[MUSIC PLAYING] We are the Association for Child and Adolescent Mental Health, or ACAMH for short, and this is ACAMH Learn. Welcome to a new series of Mind the Kids. I'm Dr. Jane Gilmour, honorary consultant, clinical psychologist and Child Development Programme director at UCL. Toseed, a professor of psychology at the University of York, focusing on children and young people's mental health and special educational needs. In each episode, we select a topic from the literature, and in conversation with invited authors, sift through the data, dilemmas, and debates to leave you with our takeaways for academics and practitioners. Today, we're taking a deep dive into a really important topic. We're discussing different types of psychotherapy, what works and why. This episode is called Rebooting the Great Psychotherapy Debate. OK, so let's talk about talking therapy. So this is a really interesting conversation for me that we're going to have today because I have specific views that I've developed over time about talking therapies. Talking therapy is the go to approach for common mental health difficulties. So I think, and Jane, you can correct me if I'm wrong here, it's the first line approach when someone experiences mental health difficulties and they seek support. And so we're recording this at a time when it's Welcome Week at my University. And we had this session where we were talking to some students, a handful of students, and I was saying, why you want to study psychology? And one person said, oh, because I want to go into therapy. So then I started my little rant about my problem with talking therapy. And I was like, a lot of common mental health difficulties that we talk about on this podcast and that are established in the literature, their origins might be social. So things like bullying, adverse childhood experiences, poverty, stuff that happens in the environment around us that then contribute to mental health difficulties. And then what we say to the individual is, oh, I'm sorry all that stuff has happened to you. Now let's try and talk about that. But that situates the problem within the individual rather than trying to address that social structure. Yes. Right, of course. And we've talked quite a lot about these environmental issues, like poverty, for example, which has a pernicious effect on mental health. And yes, we also know that for depression, conduct disorder, anxiety, trauma and so on, there is an evidence base for psychotherapy. In depression in particular, there's a range of different psychotherapies that is very pertinent for our conversation today. And I take your point, but you could also argue that the idea of-- we're not situating a problem within a person. We're giving them the opportunity to feel better. We're giving them the agency for change. And in particular, for example, cognitive behavioural therapy changes the perception of an existing circumstance. So the power to change where environmental shifts aren't feasible, we are not in charge of the economy or that we are in charge. So what else are we going to do? We can make a change-- and I think it's about delivery. So for example, most of my clinical work is with young people with tics. And I would offer psychotherapy. So habit reversal training or exposure and response prevention. And I'm not giving the message I have to fix you. I'm giving a young person the opportunity to have agency over their body. And I think that has huge repercussions for a young person. So I think it's not either, or, I think it's both, and. Yeah. I agree with you. I think my point is more we spend a lot of resource, whether that's time that we spend talking about it or research or whatever it might be, on what to do after the fact. But actually there's lots of preventative stuff. And we know we're not in charge of economic policy, but actually we do have a role in advocating for that as well. Yeah. But I think this is not about diminishing the power of psychotherapy. I think it's about recognising that intervention can take a variety of different routes. And I think one of the things about Ian's paper that we'll go into a little bit more is it starting this debate about different sorts of psychotherapy and whether there is specific efficacy or effectiveness. And I'm really interested in these differences between psychotherapies potentially that are raised here. And it sounds to me as if you have a concern about all psychotherapies as a sort of philosophical stance. So I guess there's slightly different things to say there. Yeah, no, I'm not against it. Obviously it's necessary. I would just prefer that we weren't in the situation where we needed psychotherapy. But that's not where we are. But let's bring Ian in because I'm sure he has thoughts. So today, we're joined by professor Ian Goodyer from the University of Cambridge. Ian is the co-author of the paper Dynamics of Depression Symptoms in Adolescents During Three Types of Psychotherapy and Post-treatment Follow Up, published in the JCPP. Welcome, Ian. Jane, do you want to start us off with some questions. Yes, because I wanted to get just to the top of the concept and the ideas that you've raised in your paper. I wondered if you could describe for our listeners Walpold's great psychotherapy debate, and in particular what your thoughts are. Bring us up to date in the context of your recent findings. Well, as know, psychotherapy is not new. And what Bruce Wampold was doing after many, many years of highly respectable research himself, was trying to gather together a set of information, probably over about 50 years of work, his own and others, and try and put to a point to all of us that there were more similarities between psychotherapy schools than the psychotherapy schools themselves wished to acknowledge. So the point here is that, yes, they look different. They have huge theoretical differences. People have used anecdote and theories and some empirical work to try and emphasise the differences. The differences therefore influence how you train and educate people in the underlying theoretical frameworks of the mind that you need to be aware of. Trainings have got longer and longer and longer over the time, as though somehow the more time you spend in some kind of obedient form in the therapy that you're learning, the better you will be, for which there's very little evidence. And he came to the conclusion that we needed to have a reset. And he wrote a number of papers, but I think everyone will probably know the book that he published I think in about 2020 or so. I can't quite recall. Anyway, he had two basic premises. First of all, the great debate centres around commonality and disparity between the psychotherapies. Common didn't mean necessarily things that anyone could do, but it did mean things that therapies shared and therapists might share. And that common mechanism concept has actually been established time and time again since that book. Meta analyses by the Dutch group led by Cuijpers have shown that probably more variance is occupied by the common mechanisms than by the specific. There are problems with that because of the nature of the level of measurement and understanding about what could be common and what could be specific. But nevertheless, this is the great kind of debate that Wampold worked on. And he talked about dropping the idea of specific therapies, which suggested that there were modular mechanisms that each therapy uniquely had possession of that would make a difference to the patient. And he said, no, there are contextual factors that all therapists have a view on but they actually share. And you're all being a bit defensive about the situation. What really matters is that we understand the common features and then we ask the question, well, are there different specific issues that each of the therapists can bring in order to obtain a similar result? The concept of equifinality has actually never been questioned that much. That is to say you can get to the end road by different pathways, but you'll all end up doing something useful. And therefore measeurement, utility measurements, are being useful when actually show the same characteristics across therapists, across therapy, across therapists, in order to get the right outcome. So that begged the question, well, what are the common features? And it was interesting listening to the two of you talk about the environment versus the internal world because the common features contain both. And for Wampold, what was important, more important than anything, it appears in his writing, is the relational aspect of the therapist and the individual knowing the environmental context from which the individual comes. So I would say to both of you and to anyone listening, first thing you should do if you're going to be a therapist is learn how to take a history, do a proper assessment, come to a formulation, and determine what in your view might be the best way forward, which you then check with the client, patient, maybe the family, in order to develop a collaborative contextual framework as you set out on the journey of therapy. I think the biggest concern I have at the moment is that we're not educating our therapists, many of whom are not academic psychologists or psychiatrists. In fact, many of them seem to me to be young. The age seems to be dropping, they're working in difficult environments in a professional sense, and they must be taught and must be helped and supported to do proper assessments. I think that would be the first place, particularly if you accept Walpold's view that context really matters, context in terms of-- And I think that's a really-- sorry to interrupt you. I think that's such an important point because, for example, if you're training wellbeing practitioners who are low intensity, interventions, but actually coming across by virtue of the complexity of mental health referrals, coming across quite complex cases and helping those younger or early career practitioners to understand the contextual complexities that may be outside their prescribed protocol is actually one of the challenges. Because we don't want to stop families getting access to treatment. But nor do we want wellbeing practitioners or early career practitioners intervening where, in fact, the presentation is more complex. So I think that's a really pertinent point. I think it's a very good summary. I think that the specificity issues within therapists are not to be forgotten, but it might be something that Walpold's work didn't quite get at, is what were those techniques be that would distinguish people? I mean, if it's a conversational approach, there's one underlying philosophical dilemma that you didn't mention that I will suggest to you, is that unlike any kind of intervention in health care whatsoever, psychotherapists, whatever their school and their beliefs and their own values and so forth, are prescribing themselves. There is nothing else that does that. So when you put yourself on the line in a randomised controlled trial, it's not the same as prescribing a tablet, a pill, a catheter, or indeed a public health outcome, such as changing the environment in a school, which is going to be tried more and more over the next 30 years. So it's difficult. The defensiveness that one is so explicitly sees amongst psychotherapists of any kind is because they've spent years learning things and now somebody has come along in saying, well, they're quite important, but they're not the only thing. And anyway, you could have probably done it in 10 minutes instead of 27.5 hours a week and so forth. So it's not surprising that it's awkward to get implementation and delivery into the places where it really matters. And I agree with you, Jane. The young being practitioners are a vulnerable group, professionally speaking, because I'm worried about what it is that we're teaching them and how they're going to be supervised in a very difficult professional environment at this time. Thanks, Ian. So moving on from that, so context, relational, and where do we go from there, in the paper that we're going to talk about today, you're specifically looking at three different types of talking therapies. Do you just want to talk us through what those are and maybe what their origins are? Sure. So first of all, let me acknowledge all my good friends and colleagues from UCL, because this study was a combination of actually four university staff groups in the end. But it started off with a gentleman called Peter Fonagy, who Jane probably knows quite well, and you've probably all heard of, phoning me up and having a conversation about a randomised controlled trial. And Peter suggested that he and I applied, together with others, applied for this grant. So it was Peter that kicked this off. And of course, he wanted to study short-term psychoanalytic psychotherapy, which wasn't getting, in his view, I think it's OK to say this, enough of a decent press, given that there was more and more empirical evidence that it wasn't just CBT, which has enjoyed this being a front runner for 30 years now, and that other things could be brought in. So he wanted to do a study of comparing short-term psychoanalytic therapy with CBT in depressed individuals. He wanted to do this at the time that the NIH, National Institute of Health Research, had issued a very particular call to try and understand better the talking therapies that might have effectiveness in adolescents. And there hadn't been much in his country, and there had been a very large study in the United States which had involved a lot of antidepressants and had shown antidepressants were really valuable for some individuals. But who and when and how and why. And that one thing we weren't no longer needed to do was to have a control group who got nothing, because it was quite clear in that kind of standard trial of something against nothing, doing nothing, not a good idea, which didn't surprise any of us, but nevertheless had now been established. So that's how we got to short-term psychoanalytic psychotherapy and CBT. But I said, well, as you know, Peter, in our 2007 published randomised controlled trial, we were kind of surprised at the number of young people who did rather well, who were severely depressed, with an antidepressant and what you might call good psychosocial support. And when we gave them CBT 2, it didn't provide any added value. Now, this kind of fed into the idea of Walpold's contextual and common issues because here we're using two quite different methods. Psychosocial supports pretty straightforward problem-solving empathy, trying to keep the relationship but not adding cognition in the way that Jane described earlier for looking at cognitive rationality and the irrational thinking that can occur in depression. Not adding behavioural activation in a formal way that has become a big deal in young person's therapy in general, and certainly not adding things like insight and empathy in a strong sense that you would see in the psychoanalytic perspectives who are trying to make interpretations about the way people were appraising themselves in the world. So what was it that was useful if CBT made no added value? And we concluded there was something about the ingredients of everyday good clinical practise that was worthwhile in pulling together and dissecting out from the general corpus of I'm talking to my patient and looking after them. So we got to the third treatment, which was standard clinical care, but we turned it into something and we gave it a name and we called it brief psychosocial intervention. We weren't the first people to use the term BPI, or that was the first time the acronym was used, but we were the first people to use it in two distinct ways. One, brief psychosocial intervention came from the experience of people working in the clinic. So rather than theory-based, it's practitioner based. Doesn't mean there aren't theories, but it was practitioner based in how we formulated what to do. And two, it very much was out of the evidence, the continuity of evidence from our first trial which was called the ADAPT study. So then we had three treatments. We had CBT, we had psychoanalytic psychotherapy, and we had BPI. And Peter and I agreed, and we got others to agree, Shirley Reynolds came on board, Professor of Clinical Psychology at UEA and then Reading, that what we wanted to do was to test it in a way that people like you guys, academics and practitioners, would believe the results. And there we move into the more pragmatic world of saying, yeah, yeah, great trial, but look at all this stuff. I don't believe it. So we had to make sure that it was believable. And that's why we went for a big study, the biggest at the time, and we went to use only people who were really credited and able to practise those three things. So everyone who was a therapist in CBT met the approval of the specialist CBT committee that we set up to organise therapists for the trial. The same was true for psychoanalytic psychotherapy and we also set up a similar committee for BPI. Although BPI, being pragmatic and out of practise world, had less, as it were, regulatory history to it. So we had to be a bit on the hoof about that. But that's what we did. Thank you. And before we move on to why their work, which I think some of the mechanisms, which is what the focus of the paper is, what is the evidence that those three types of psychotherapy do work for adolescents with depression? OK, so now you get into the thin ice that is called the evidence base for the treatment of depressed young people. It's thin, but it's not negligible. So actually, you can trace about a 50-year history of the anecdotes of treatment. And if you take a qualitative view by accumulating all the qualitative, anecdotal data, you have to come to the proposition that it's reasonable to form the hypothesis that there are active elements in a professional conversation that do not exist in an over-the-garden-fence piece of advice or a chat with your mum and dad. So first of all, let's step back and not get too clever and say, look, the heuristic that is out there is that there's something about talking to people in a professional way that seems to be useful. I think the epiphenomenal error that's been made by all of us is to think that all the theory that all of those useful people thought they knew, and all the mechanisms that they thought that they were acting through their very special skills might not be true, at least might not be true in full. So I do think there is evidence. What about the empirical evidence? Well, the NICE guidelines had been about for a bit, but didn't say much about child and adolescent psychiatry. Since I was on the NICE committee at the time, I'm aware of that. The evidence was pretty reasonable that something about this concept called CBT and this construct called CBT therapy was doing something. The evidence was a problem. And this is nothing to do with CBT therapists. But the trouble is CBT was the only thing that seemed to have been studied scientifically in any particular quantity. But it seemed to be fairly clear that despite some of the grandiose claims of the CBT godfathers, there was no doubt that CBT was doing something that was pragmatically effective. And I read that, I think it was 2019, the NICE committee were moving somewhat, but very cautiously. There was a lot of couch language that psychodynamic psychotherapy might be appropriate in some cases of childhood depression. But it was very cautious recommendation, I would say. Yeah, I think that's a meeting of the great churches asking whether or not there isn't a common way of coming towards this concept of agency in God. They do it very cautiously, let's be honest, because they've got a congregation to worry about, not to mention their jobs. So I think it's the same with psychotherapists. That coming cautiously, it's still the case, and it's still the case in the UK that the dominant landscape for psychotherapy is CBT, in the very general sense. I think as David Clark, my good friend in Oxford, who's one of the grandfathers of CBT in this country, has said CBT is a family of conditions, but we're not dissecting them out properly. We don't really know which elements are good for what condition yet. So the problem was looking at other things caused a bit of a dilemma for a committee that, quite frankly, contained almost all CBT therapists. One of the interesting things in the 2005 and '15 reviews was that interpersonal psychotherapy, which is hardly practised in this country and very much, by the way, represents a kind of fusion that Wampold was interested in, developed at Yale, operated through adult depression only. Gerald Klerman, Myrna Weissman, and a man called Jean Paykel, who came over back to England and eventually to Cambridge, worked on this. And it was a brief intervention. Not as brief as we really mean. Could be 52 weeks worth. But it was very relational, it was very problem-solving, and it went down well in America. But it didn't travel, as Peter Fonagy said, didn't travel well over to Britain or many other places. So that got into the NICE guidelines, because there were six studies, and a psychologist called Laura Mufson was at the forefront with Myrna Weissman, showing that interpersonal psychotherapy is quite good in schools and quite good in community mental health clinics, not so hot in the child psychiatry clinics. So that got into the guidelines. And there was a very good review by a good clinical psychologist, well, I say she was good, she was here, so I better be careful, a Cambridge-based clinical psychologist who now is an Australian and very famous, who said that it was a bit tough because 90% of the studies that were being reviewed by its scientists were CBT versus nothing and 10% of the studies that were reviewed was any other psychotherapy versus nothing. So what happened if you took up the cudgel and said, well, how much better is a form of CBT than, say, psychoanalytic psychotherapy? And that has not been done. And that's why what we were suggesting caught a bit of attention. We didn't know. But there was a very small group of studies from America, again, and Europe, and that has suggested there was a signal in psychoanalytic psychotherapy for young people. And finally, the last thing was family therapy. Now, unfortunately, the studies on family therapy were quite good and they weren't getting much of a signal at all. So when the NICE guidelines came up for review in 2015, again in 2019, I don't think family therapy was going to do very well. And it kind of-- people wanted to acknowledge that it must be important, because how can you be in a subject involving child and adolescence and not be in a subject that involves family? And it seemed philosophically a problem to say, oh, well, involving the family can't be right. But direct family therapy so far has not done very well. So we ended up not choosing family therapy, we argued why. We ended up not choosing interpersonal therapy, which wasn't difficult because there weren't any in this country. And so we had our two main therapists, psychoanalytic CBT and our reference treatment, which was termed BPI, because we hypothesised that both the specialty therapists, unlike Wampold's assertion, would be better than the reference treatment. So we had a hypothesis that proposed there were, although we didn't explicitly state this, there were special mechanisms in psychoanalytic psychotherapy and CBT that were distinct from what was going on in BPI, and therefore they should get a better signal in outcome than BPI did. And then you've set that up really nicely. So what did you find when you compared the three? Well, so I'll tell you. But first of all, let's get the context, right, since we mentioned that. Our study isn't about people who get to child and adolescent mental health services. We've become very, very community-oriented in the last 10 years. This study was conceived in 2009, '8-'9, and started in about 2010, '9-'10. The paper was published in 2017. That took two years to get into the Lancet. So that's the kind of time scale, the context of time. And the context of the environment that we worked in was we stuck with child and adolescent mental health services, the tier three or tertiary stage patients that probably Jane sees at Great Ormond Street that will come to your specialist services in York that I'm aware of in the whole York area. And we're coming into Cambridge in East Anglia, and we chose three parts of the country. We had the whole of East Anglia, North London, and Manchester, Cheshire, and a bit of Liverpool. And we had 15 child and adolescent mental health services, and we spent an awful lot of time in getting those services to become research sensitive. And then the collection of some 470 patients, of which 465 got into the study, and the three arms had roughly equivalent numbers, I think, 155, something like that. And we got about 83% of each arm through to the final stage, where the intention to treat analysis was therefore going to be pretty valid. And-- I think that's so it's so impressive though, because having just been through some recruitment and so on through a variety of trusts. And each trust is a different country when we were trying to run a systematic programme throughout different clinics. That's incredibly impressive. But all I would say at that, I think the idea of having that degree of getting clients through, that percentage, is really impressive for anybody that hasn't done this sort of work. I'm impressed. I mean, as you can imagine, the number of people involved in making sure that those clinics felt happy and so on. We were all over the implementation and delivery. As you may know, there is this study called the STADIA study, which was published by Kapil Sayal and colleagues at Nottingham, which is extremely difficult to read because in their 1,240 adolescents that they followed through services, they showed literally no effects of service on mental state from the beginning to the end. There is no significant difference in-- and you should probably get Kapil to talk to you about this study. It's only just come out. So-- I actually, one of our podcasting colleagues was lucky enough to speak to professor Sayal about the STADIA study. So please do check that out. It's from just a few weeks ago. So back to the plot. We had good agreement. We had leaders. Every clinic had a research champion who was either a clinical psychologist or a mental health nurse or a psychiatrist. Every champion talked to each other. We had meetings all the time. You can imagine the amount of organisational work that-- It emphasised to me the importance of getting your clinic in good shape and I think it emphasised to me, we might talk about this, that what's trying to be going on in community at the moment, it has the same principles. If you're going to enter into an organisation like a school, the assumptions that you can make about all schools doing the same thing are virtually nil. So the idea of not having implementation and delivery for that group strikes me as just silly. So if you go back to the study, we got it set up right, and we had 465 individuals and we randomised them each to a relatively proportional groups. They got their treatments. We got to the end. We had a 12-month follow-up post treatment, which most studies, most studies don't do because they want to stop at the point of best effect, which is always short. So we followed up for 12 months and disproved that hypothesis too. And we were all delighted. I mean, I can't-- really, if you want to see the 2023 summary paper in the American Academy Journal, it's hard to believe that all those years later, we've had over three books, over 60 papers. People's careers have been made. I don't think anyone's career has been lost. I haven't asked that question. There's professors and consultants and consulting psychologists. All research Fellows have all come out of this extraordinary amount of work for which everyone deserves. You've given us lots of great context there. And I think one of the interesting parts of the study that we're going to talk about is you've taken this network approach. So my understanding of the network approach is that rather than thinking about symptoms of mental health difficulties resulting from a latent factor or an underlying psychopathology that we can't measure, so we measure the symptoms, but we assume that something latent there, the network approach suggests that these symptoms of mental health difficulties cause each other. Is that your understanding too? I think it's good enough. When you get into it, it's really a very complicated theoretical framework. And I think maybe the three of us would enjoy spending the rest of the afternoon talking about that. But to try and get across to people, a really different way of thinking about mental state dynamics is not the easiest. First of all, I think we should get, for context, we should say what impact showed was no significant difference between the three treatments. So I think Bruce would have-- I expected Bruce to send me a birthday card or something because-- I mean, it's clearly obvious that young people's RCTs were going to show the same thing as general adult RCTs. They're worse. They've got to be some common mechanisms across these treatments. We don't know what they are. We were proposing this so-called dodo effect, where everyone takes part and everyone has a prize. We had some really good critiques. We chose some people I knew would not bow down to this easily. And we've got some great critique writing from two colleagues from America and one from here. And they said, well, it is a fantastic start, but don't believe you can walk away from this and think you've got the answers. And one of the key issues in not getting the answers is that we were very fortunate, I think, to see that the difference between therapists in our study, there were about, oh, I can't remember. I think it's about 65 different therapists. And what's called the interclass coefficient, intraclass coefficient, sorry, between therapists was virtually nil. So that means we're taking out the individual differences that therapists might bring to the treatment. And we can truly examine the results as if there's something to do with therapy. What's very interesting about that is that also implies very strongly there are common effects. Because as Wampold points out in the studies, individual differences can drive huge differences in treatment outcome if you don't get that right. So here we've honed down and diluted any chance of interaction between therapist effects, giving us perhaps a better chance of arguing there's some common factors. I must say to you both and to anyone listening, of course, when we presented this to ourselves in the group, the two camps, CBT and psychoanalytic psychotherapy, were a little bit shaken, because I think everyone expected the specialist therapist to do a bit better. And there was quite a few weeks when people were trying to dissect out a bit of signal that suited CBT or suited psychotherapy, whereas of course I was completely happy because I thought there would be no particular difference. It was striking how similar the findings were. And they also showed this curve, which is why the network thing becomes important. I'm getting to it. People respond very quickly. There is a really rapid improvement. About 30% of the variance of improvement occurs within 12 weeks. We're following people up for 52 weeks. That's remarkable. And all of the therapists-- all of the therapy manuals predicted twice as much therapy would probably be required as was needed, including BPI. CBT had, I think, a 24-session or 20-session model. Surely believed the 12 sessions would be about right. STPP, short term psychodynamic therapy, they wanted a 30 session to 28 to 30 session model. They thought that was about right. Not now, Jane. And BPI, we saw a-- 12 was the maximum, probably 8. Well, we were all wrong. In BPI, the median was six. In CBT, the median was eight. 9, 10, depends where you drew the interquartile range cut off. And in STPP, it was half, half the therapy. So it was about 12 to 14. So we all got that wrong. And that really made me think too about what were we doing as therapists, as researchers bringing these notions to an empirical study. Where did we get the idea that you needed to do more? That's a question I don't have an answer to. I think it's worth, why do we do that? And that made me, as I've been doing for the past five years, when we're teaching BPI, which we do around the country a bit, I wanted to say to people, what do you think when you're learning whatever you're doing, how do you learn to know when you're going to stop? Tell me what are the psychological, the psychosocial, the contextual features they're going to tell me that you know this is the end of treatment. And you wouldn't be surprised. Virtually no young therapist has the idea of stopping worked out in their head. So we gave people a little task when we teach BPI, which is, you're going to do two sessions with your therapist. You're coming back. We were seeing them for supervision every month. And by the second session, I want you to tell me how many sessions you think you need and how that you're going to stop. And they say, well, I can't do that. I said, yes, you can. You're going to work out from your assessment and your formulations and all the information you've got, the predictors for stopping treatment and not for continuing the treatment. Interesting. Yeah. For efficiency, it would be really interesting to look at that. Especially if there's no difference in outcome. Why did we say that? Well, if you follow the curve down, then by 28 weeks, in terms of the outcome measure of symptom reduction, which was trans-symptomatic and diagnostic-- so it's depression, it's anxiety, it's obsessionality, it's well-being. it's antisocial behaviour. We measured all of those things multidimensionally. And when we looked at it, by 28 weeks, there were 90% of the people were below a quantitative threshold of 50% reduction in symptoms, there or thereabouts. But the thereabouts was what got the statisticians and me going. It wasn't really. We had a range in which there was still quite a lot of people with symptoms. So we followed them up for 12 months without any treatment. And when we did that, they continued to get better after treatment. So now we don't even need more treatment. What do we need? We need more time. The rehabilitation from treatment, the recovery coefficient or curve from treatment is still going on for 12 months after treatment. Only 16% of the subjects relapsed towards the levels of symptoms that they had at the time of therapy. That's a lower relapse rate by standards. So what happened with treatment was that everyone continued to get better by and large, except for the 16%. And by the time we get to 12 months, less than 5% of this whole cohort has used health care or social care services since end of treatment. Now, it's said that we're not very good at treatment. John Weisz has done fantastic work in showing that the real effect sizes, particularly in CBT, but not only in CBT, are much lower because we don't take into account the non-specific changes that occur in control groups with sufficient clarity to show that, yes, there is something about CBT, there is something about STPP, there is something about-- and everything else. But there's something about time and recovery and we need to know what that is. And I think that's begun to be signalled in work such as that done by Jessica Schneider, who used to work with John Weisz, now a big wheel on her own, and the Jessica Schneider equivalent in Britain, who is Maria Loades, who's at University of Bath. And they're trying to figure out how does one session of therapy go. And what Jessica Schneider shows is that psychoeducation or behavioural activation from the CBT programmes are as good as each other, and they're both better than doing nothing, in one session. They followed up their patients for nine months and they did not get much relapse rate. But they were mild. And an epidemiologist have pointed out an awful lot of people would have recovered anyway. But Jessica still got an effect size of about 1 and 1/2, which given that they had 2,000 people online in the study, I thought was pretty good. We had Professor Maria Loades on a few weeks ago. So check that out. We're really spoiling you with lots of relevant content. Our problem with a symptom is that a symptom isn't really a symptom. What it is an extension of a normative construct beyond the population range that we consider to be all right. Everyone gets sad. Everyone can feel worthless. Everyone has poor sleep. If anyone said to me by the time they got to about 16, I've looked at your symptom list and I don't have any of those things, and I would say, don't be ridiculous. Can you read? Let's just sit down for a minute. Nobody's like that. We have chosen to work in a field where that kind of moving slide rule, as it were, across normative to non-normative to atypical to abnormal to completely wild, unacceptable psychotic features is all on a continuum. Hence the latency concept, Umar. That's why epidemiologists in the 60s and 70s thought latent approaches would be reasonable, because these are quantitative trait changes. But we call them symptoms because we've got a descriptive categorical term with a lot of holes in it. Now, in the model of symptom counts, the assumption is that every symptom has an independent effect. So if you've got five symptoms and therefore get a diagnosis of depression, it's because it's assumed, not that even a lot of scientists will quite appreciate this, it's assumed that each of those symptoms not only has an independent effect, but they're equivalent. Because you don't have to count mood as being twice as important as poor sleep. They're just important as opposed to not. So you reach the end of the distribution threshold where it's no longer normal and you're counted as one unit of thing. Five of those, you're depressed. However, when we were working on this, it struck us all the time that none of us knew what the common elements in the symptoms were and what the specifics were. So that's when my group at, where I was, by then I was at the University of Toronto and as well as at Cambridge. And the University of Toronto group, very good at statistics, you might-- And the person who's the first author on the paper we're talking about, Madison, we talked together, and she's a clinical psychologist now, director of training at York University, Toronto, and she said, I'm quite interested in this. I said, great, I've got an idea, but I don't have the technical skills anymore because I'm getting old and can't remember the name and whether what's left or right anymore. So we got together and she did something called a by factor analysis. We wanted to see, if we took out the common element, did we have anything left? Because if Wampold was correct and everything's common and it's all latent and all these things are independent, but actually they're not because at the level of latency, they all count, then that's it. We don't have to do anything else. Well, that wasn't true, of course. So common by factor variance accounted for about 60%. But there we were, it was published in the JCCP, there were very, very clear signs that there might be some precision in treatments, but we don't know what they are if we look at the more specific factors that were left from the bifactor variable. The big argument you all might know, Jane, you may know that specific factors are just redundant noise. I don't believe that. And we don't believe that. So we published this paper and made some statistical criticism, as you can imagine, and we showed there must be some specifics. So now we have the theoretical dilemma, Ummar, because we've said there is a common feature, like Wampold says, but we've also got specifics. What are we going to do with those? So that was when I started thinking, there's this new guy-- this new paper came out in something called world psychiatry. And in that paper, a very, very thoughtful theoretical psychologist called Denny Borsboom at Amsterdam University-- anyway, he and together with his mathematical psychology friends started to look at something quite old. There's nothing new about network analysis. It's quite an ancient theory and quite an important use in big subjects like sociology and psychometric work in large number, even in large number things like biology. So I got excited, but worried because what Denny was suggesting is what you said, Umar, We think that psychopathology is about adding up atypical, abnormal, severe levels of symptoms that derive from normative behaviours, thoughts, and sensations. But what if it isn't? What if it's actually-- what if psychopathology doesn't exist except when things go wrong in the mind that involve those features of the mind, thoughts, feelings, and behaviours, influencing each other in ways that they shouldn't? Now, the idea that one thing leads to another is hardly new in behavioural science. But the idea that one thing has some latent mechanism, have to use the word, that is self-sufficient strength to cause another symptom or another item to turn rogue in the mind, that is both alarming and exciting. So, for example, we might now start to think that you can't get to five symptoms to make a diagnosis unless you absolutely have to have an interaction between two symptoms. Because there is a functional relationship between different compartments, and those compartments are doing things they should not be doing in the normal mind. And that's what really got me going about it, to the extent that I wrote a theoretical paper that was chucked out because it was thought disordered. I couldn't actually create the right theoretical space to say the things that I wanted to say. But I understand these are so disquieting to have the sense that this is-- it feels a bit liminal. It's because it's about to explode all over the diagnostic and therapeutic world. I think it has huge implications. Yes. And that's one of the reasons why this paper is so thought provoking. Well, it's very kind of you to say that because you won't be surprised to know we had an awful time trying to get it published. And the reason for that is precisely because it goes against the orthodoxy, in my view. This paper was reviewed seven times by statisticians, by different statisticians. And I'll tell you why. What's important about the paper is not doing the network analysis. Almost all network analytic papers are cross-sectional. And that's a problem, isn't it, if you're going to start talking about prediction and change and so on? There are two elements in the network analysis in network theory. This is terribly important for us or for me. One is, if it's true that symptom x is good enough to predict symptom y, then it can only do so over time. You cannot avoid a temporal feature. So you need longitudinal data. We had longitudinal data, because in the IMPACT trial, there were five time points. We're all going to have trouble over time because we didn't design it to have lots and lots of time points, which is what you really need to show trajectories over time. But we had five. And in theory, it's the minimum of three that you need to do anything useful. So we had a longitudinal network analysis design. And then it turns out there is no longitudinal network analysis statistic. So when we first got there and we talked to Denny Borsboom and others in the group at Amsterdam and at Leiden, nobody had yet worked out how were we going to actually analyse data over time. If you wanted to show that interaction between two items at time 0, we're going to predict a new phenomenon, not the same phenomenon, a new phenomenon at time y at a temporal distance between the two points. It didn't exist. So we got someone to work on it. We have a person in Cambridge who's also worked with us for a long time called Sharon Neufeld, who's a statistical psychologist, and she is now a Wellcome Trust senior fellow pursuing these things. And we had Madison, who was very, very competent statistically and mathematically, who had me chuntering away in the background, and we had a Dutch advisor, who was, in fact, a mathematical psychologist. And between us, having changed a few of the algorithms and knocked around a bit with the software, we came up with a longitudinal model that allowed variance and invariance estimates, and that meant you could follow things over time. And it wasn't just going to be an effective time. We hoped it was going to be something meaningful. So that's how we got to the longitudinal dynamic network model, which is now published. I don't mean our data. The model is now published. Anyone who wants to do longitudinal network analysis will now find stuff out there. And I'm sure it's going to get better. There is an issue with sample size because network analysis is intensive, and we just make it. And that's the other reason it took a while to publish. The JCPP were very interested in publishing it, and thanks to them, they stuck with us, because one or two other journals gave up, as it were. And we got through and it is what it is. And it is remarkable in its provocative findings. And thank you, Jane, for your comments. So the thing about it is that, which I really like, is the intuitions that we all wrote down on a piece of paper, put it in an envelope, and shoved them in a drawer. None of them were correct. The findings that we got were not what we expected. And it made us it-- and it came up against our own belief systems. It really did. It took us two years to get this paper written. It took us another year to get it published, I mean, get even close to being published. And it took us that time because it showed that the most important symptoms at the beginning of the study, these are the 465 depressed adolescents randomised into these three treatments, were now being treated as a cohort where treatment is now a cO-variable, co-factor. The most important features were fatigue and insomnia. None of us wrote that down. We all had psychological features as the prime driver, and it's not true. And that's about the most robust finding that I want to get across, is it's not true. In moderate to severe depression, the driver for the network relationships between items is fatigue and insomnia. Now, you say, OK, but not alone. Jane, you're so right. That's what you've got to keep reminding yourself. They're important because of their interrelationship at time 1 for predicting what happens at time 2. And that is a shift in the way we think about things. It should be a shift eventually in practise, because the way we assess people should be not just oh, yeah, tick fatigue, tick insomnia, tick worthlessness, tick mood disorder, tick self-harm. No. What are these things doing? Can we find new ways of doing clinical interviews to say, do you think in some way that not sleeping well or sleeping more, it could be hyper or hypo, and feeling tired are connected? Now, it's a really difficult thing to go to the subjectivity, isn't it? I mean, asking the patients that is in itself a real problem, but it's not been done. So I think we should do it. Although I recognise the difficulty of getting people to try, young people, to describe their mental state. And I was just going to say that, I mean, I am so struck by the idea, as you said, none of these great minds were predicting what may or may not be influential. And in fact, one of the key findings was fatigue. And I was mindful of the Shakespeare. I think it was Hamlet, wasn't it, who said that sleep is the balm of hurt minds? So perhaps he was way ahead of us all. But you're absolutely right. And that means that there's a modular thing that says what is fatigue. Because we don't know. I mean, the study of fatigue has never really taken off. And there are lots of reasons for that. I think the study of sleep has taken off, but I don't think we've translated it yet into our subject. What does it mean? How do we get-- and again, Umar, remember, I was quite preoccupied with HPA axis research for quite some time. And what I now realise is that we had a correlation between the loss of diurnal rhythm in the HPA axis control centre and the presence of insomnia. And I never did anything with it because I didn't make any sufficient theoretical connections. And can I have 30 years back please because I would certainly say, right, where are the young folks? Let's do that. To finish off, you've got to also understand that networks are dynamic and they change over time. So that means you have to understand what are the influential symptoms. And in the paper, we show how influential fatigue and insomnia are. But we also show how uninfluential symptoms we often think might be influential are not. Poor concentrations, very important, but does not produce new networks over time by itself. It does not. And social cognition, I don't see much network work in the early stages of depression, but my word, by the end, the biggest observable network is between the things that become important for CBT. The cognitive triad, worthlessness, the future, myself, these have emerged. So I'm going to make a prediction that might or might not be true. I'm not sure that the cognitive vulnerability hypothesis is a primary hypothesis. I think that adult depression emerges from individuals who've had episodes of mood disorder in younger years, and they have, as it were, developed their social cognitive triad, which I believe, certainly do believe, studied it enough myself, I believe, but I think it's not as primary as people thought. There's so much controversy and so much to do and so little time. What an extraordinarily thought-provoking conversation, Ian, is exactly what we hoped we would get from you from such an expert. It really is something that has got us thinking very differently, reviewing our academic and our clinical work in lots of different ways, or at least asking questions. And that's the best position to put us all in. Yeah, absolutely. Well, thank you very much. It took me back to, what, 2012 when we used to work together, where actually these are the conversations that were the most helpful. Like you think of post-doctoral training as the formal things that you do that are your job, which is data analysis or whatever it might be. But actually, these conversations are also part of that experience. And I hope that what we've had today, our listeners also benefiting from that because it's been absolutely fantastic. Thank you so much. It's been a real pleasure. And thank you ever so much for being interested in the paper and inviting me to do this. Thank you. Thank you. So that was a very, very fascinating conversation. I think what I really liked was, when you start a podcast or you start any conversation, you have a view of how it's going to go. And you start off with a structure. But actually, when you have someone like Ian, and I think that it really demonstrates when you have giants of the field, they just have so much knowledge that you're just like, OK, I just want to listen now. Because this is all very relevant. And what it made me think about was, when you're new to research, you do some research and it's not one level. And then as you progress through your career, there are layers underneath your thinking. The paper that Ian has just described, when I saw it, I was like, oh, why didn't I think of that? That's a really good idea. I have the skill set. I could have done it. But actually, he's done that paper. And underneath what's written in that paper, there's layers and layers and layers of theory, thinking, experience, knowledge. And it comes together in that paper. And if you just read the paper, you think, oh yeah, this is cool. This is really nice, this is challenging, this is thought-provoking, whatever. But actually having that conversation with him, you realise how much thought, experience, and knowledge and how many years of progress kids even has gone into that paper to get to that point, I think that's such an important point to make, because it's not just the paper that's written, and you can only write that paper if you've been through a variety of different experiences, ask questions, had a hypothesis that just didn't work out the way you thought it was going, and so on and so on. And that depth of thinking and that breadth of thinking was really, really interesting and exciting to hear to concertina out a thought process over a variety of different experiences and studies. And I hope that everyone else had the sense of that process of thinking that is great to be witness to. It's great to be alongside it. It's often those spaces in between that we learn so much from, as you say, just being alongside. It is great. What I thought was perhaps the most interesting was that theoretical paper that he can't quite-- talked about not being able to articulate it just yet because-- so it's turning the table over in so many ways in mental health services and mental health theoretical approaches. And so it almost feels too big, but it feels-- as I said, it's liminal. It feels like it's coming into consciousness and coming into articulation. And it will be very interesting period of time, as there's so many levels of need and service provision and so on. Does it need a radical rethink? Probably it does anyway. Could the theoretical position change the rethink in a different direction? Possibly. And that's very interesting, isn't it? And I can definitely relate to what he was saying about the theoretical bit because there's-- I mean, he mentioned the problem with diagnoses, and he was like, oh, let's not get into that, but actually, we're going to get into that in a few weeks. But I think that, for me, the fear of not being able to compose your theoretical position into a coherent argument is very, real for me. Because I've been thinking about diagnostic labels and all those things that I've not quite been able to articulate it well. And every time I try and write it, I'm like, oh, I don't know. And then about two days ago, I just woke up and it just occurred to me and I was like, you know what, I'm just going to write this down. And I wrote it down and I was like, yes, that articulates my position very well. And actually, maybe Ian will have that moment. Maybe it's just when you're consciously trying to do it, there's a block. And it will just come to him. But the power of sleep isn't interesting. You walk up and there it was, which is-- and I actually want to just very briefly, although-- I think the theoretical and academic repercussions are somewhat more powerful in some ways, but I also just want to just underline the importance of one of the findings about the potential power of fatigue and insomnia and how addressing that clinically may have a differential impact on lots of young people. So just to highlight from a clinical point of view that there are a variety of evidence-based sleep programmes and sleep usually using a cognitive behavioural model, somewhat emerging literature looking at a mindful cognitive behavioural module looking at increasing both the quality of sleep and the length of time asleep for young people and adolescents in particular. Because of course, we know adolescents are more likely to have depression and the proportionally more likely to be vulnerable to sleep problems, too. Some of those are physiological in terms of having high emotionality and a slight delay in sleep hormone, which means they're sleepier later, but they have to get up at the same time as the rest of the world. And some is environmental. Just having a phone in the room, even if it's switched off, has an impact on sleep. So those changes right there, very small changes. But again, there's an evidence base to think about sleep as a really important part of your treatment plan, potentially with a young person who's showing features of depression will be very important. So for lots of reasons. Theoretical and clinical repercussions are quite profound, I think, in that paper. Yeah, and I think that the finding is particularly interesting because in my department, there's a whole group of people who do research on sleep, and a lot of the stuff that they used to do was sleep and cognition. But I think that in recent years, they've shifted that focus to sleep and mental health. This paper demonstrates that, the conversation with Ian demonstrates quite the central role of sleep here in children and young people's mental health. So I'm hoping that through that body of work in my department, there will be some exciting findings around sleep, because they are excellent sleep researchers and now excellent mental health researchers and it just seems like the right recipe for some success in that area of some advancements. But it's interesting about fatigue. Our funders interested in this? Is there a motivation or a need, what's the word, motivation and to move that forward? Well, I mean, I'm not the right person to ask, but it's a question that should be asked because it's clearly-- it's not just about insomnia. It's also the idea of fatigue and what that implies and how that can be addressed. It's very interesting. So much to do, Umar, so little time. What's your academic takeaway, do you think? My academic takeaway. I think it is that what I said just after Ian left, which is you can do answer the same research questions. And on paper, it looks like a study that you could do and that somebody else could do. But what the thought process behind it and the theoretical underpinnings and the layers and layers and layers of knowledge that has gone into that is not always apparent. And the second takeaway, and actually I just thought of this, is Ian seems to be reasonably-- well, actually, he said network modelling was an ancient approach, but he seems to be the first time he's applied it in his research. I've applied it recently in my work too. It's interesting how two people who have come to this method recently have very varying levels of understandings of what it means and how they explain it differently. So the way I explain it to people is not how Ian explained it to us, but they're both correct and they are both making sense. But it's just indicative of where we've come from and what our levels of experience and knowledge are in the field. And I wonder if it's significant that both of you have come to network approaches at the same time. Is there something in the zeigeist? But again, a different question for another time. From our clinical takeaway point of view, I would say it's about the Zoom out, first of all, thinking about these wider psychotherapy debates. We might need to, and I suggest that we should revisit these age old questions, in the case of depression certainly, when we ask what works, for whom, and why. But in terms of zooming in from our clinical takeaway, consider sleep and fatigue as an important priority area of investigation with young people who have features of depression. [MUSIC PLAYING] Next week we'll be joined by Vivian Garner Rakesh about poverty and child and adolescent mental health. Lots of thought-provoking discussions about the times we live in. And it's a fantastic episode, so please tune in. Don't forget to subscribe, like, and share with your friends and colleagues.