Part 3 in a four-part series on solving the toll of depression on populations. A talk given by Professor Pim Cuijpers, Professor of Clinical Psychology at the Vrije Universiteit in Amsterdam, and Director of the WHO Collaborating Centre for Research and Dissemination of Psychological Interventions.
In this talk, Pim Cuijpers focuses on the psychological treatment of depression and gives an overview of a meta-analytic research domain.
Useful links:
– Read Pim’s Mental Elf blogs on solving the toll of depression on populations https://elfi.sh/DepressionSolvingTheToll
– Visit Pim’s personal website https://www.pimcuijpers.com/
– Read an interview with Pim recorded shortly before he retired in September 2023 https://vu.nl/en/news/2023/interview-with-pim-cuijpers-professor-of-clinical-psychology
– Visit the Meta-Analytic Database of Psychotherapy Trials https://www.metapsy.org
So welcome at this third lecture on solving the toll of depression on populations this lecture will be on treatments of depression because our treatments for depression they work but not good enough and that’s uh one of the main subjects I’ve been working on over the past couple of decades and this is
The third lecture as part of a series of lectures on solving the toll of depression on populations um uh and part of the uh the this this this series of lectures I give those because I have to retire from my position as professor at this University so I will briefly again give
An introduction uh of the diseas bur and of deess I do that because not everybody wants to listen to All um uh lectures and this may be boring for the ones who already heard it in the previous lectures but I still think it’s important to give the
Context of what we’re talking about uh so please bear with me then I will talk about the idea of meta analytic research domain stepping up from normal meta analysis to much broader domains then I will talk about psychological treatments psychotherapies for depression what have we learned from these uh uh uh of the
Effects of these interventions and I will focus on the difference between the effects of psychotherapy pharmacotherapy and combined treatments because that’s the most used type of of treatments and then I will give finally some ideas how we can get to better treatments because they work but not good
Enough so briefly the introduction of the uh of depression as a public health uh problem major it’s a major it’s the most prevalent mental disorder we have 200 million 280 million people uh with depression if you look at the numbers that’s 63% of the total EU population 80% of
Those people live in low and middle income countries and the total cost of depression is$ trillion uh I also refer to the Lancet the the the the paper we wrote for the Lancet which was a collaboration between the the Lancet and the world Psychiatric association if you want to know more on
Depression you really should read this uh paper I was mostly involved in the writing of the treatments section but there’s anything on prevalence on incidents on risk groups on prevention uh so I can really recommend this one I showed this also in my previous lecture but I do think it’s very important
It’s shows that depression is a disorder of the working age meaning that that’s the reason why it’s so expensive 1 trillion Us doar lost in mostly because of production losses if we could use that money to to develop treatment services that would be a very good
Economic investment so to say in the UK they use used this reasoning to develop the IEPs program the increasing access to psychological treatments and they invest lots of money every year just to increase the number of people getting treatments as an economic investment uh in in the UK treatments are effective but the
Uptake is low um even in optimal conditions uh the the and so 100% uptake 100% of evidence-based treatments then we still only have 33% of this reduction of the disease burden so we need basic research because we don’t really know what depression is and where it comes
From if we have better knowledge then we may find better treatments and better preventive interventions but that’s a long shot that’s a long-term investment and we don’t you cannot expect that within the next 10 years or 20 years or 30 years that will really lead to new treatments or preventive interventions
So if we want to do something about the disease burden of depression uh we have to focus either on prevention that’s what my first lecture was about we should focus on dissemination simplifying treatments so that more people can benefit from it uh for example through digital interventions
That was the previous lecture and in this lecture I will focus on more effective treatments how can we improve the effects of treatments in order to do that we first have to know if they are effective do they work and uh what we have done uh in the past 15 years in my
Group U is that we have developed what we call a meta analytic research domain and that means that we do not do one meta analysis on one subject uh for example what’s the effect of psychotherapy for adults compared to control conditions or what is the effect of CBT compared to to other therapies
What we do in uh meta analytic research domains is that we collect all the research in a whole field uh every regularly so when this uh for depression we do that every four months now and we collect all randomized trials with psychological treatments whether they’re compared to other therapies to
Anti-depressants to combine Bine treatments in Youth and older adults and perinatal depression whatever as long as it’s a psychological intervention uh for depression we collect it and that has all kinds of advantages so we it’s a lot of work um a ton of work you can say but
It’s uh but what the the main advantage is that you really have a broad overview of of this whole field everything that can be known from randomized trials on psychotherapies for depression we know that because we have all the randomized trials and we integrate them all into specific questions and so anything you
Want to know about the effects of psychotherapy for depression we know that because we have all the trials we have the results of all these trials we because we have all these trials we can also look for example at secondary outcomes that’s not you that’s not possible when you do a normal Med
Analysis because then you will because many papers don’t report whether they have looked into the quality of life for example but we just go through all the papers and see if they’ve reported quality of life and um uh another Advantage is that we do that in a very
Consistent way so we have all these trials and we do M analysis but they are all all the data are extracted in the same way we do the analysis in the same way we use the same quality criteria so we have a very cohesive picture of what these therapies can say so um
Uh another Advantage is that we can see what the emerging topics are so we can see whether there are new developments because we we every four months we look at the new trials that are coming up and for example 10 years ago there was more and more research on depression or
Anxiety so trans diagnostic interventions which are not specifically at at depression but um and since 2005 for example the number of trials in low and middle income countries has risen explosively and we can see that because we and then we can also look we can always look at these new developments
And do a met analysis when there is enough uh research okay but there there of course there are Al always uh dangers and disadvantages the the main disadvantage disadvantage is that it’s a ton of work to keep it going uh it’s it’s very easy to drown in all the work that needs to
Be done to get this updated um it’s very difficult to get funding for it uh because many funders consider this as secondary data analysis which is not very uh uh interesting another risk is that because we have these very big data sets uh nobody else does that in the world
Because it’s a ton of work to develop it and they know that we have the data so they don’t do that now we have made the data open so to then it’s it’s easier for people to add things do things differently etc etc well it’s not just depression that
We do this for depression is the biggest one uh but we are also working on all kinds of other mental psychological treatments for all kinds of other mental health problem here’s an overview uh the green ones are ready uh the the and the the orange ones are people are working
On it uh so it’s it’s uh work in progress but as you can imagine this is a big project here we have made all the data open uh which are ready to be made open I won’t go into it I won’t show it to you but if you uh click on start
Analyzing you can you see the different data sets which are there and you can uh make selections out of the data and run met analysis online so if you want to know whether group therapy works in perinatal depression uh you just go there select the trials and run them Med
Analysis online if you want to know whether problem solving therapy works in older adults you go there select the studies and run it so it’s um it’s a pretty cool project so what what have we learned over these past 15 years on psychotherapies for depression um well
We now have more than 900 randomized trials and um they have compared psychotherapies with control groups they have compared one therapy with another therapy compared with anti depressants with combined treatments uh there are separate studies on inpatients there are all kinds of comparisons between treatment format uh digital versus face tof face
Individual versus group um Etc here you see the development over time and what the what’s the most important development is is that the number of Trials is in still increasing rapidly and um so um uh this whole field is booming uh here you see the subsets so we have separate subsets comparing Psychotherapy
With control uh psycho theapy versus Psychotherapy uh pharmacotherapy and a a rest group uh and the number of Trials comparing Psychotherapy with control groups that’s by far the biggest and is still uh increasing so this is one um where we did uh so we did some historical analysis not published
Yet uh to see where you can see how this development goes over time we did that with just with Pon uh regression analysis so no meta analysis and as you can see most studies until the mid 1990s were done in the US and since then more
And more research is done in Europe and now there’s more research in Europe than in the US but you also see that uh uh the number of trials in East Asia is coming up and the number of trials in other countries so outside Europe North America Australia and East Asia is
Increasing uh rapidly so uh before the 19 before 1995 75% of the trials were done in the US and now that has gone down to 20 5% so uh it’s really changing so uh this is one trial in which we I think this is an error
Um but this is one met analys one network met analysis in which we compared uh Psychotherapy pharmacotherapy and and combine treatments uh with each other and so so what we wait wait sorry about this so this is a met analysis we published uh this year on cognitive
Behavior therapy uh which is by far the best examined type of therapy um and well it includes more than 400 randomized trials with more than 50,000 patients and so it’s CBT is by far the best examin type of therapy it’s effective it works the effects are not very good but they’re good enough
It’s not more effective than other therapies that’s what some people think uh but that’s not true there’s no evidence for that it has been suggested that the effects are de declining over over time uh which is not true there there has been a study met analysis in 2014 that suggested that that the
Effects are declining but that was that met analysis was flawed and here we showed that it’s not uh declining the effects are smaller in children and Adolescence and when there is no guidance but that’s also not different for CBT uh than for for other therapies so then and then I come to
This other network met analysis in which we compared all these different types of therapy with each other and um this is the the the network so you CBT is the big one but you also see behavioral activation problem solving therapy third wave therapies uh psychodynamic therapy interpersonal Psychotherapy supportive couns in and the
Um yeah basically what we found is that all therapies have comparable effects except for non-directive counseling but that could be an AR effecta be because that’s used a lot as a control condition and you don’t know whether counseling is delivered as it should be in Trials where it’s
Explicitly uh used as a control condition um but usually these effects are reported in terms of effect sizes which indicates the difference between the treatment and the control group in terms of standard deviation so uh if you find an effect size of 0.5 the treatment and the control group differ 0.5 standard
Deviation after treatments from each other um you have never explained that to a patient right because that’s impossible to explain uh it’s even for many clinicians it’s very difficult to understand what that means what that what that what that uh uh what what is an effect size of 0 five a patient will
Ask yeah but what’s the chance that I will get better and so what we did is we looked at the ex exact response remission change and deterioration rates in one met analysis in of a couple of years ago we included 228 trials um uh we looked at response and
Remission and this is what we found for response um within the treatment group so we looked at how many people get better within the treatment group and within the control groups and we found that uh uh for most treatments on average the improve movement was 41 uh% so meaning that 41%
Of people get better when they get a treatment meaning that 60% do not respond and response means that the uh that the that the symptom level has decreased with 50% compared to Baseline so there are uh the half of the symptoms are gone um and if you look at the control
Groups that’s in carage usual control groups the response rate was 177% so the actual benefit of getting treatment is 41% minus 177% not so very good uh and it means that the majority of people don’t get better even when they got get an evidence-based treatment better than doing nothing better than not getting
Treatments uh because then the chance of getting better is really very small um and you can try another therapy when this one doesn’t work but overall the therapies are not that good and so 40 41% response reduce a reduction of 50% of symptoms but if you look at remission
That’s even worse remission means that people are well so they’re not depressed anymore they meet all the criteria that healthy people also meet healthy again and that’s only 26% of people getting treatments are remitted and that’s 12% in car as usual and 9% in weight list so
75% of people who get treatment are not well after the treatment um I will come back to that when I talk about recommendations well the the next question is how can we apply this uh how can we apply that we limited this to CBT and we looked at all the treatments uh
Format so guided self-help individual group I also talked about this in my previous lecture but briefly it doesn’t matter how you apply CBT that can be done individually in groups by telephone by digital interventions as long as there’s human support if there’s no human support then the effects are
Smaller then we got the question does this work in all age groups and so we did one big met analysis together with a group of John Weiss who has focused a lot on Research in Psychotherapy in children and Adolescence and we looked how do these therapies work in the
Different age groups what we found is that when people are adult the effects don’t differ so younger adults uh middle-aged adult older adults we now also have enough studies to look at older old people so 75 years and older um and it really doesn’t matter the effects are very comparable but in
Children and also in adolescence the effects are significant ly smaller we looked in one study which I don’t report here also at a response and remission rates in Children and adolescents and we found that the response rates in the treatment groups were the same in adolescence we didn’t have enough
Studies in children uh uh but the response rate in the control conditions in adolescence was much higher than we find for adults so which makes sense so be adolescence do respond but if they don’t get treatment the natural recovery so to say is higher than what you find in uh
Adults we have also looked at all kinds of other outcomes like quality of life social functioning anxiety self-esteem hopelessness dysfunctional thinking and what you find is that if a treatment for depression is successful that also improves all the other things social supports quality of life Etc although the effects are
Smaller um there are an there is a number of studies looking at suicidality so if people are suicidal and you treat them for depression and we find no significant uh result on suicidality but the number of Trials here uh is also very small uh but there
Is some somebody working on an update of this and they still find no significant outcome which really is an important finding there is also a small group of studies in mothers with small children and then you see that if you treat depression in mothers that also has a
Beneficial effect on their children on the mental health of that children of the interaction between mother and child and parental functioning in General so we have 900 randomized trials examining the effects of psychological treatments of depression but only a handful of co-orbit people with depression and coorbit other conditions
While we know that the majority of people with a depressive disorder also suffer from other uh coorbit uh conditions and so for anxiety and insomnia the the results were pretty good so if you have depression and anxiety and you get a treatment that works both on depression and anxiety
Same for insomnia so your insomnia gets better and your depression gets better for people with substance use uh problems the effects are pretty small much smaller than we find for other treatments and also uh s small small effect on substance use problems uh so uh substance use is really a
Complicated factor in uh uh treatment of depression but this also illustrates major problem of the whole field that we have 900 trials but a very clinically relevant aspect namely comorbidity we don’t have answers to that so uh people working in clinical practice know how important for example personality
Problems are if you treat people with depression and a comob with personality problem we don’t know how treatments uh do it for this population so there’s a there’s a lot to be done here so we have examined all kinds of other things uh the uh the no differences for in students that works
Also in students in when people have coobs General Medical disorders it’s work than older adults and perinatal depression maybe we have smaller effects in primary care I don’t have the time to go into that a chronic depression maybe the effects are uh smaller I already talked about com but substitute problems
Smaller effects and sub threshold depression which makes sense uh because there is not so much room for improvements uh but we do uh as I said in my first lecture if you do treat people with sub threshold depression you can prevent the onset of major depression later on we did not find an
Association between the number of sessions so um uh we’re repeating that now to see if that uh with all the new trials if that still uh is true but we did find an assotiation between the frequency of sessions and the outcome and that’s important I will come back to
That later too um so if you give two treatment sessions per week the effects are larger than when you give them once a week um so and then I come to the work led by my colleague irini kotaki I already talked about this in my previous slide
Uh this is an individual in my previous lecture uh and this is an individual ual patient Data Network met analysis of digital interventions for depression with guided interventions unguided uh and control groups and then you can compare all three of them and you can um
Uh when you go to this website you can fill in the characteristics of one individual and see what the treatment does for this individual whether this individual with this these characteristics needs guided interventions or an unguided intervention very important for the development of personalized treatments well this was the good news
About psychological treatments it seems as if they work and are very effective which is true uh but we have also a whole series of papers on all the problems of the field showing that uh the the if you just look at all the studies done in this Fields you heavily
Overestimate the true effects and we know for example that weight list control groups overestimate the true effects um uh risk of bias indicates the weak spots of randomized Trials so were the trials done methodologically in the right way uh we know the problem of publication
Bias uh the so when studies are not show no significant effects of an intervention they’re often not published we know that for sure so uh Alan d who worked in our group she looked at the NIH database uh to look for published uh for funded projects on Psychotherapy for
Depression and looked which were published and which were not published and so aot a quarter was not published and when we went after the people uh doing these trials and they gave the data so we could also see what happens when you adjust for the uh publication
Bias um and then yeah the effects drop with about 25% uh if you adjust for that and that’s um uh also what we find when we when we adjust for publication bias in a statistical way because you can do that in statistical ways and selective outcome reporting uh where people change
Outcome measures do other analysis afterwards because these are give better outcomes than uh when you look at the original outcome measure so if you adjust for all these problems then the effect sizes go down with about 50% so how do Psychotherapy and pharmacotherapy compare with each other
We did the network met analysis in which we compared Psychotherapy pharmacotherapy and our combination in depression uh and yeah 100 trials 12,000 people response was the main outcome and what we basically found is that at the short term Psychotherapy and pharmacotherapy have comparable effects combined treatment is better and at the
Longer term combined treatment is still better um uh and um uh Psychotherapy is probably a little better than pharmacotherapy uh this is for chronic this is also true for chronic and and treatment to resistant depression for severe depression uh so combined treatment is definitely better this one
Is a very this is one was L by Toshi Furukawa it’s not only based on our database but it’s it includes also uh several other uh randomized trials so here we looked at the acute phase treatments and that could be Psychotherapy pharmacotherapy combined treatments followed by maintenance treatments which could be also people
Could switch from Psychotherapy to pharmacotherapy from combined to psychotherapy from combined to pharmacotherapy so all the combinations were possible and so um uh 81 trials 30,000 patients um main outome was response at 12 months of the Baseline this is the the the the the network plot
And this is the outcome and to summarize this it shows that at a long term at one year um with u with sequential treatments combined treatment definitely itely is the best uh treatment compared to pharmacotherapy Al alone and Psychotherapy alone is also better than pharmacotherapy Alone um I think this
Has major consequences for uh treatment guidelines and this is one important paper that was also included in uh uh thinking about updates of the wh guidelines which are going on now and again I want to show you this idea of more personalized treatments this is an individual patient Data
Network met analysis of cbasp cbasp uh a cognitive behavioral analysis system of psychotherapy it’s the only therapy specifically developed for chronic depression and so we collected all the data we had we had only three trials uh but the trials are very big and so we had more than 1,000
Participants and so you had you you could have seasp anti-depressants or the combination and again uh if you go to this website uh here you can fill in the characteristics of the patients and you can see what what is the best treatment for this patients uh in this case you
See that uh it doesn’t really matter very much whether you give anti-depressants alone or fical therapy alone or the combination of the two and then you can talk with a patient which is the best of the three but if you for example uh increase the level of anxiety
It’s very clear uh uh that the combination treatment is better than either seasp or anti-depressants alone so again this is a important tool towards personalized treatments so how do we improve treatments uh because we saw that they are they are effective psychological treatments probably better than pharmacotherapy especially at the longer
Term uh but they’re not that good right 60% of people who get treatments psychological treatments are don’t respond so what do you do with this um so I I always say to people working uh as clinicians that they what happens is that when you work as a clinician you
Think that you’re treating somebody with an evidence-based treatment and then you see that that person doesn’t get better so then you think oh maybe didn’t I do the therapy right or maybe the diagnosis was not correct should I do the diag should I go back to the diagnos
Diagnosis phase and see what what’s really going on because this is an evidence-based treatment and it should work but that’s not the reality you can do with a depressed person you can do a uh treatment but it doesn’t work that’s for most people that’s what what you can
Expect so how how can you approve that I prevention dep prevention of depression is of course an option I talked about that uh scaling up and simplifying treatments is of course also important uh I already talked about that in my second lecture um I think we should develop no new
Treatments uh formats or Target groups there’s no reason to assume that digital interventions won’t work for older adults with diabetes while we see in our database all these new trials coming up saying this is the first trial in all adults with diabetes yeah but why wouldn’t it would it not work
Um um I will come back to that and we need research on sequential treatments because what do you do once when with these 60% of people who do not respond uh and there’s hardly any research on it where do you start with what do you give second and what do you
Give third we don’t know um you we need we need to improve treatments for specific Target groups we need to personalize treatments stratification is important and we should focus on the process of treatments so the frequency question I will I will work out a few of these
Things so no new treatments formats and Target groups I think the Innovation cycle we use in psychotherapies is wrong so what we what happens now is that people working in clinical practice see all these people not getting better and they want to learn all they want to help
People they learn all these skills that they and then all still all those people don’t get better or it takes ages before people get better so they want to learn new things and are very eager to learn new and Innovative things so they and then you have people saying okay I’ve
Got this new treatment which is much better uh than all the therapies you had uh because based on my clinical experience problems are caused by this in this way and are related to that and if you focus on this issue much better and we can see that in our database we
Come all these treatments coming up and uh uh going um uh being implemented being researched but they’re never better than the ones we had so then you get this new treatment it’s implemented trials are done showing that it works but then after a while you can do large
Trials or you can see in meta analysis yeah it’s new and it’s different but it doesn’t work better so in the end you have this whole Innovation cycle uh with a lot of energy people are trained people you know you get a Professional Organization but no patient gets gets
Better because of it and so um uh I I think it’s a waste of time the Innovation cycle is wrong and what we what we what we really need is for example focus on sequential treatments what do we do when do we start with CBT or do we start with behavioral
Activation or do we start with a third wave therapies and how do you decide that we don’t know and there is no way of deciding we don’t know who benefits from which therapy and we so we we don’t know uh uh why we should start with CBT and if
People don’t respond the 60% who don’t respond who should we give behavioral activation so there’s clearly more research needed for that and then I think we have all the problems clinical problems that have not been solved like for example comorbidity uh most people with a depressive disorder also have a social
Anxiety disorder or a substance use disorder or a personality problem or whatever we don’t know what that’s we hardly have any knowledge on what that what that what the implications of that are and should we adapt the treatment or not um uh what do you do when those
People don’t respond to treatments do do you give the treatment for the personality problem although it’s a secondary problem we don’t know and while these are the uh the the basic clinical questions that need to be answered not another randomized trial on another we already have 900 of those uh
While the basic clinical questions have not been answered same is to for chronically depression so the one Big Driver of the disease burden of depression is that a lot of people don’t get better they just remain chronic depressed for years and we don’t know what to do with it there’s one therapy
Specifically developed for uh chronic depression seus as I showed you U another thing is relapse prevention we know that people get better but then lots of them uh uh relapse again um stratification is another thing and that’s I showed you the work of irini kotaki uh my colleague who has
Done a lot who is doing all these ipd met analysis but there are other ways to stratify um there are also several large randomized trials now going on with several thousands of participants because that’s the number you need to really look at moderators and um uh one
Big disadvantage of ipd Med analysis is that you have to work with the predictors and moderators which are which are examined in the trials but that’s pretty limited so we need these very large expensive randomized trials with thousands of people where we can really look at all these potential
Moderators of outcome with sufficient statistical power um and then I think yeah we we know all these therapies have comparable effects but if you deliver them twice a week they’re better why don’t we do that we also know for example we have not done research on that but there is
Research that if you give uh feedback structural feedback to patients about how things are going and you do that every couple of sessions and then the patient and the therapist talk about how things are going that increases the effect size it has nothing to do with a
Content but more with a process and so these kind of Innovations are uh that’s another way to improve uh outcomes and if you’re interested in these things I will work these out in the next lecture where I will uh uh focus on the future thank you for your [Applause] attention