Skip to main content

Insights: Difficult-To-Treat Depression – Dr Stephen Barton

Rachel Allman
30 January 2024

Welcome to Psychology Tools Insights

Insights explores thought-provoking new research, translating findings into accessible, clinical takeaways for mental health professionals. Hear directly from the authors about their latest ideas, and how to effectively incorporate findings into your work, keeping you up-to-date and evolving your practice. 

Many of us have seen clients struggle with difficult-to-treat depression. Whilst CBT is an effective treatment for many, there are some who don’t respond to standard forms of CBT, as well as having chronic difficulties and/or multiple recurrent episodes.

Stephen Barton is a clinical psychologist, trainer, researcher and author based at the Regional Affective Disorders Service in Newcastle. He has led the development of a self-regulation model of depression aiming to help clients with difficult-to-treat depression. Stephen’s approach is outlined in his recent paper, ‘CBT for difficult-to-treat depression: self-regulation model’, published in Cognitive and Behavioral Psychotherapy (2023).

While it’s still early days testing the model, it’s an original and fascinating approach which we were keen to learn more about. We sat down with Stephen to discuss the strategy he used to develop the model, which elements he thinks will make a difference to this client-group, and why he has broken the conventions of current treatment parameters.

Developing the model

Can you explain a bit about the strategy you used to develop this model?

For a long time, the depression field was pretty much limited to Beck’s cognitive therapy – and that was a bit frustrating. From the 1990s onwards, we saw a lot of differentiation, such as behavioral activation (BA), mindfulness-based cognitive therapy (MBCT) and an increasing number of third wave approaches. These approaches (such as ACT and CFT) might no longer self-identify with CBT, but they grew out of the CBT school. So, we went from things being quite limited to a broader range of possibilities.

If we’re honest about the current evidence base, the newer treatments aren’t clearly more effective than Beck’s therapy. They’re just different, which is good and rich because we now have approaches that work for some patients, even if they don’t work for others.  But it does put the field in an odd position.  At a recent CBT conference, there was an informal question to the audience asking how many use Beck’s cognitive therapy for depression… and no one put their hand up. Most of the evidence base is grounded in a treatment that isn’t being used that much, at least not in its pure form. This seems like an unusual situation, but also an opportunity for integration.

In Newcastle, we set out to explore whether these different approaches could be synthesized into a coherent, integrated therapy that would benefit from the richness available across the broader field.  We hope it will result in a more effective and potent treatment, particularly for challenging cases, but it’s too early to tell. We’ve had encouraging results in single case studies, but the approach is still developing and needs more empirical tests.

Why the focus on difficult-to-treat depression?

In 2003, I moved to a specialist CBT service in Newcastle. At that time, CBT was doing a solid job in primary and secondary care services. Relatively straightforward cases were being served well, but the available therapies were missing the mark when it came to the most difficult-to-treat and complex cases. By that, I mean clients with multiple depressive episodes or chronic depression, a high probability of a trauma history, comorbidities, heightened risk and current social difficulties. The majority of these clients had a first onset in childhood or adolescence. These clients tend not to respond to CBT, or when they do they are prone to relapse. It’s a difficult task developing a therapy for these clients, but this is what we set out to do – we thought it would be a 5 year project!

The self-regulation model of difficult-to-treat depression

What are the key assumptions the model makes regarding depression?

There are two basic assumptions. Firstly, depressed mood is a normal human experience – it is not a disorder. However, depressed moods either self-correct or are perpetuated, and when they’re maintained there is a risk of a disorder developing. No-one is immune from depressed moods; we all experience them from time to time. This is analogous to anxiety and anxiety disorders. Many of us will say, “I’m phobic of this” or “I get anxious about that”, but most of us do not have an anxiety disorder. It’s a similar situation with depressed moods – we might feel uncomfortable about feeling down, but it’s normal to feel miserable from time to time. Secondly, a Major Depressive Episode (MDE) is a disorder – but first and foremost it’s a problem of self-identity. What’s matters is whether our self-identity stays intact during depressed moods.  Vulnerability to depression lies in the structure of self-identity – whether it can withstand life’s outrageous misfortunes or periods of disappointment and dejection.

Can you outline the model and explain about the link between self-regulation and self-identity?

Self-regulation is a normal set of processes needed to function as a person. Most of us have a sense of self-identity – of being ourselves. We need an identity to motivate ourselves and interact, to think and reflect, to make decisions, to process our history and learn from it, and to project ourselves into the future and set goals. This is self-regulation. Having a ‘self’ is part of what it means to be human. Other creatures have a less elaborate self-identity, whereas humans are sometimes preoccupied with who we are, what choices to make, what values to live by, the meaning of our experiences, and so on. At worst we can be self-obsessed – but we’re trapped in an intimate relationship with ourselves, whether we like it or not. We can’t escape it, so it’s desirable to have the best possible relationship with ourselves.

When the process of self-regulation is going well, people are not usually depressed. Positive self-representations generate motivation which keep us engaged and interested in the world. This keeps our mind in place, able to reflect and make decisions when we need to. This, in turn, helps us to link past, present and future, access relevant memories and move us towards our needs and desires. But no-one flows through life without challenges to self-regulation. There are inevitable times when we feel lost, don’t know who we are, when we can’t be bothered or aren’t motivated, or when we’re thinking ruminatively rather than clearly and decisively. Becoming dysregulated is not that unusual, but it’s a fork in the road.  How to respond in those moments? Do we hold onto a positive sense of self and reflect, re-motivate and re-engage? Or is there a downward spiral into dejection, self-dissatisfaction and, much worse, self-loathing? This is when people can get stuck in dysregulated states and that’s what we believe perpetuates major depression.

How does the model explain people getting stuck in self-dysregulation?

When people get stuck in depression, there’s an interlock between different dysregulated processes, and it’s very hard for people to find a path out of this without help. It’s as if the depressed self is in a state of capture; it is very entrapping. The model proposes there are six key processes:

  1. Self-identity disruption
  2. Impaired motivation
  3. Behavioral disengagement
  4. Cognitive rumination
  5. Intrusive memories
  6. Passive goals

In the therapy, there are six treatment components targeting these processes:

  1. Self-organization
  2. Approach motivation
  3. Active engagement
  4. Mental freedom
  5. Memory integration
  6. Goal orientation

The four other components share a lot with standard CBT: Alliance Building, Treatment Rationale, Risk Reduction (not to be confused with risk assessment), and Staying Well (relapse prevention).

One of the distinctive features is there is no protocol that specifies how to sequence the components. In fact, there’s not even a requirement that all the components are delivered. This is because depression is so heterogeneous; clients are dysregulated in different ways. For example, some of our clients have prolonged trauma histories, but they manage to stay engaged in daily life and function remarkably well. Their depression is maintained by intrusive memories, identity disruption and cognitive rumination. Other clients, particularly those struggling with adolescent developmental tasks, have an unformed self-identity and find it much harder to motivate themselves and engage behaviorally. It depends on the client’s formulation which components are emphasized, so the treatment is highly individualized.

The approach in context

Can you outline where this approach differs from standard CBT treatments?

The main difference is that there’s no prescription in terms of the stages of treatment. It’s an individualized therapy (idiographic) based on ten treatment components (nomothetic). Only those components can be provided, but they’re allowed in any amount, combination and sequence, as long as that is guided by the case formulation. In this sense, it aims to strike a balance between the needs of researchers and therapists. Researchers are – quite rightly – concerned about treatment adherence and fidelity. How else can we know what treatment is being delivered? But they’re open to the criticism that protocols are insensitive to differences between clients, especially with heterogeneous problems like depression. In contrast, therapists are concerned about tailoring therapies to the needs of the individual client. But they’re open to the criticism of drifting from the evidence base and integrating in ways that are difficult to describe and replicate. We’re trying to strike a balance between these two positions, both of which are valid.

The second difference is that it is a very high-intensity treatment. Standard CBT usually provides 16-18 sessions over 4-6 months, whereas our clients typically receive at least 30 sessions over 12 months. We’re committed to being as time-limited and efficient as possible, but we’ve broken with conventional treatment parameters. The reason is, when we experimented with longer-term treatment we had more success. What worked best was taking time and care with the foundations of treatment and setting conditions for change: building the working alliance, encouraging client engagement, developing an accurate formulation and rationale for change. In standard CBT, this is often achieved in 2-3 sessions, whereas some of our clients need 8-10 sessions for this. Skimping on the foundations just didn’t work, even though it helped therapists feel they were getting on with things. If you’ve been depressed for 30 years, are you really going to start feeling better in six weeks? It’s unlikely. What worked best was suggesting to clients that they have a long-term, life-limiting condition that is unlikely to change quickly or easily. Instead, there’s a commitment to work with them for at least a year to try to find a path out of depression. Several clients have said how empathic this is: it’s not a comfortable truth, but it’s realistic. It also takes the pressure off having to get quick wins.

Is there more danger of drift, given the ‘loose’ element of the treatment approach?

Protocols run the risk of being too tight; clinical practice runs the risk of being too loose. We’re aiming to get the balance of the two, though we need to conduct more studies to find out if that’s what actually happens. We want therapists and clients to be creative within the parameters allowed by treatment components. What’s different here is the treatment is specified at the component level, not the protocol level. Apart from that, the requirement for good fidelity is the same as any other treatment. Time will tell whether this helps to create a more effective therapy.

If a therapist started seeing a new client who they class as ‘difficult-to-treat’, would you advise following a standard CBT treatment first or starting with the self-regulation model straight away?

It depends on what service they’re in and the client’s treatment history. There is far more evidence for standard treatment than there is for what I’m describing. It’s not fully evidence-based yet – it’s still being tested. I would always suggest clients are given an evidence-based sequence of treatments. They should be offered Beck’s cognitive therapy (CT) or behavioral activation (BA) first, which are probably equally effective, on average. Clients should also be offered evidence-based medication: combined treatment is the top recommendation in NICE guidance. I would also encourage social prescribing and looking at the social context and support available. Even when CBT is effective, we shouldn’t overlook the other elements in the care plan.

The question is what to offer when CT or BA hasn’t worked, or hasn’t had a lasting effect? Is it best to repeat them, or maybe switch to interpersonal (IPT) or cognitive analytic therapy (CAT)? In this situation, we’re trying to offer an alternative CBT approach, and we’re explicit with our clients that it’s still in development. To be honest, most of them want to try something different rather than repeat. This is also part of the health-economic argument for a very high-intensity therapy. If it turns out to be clinically effective, it might be more cost effective than repeating brief therapies across the lifespan.

Are these treatment parameters realistic in real-world services?

At the moment, treatment development is strapped by conventions and economics. In CBT, there’s a convention of 16-18 sessions over 4-6 months – this is what worked in Philadelphia in the 1970s and it’s become the norm in treating depression worldwide, even when the context and clients are significantly different. In RCTs, it generates a drop-out rate of 20%, a response rate of 50-60%, and a sustained recovery rate of around 30% 2 years post-treatment. This is what effective is, as we’ve collectively defined it. When that evidence base is translated into resource-limited services, the treatment dose tends to be eroded, especially when demand for therapy exceeds the supply of therapists. So the current fashion is for briefer, cheaper interventions that can be delivered at scale and are affordable.

I understand the logic, and it’s probably okay for clients with straightforward problems, but it’s adding to the difficulties of clients with complex needs who are receiving briefer treatments along with everyone else. How much harm is caused by failed treatments? We don’t really know, but clearly some clients learn that CBT doesn’t work and give up on it. When thinking about affordability, we need to factor in the long-term costs incurred by less expensive, less effective treatments. More expensive, more effective treatments might be worth it in the long run, especially with life-long conditions. There’s also an ethical issue here: it’s one thing providing a treatment that doesn’t work, it’s quite another offering a treatment that can’t work. I honestly believe we’ve reached that point with some clients in some services.

My argument is this: in relation to clients with complex needs, firstly find out what it takes to make the treatment more effective Once we have that, then try to make it briefer, more efficient, more scalable, etc.  But don’t miss out the first step. This is why treatment developers should break conventions and experiment with different parameters. How else will change come about? I honestly don’t think the current paradigm is working for clients with complex needs; if it was going to work it would have worked by now. 

Psychology Tools Membership Sign Up

Practical takeaways for professionals

What are the key elements that you believe will make this model effective for this subset?  What elements do you want clinicians to know about and keep in mind when using this model?

There isn’t space here to unpack the hypotheses about change mechanisms, but mostly it’s about setting up tensions between dysregulated and regulated states and tipping the system towards the latter. I’d rather emphasize the setting conditions for change, because I think these are challenges for any CBT treatment with this client group:

  1. Create a tenacious, hopeful service. If you’re working with more difficult-to-treat cases, you need a tenacious service. You need to be in a service setting with colleagues, a supervisor, and a clinical lead who are fundamentally hopeful, tenacious, and won’t give up easily. This is important because clients can sense it. Are they coming into a hopeful space with a genuine commitment to work together?
  1. Address possible complications explicitly and proactively. We bring a lot of attention to addressing complications that might get in the way of the treatment working. It might be comorbidities, a trauma history, a current level of alcohol use, or a family member who’s trying to dissuade them from attending. It could be a previous unhelpful experience in the healthcare system or perverse incentives from the benefit system. There’s a whole range of different complications that create barriers to change and we try to address all of them very explicitly.
  1. Take the time needed to build a strong alliance. If there’s any barrier to the alliance, that’s what you should work on first. You could ask, “If you feel you can’t trust me with certain issues, do you know why not? or “Is there something you’re holding back and not telling me? If you have doubts about me or the treatment, I’d really like you to tell me about them”. We assume the alliance isn’t strong enough until we’re convinced it is. We don’t assume it’s okay because the client isn’t complaining.
  1. Get a clear formulation and rationale for change. The formulation needs to be individualized and explanatory, of course, but mostly it needs to create a path out of depression. This was an important piece of learning a few years back – the formulation was generating hope and a change rationale for therapists, but not for clients. So now we ask clients much more explicitly: “does this give you clues how you can get out of this?”. On its own, the formulation can be demoralizing for clients when it spells out what they’re trapped in. The most important formulation is the path out of depression: formulating problems only works if it helps to formulate solutions.

Gathering the evidence

What is happening regarding building the evidence base and understanding the effectiveness of the model?

We’re currently running a multi-site case series. We have 10 therapists across Newcastle, Cambridge and Glasgow sites, and we’re treating 20 clients over a three-year period. The study is due to finish in 2025, and then we’ll find out how replicable the findings are. We want it to work well, of course, but we believe in doing the science. If it doesn’t work as well as hoped, that’s still an opportunity to learn something – we’ll try and figure out what the barriers were and iterate the model. Next steps depend on the results; we’ll let the science decide.

Further learning and reading

How can clinicians learn more about this approach?

Any CBT therapist can learn the model. There are no special techniques: it relies on good CBT skills guided by the treatment components. We’ve written a book on the approach and also some research papers (listed below). We have a small network of former clients and therapists who’re developing the approach, mostly in Newcastle, Cambridge and Glasgow. The training is five days, spaced out over a year, with a supervised training case. People are welcome to get in touch if they’d like more information: [email protected]

Barton, S.B., Armstrong, P., Freeston, M.H. & Twaddle, V. (2008).  Early Intervention for Adults at High Risk of Recurrent/Chronic depression:  Cognitive Model and Clinical Case Series. Behavioural and Cognitive Psychotherapy, 36, 263-282

Barton, S. & Armstrong, P. (2019) CBT for Depression: An Integrated Approach. Sage

Barton, S.B., Armstrong, P.V., Holland, S. & Tyson-Adams, H. (2022).  CBT for Difficult-to-Treat Depression:  Single Complex Case. The Cognitive Behaviour Therapist, Vol. 15, doi:10.1017/S1754470X22000319

Barton, S.B., Armstrong, P.V., Meares, K., Bromley, E.H.C. & Whitton, D. (submitted) Incorporating a Trauma Focus into CBT for Recurrent Depression: A Single Case. The Cognitive Behaviour Therapist


Barton, S.B., Armstrong, P.V., Robinson, L. & Bromley, E.C.H. (2023). CBT for Difficult-to-Treat Depression: Self-Regulation Model. Behavioural and Cognitive Psychotherapy. doi:10.1017/S1352465822000273

Read the paper here ᐅ