, I am excited about pursuing a more data-driven approach, which is really the focus of my research. Every time we have this conversation, I think of a quote in a paper that a Clinician said, “We have loads of ingredients, but we have no recipes.” That has stuck with me, and I think that’s – build my research around that. I would like to provide Clinicians with the recipes, in a data-driven way, to help them provide care for the children who need it most. Jo Carlowe Jen? Professor Jennifer Hudson I think one of the take home messages is around the model of data sharing that we have created here. Really show others how you can achieve more through collaboration and bigger data and how this can be done with sensitive data, with child data, with mental health data, and, you know, doing it in a way really that allows others to benefit from the hard work that goes into a clinical trial and an IPDMA. In particular, I think it’s going to be a real benefit for Early Career Researchers, and it will be a resource in the future, you know, we can set it up in a way that benefits all. So, another take home message is really around the need for the science behind personalised mental healthcare. From the research so far, you know, which clients might have less optimal outcomes, but at the moment we can’t tell you what else you can try. We can’t tell Clinicians what else to try. So perhaps for Clinicians as well in the meantime, taking a science-based approach to their own personalised treatment that they provide for their clients, you know, measuring and tracking their decisions, you know, and understanding their own decisions around personalised care, and in a way building their own science behind it, while we wait to catch up with the science to support you in those decisions. Another thing I just wanted to say too is a big thank you to all the Researchers who’ve been patient with us as we build this dataset slowly, and in trusting our – in trusting their data to us, and it’s been, you know, a real privilege to work on this project. Jo Carlowe Excellent, thank you ever so much, Lizél and Jen. For more details on Lizél-Antoinette Bertie and Professor Jennifer Hudson, please visit the ACAMH website, www.acamh.org, and Twitter @ACAMH. ACAMH is spelt A-C-A-M-H, and don’t forget to follow us on your preferred streaming platform, let us know if you enjoy the podcast, with a rating or review, and do share with friends and colleagues.