What is it that changes in therapy?

W When someone sits in front of me with memories that intrude, a body that still reacts as though danger were close and a life narrowed by avoidance, the question is not only what diagnosis the person has. The crucial question is what keeps suffering alive in this particular person.

Is it the trauma memory itself and the avoidance of it? Is it the hours of worry, rumination and threat monitoring that follow when the memory appears? Or is it feelings such as grief, rage, guilt, shame and a longing for closeness that arouse so much anxiety they have to be kept away?

These are three different questions. They may be present together, but they do not necessarily lead to the same treatment. Yet psychotherapy research has often been better at comparing complete treatment packages than at understanding what actually changes during treatment.

We know more about the treatments than about the change

A new systematic scoping review by García-Mejía and colleagues examined how change during PTSD treatment was studied between 2000 and 2023. The researchers found 115 studies with measurements before, during and after treatment. The selection required at least one established PTSD intervention and excluded, among other things, pilot studies and case studies. This is a substantial body of material, but also a bounded part of the research.

Eighty-two per cent of the studies were conducted in the USA. Half the samples consisted of veterans or military personnel. Prolonged Exposure, PE, accounted for 50 per cent of the studies and Cognitive Processing Therapy, CPT, for 31 per cent. No study used ecological momentary assessment to follow what happened in people’s everyday lives between sessions.

A few treatments and patient groups therefore dominate this review. That limits the answers the material can provide, but also reflects the review’s selection. For me, it raises a larger question: which processes of change do we miss when measurements mainly follow treatment sessions and rarely what happens between them?

The authors therefore call for clearer theories of cause and effect. What changes first? What follows? Is it reduced fear, changed interpretations, less avoidance, less rumination or a greater ability to feel and regulate emotions that sets the rest in motion? Today, we know surprisingly little about that sequence.

PE and CPT help many people, but they are not enough for everyone

PE and CPT are not straw men to be knocked down. They are well studied and recommended alongside EMDR in several guidelines. Many people are clearly helped. But first-line treatment does not mean the same treatment suits every person, and most studied does not mean best for everyone.

The new review starts from the position that around 40 per cent do not respond sufficiently to established PTSD treatment. Dropout is also a real issue. The figures need to be read honestly. A meta-analysis of 116 randomised studies estimated dropout at 16 per cent for psychological PTSD treatment overall, 22 per cent for PE and 30 per cent for CPT. The researchers used absence at post-treatment assessment as the measure of dropout. That is not always the same as having discontinued treatment itself. Differences between studies were large, and the figures do not in themselves say why someone stopped or did not attend. Burden, practical obstacles, life circumstances and preferences are different possible explanations.

An even newer meta-analysis of 181 articles and just over 124,000 US service members and veterans makes the picture more concrete but also more complicated. Weighted dropout was 34.7 per cent for PE and 40.1 per cent for CPT when the treatments were delivered in weekly formats. When the same methods were part of intensive outpatient programmes, the estimates were 5.5 and 8.5 per cent respectively. This was not a direct randomised comparison, and differences between studies were very large. The pattern therefore does not show that format alone causes the difference, but reminds us that a treatment name does not determine dropout. Pace, format, patient group, practical conditions and other contextual factors may contribute. Whether individual matching reduces dropout is a separate hypothesis that needs to be tested. Even within an effective method, then, there is no single solution that suits everyone.

It is just as important to be careful with the terms negative effects and deterioration. Research does not support the idea that PE or other trauma-focused CBT generally makes patients worse. A meta-analysis of 56 randomised studies with 4,230 participants found that deterioration and adverse events were generally uncommon. Trauma-focused CBT had no higher risk of deterioration or adverse events than other psychotherapies and had a lower risk of serious adverse events.

That is an important safety finding. At the same time, the meta-analysis could include only studies that actually reported some form of possible harm, and other systematic reviews have found that such information is missing from many treatment studies. An absence of reported deterioration does not always mean that it has been investigated well.

MCT does not organise treatment around prolonged, repeated retelling of the trauma. That is an important difference in approach, and one reason for my interest in the method. But the difference does not show that MCT entails less burden, fewer dropouts or a lower risk of deterioration than PE. Such questions must be examined directly. Among established PTSD treatments too, there are alternatives that do not require a repeated trauma narrative.

The question of cost needs the same sobriety. A US study of 200 adults with PTSD compared PE with sertraline, and also compared choosing one’s treatment with being allocated to it. Over one year of follow-up, the analysis of costs and quality of life favoured PE over sertraline, and favoured being able to receive one’s preferred treatment. This is a concrete health-economic comparison, not an argument for the same treatment for everyone. For me, patient preferences, treatment outcomes and costs belong together. Healthcare needs to offer considered alternatives and follow what they actually lead to.

MCT: what if the memory does not need more processing?

Metacognitive therapy, MCT, asks a radically different question. Rather than starting from the assumption that the trauma memory needs more exposure or processing, MCT asks what the person does when the memory, thought or bodily signal appears.

Does she begin to ruminate over why it happened? Try to fill gaps in the memory? Monitor her surroundings, her body or her own thoughts for new signs of danger? Try to push the memory away, control it or prepare herself mentally for its return?

In Adrian Wells’s model, it is these prolonged strategies – worry, rumination, threat monitoring and thought control – that can block natural recovery after trauma. MCT therefore does not try to make the trauma memory harmless through prolonged repetition. Instead, the patient can discover that a memory can be there without being analysed, controlled or becoming the starting point for hours of mental processing. This is not avoidance. It is another way of taking control of cognitive strategies and giving the patient the power to direct attention.

The direct research evidence for PTSD is limited but difficult to dismiss. In a preliminary randomised study with 20 participants, eight sessions of MCT produced clear improvements compared with a waiting list, and the results were maintained at follow-up. In another randomised study with 32 participants, both MCT and PE were effective. MCT produced faster change and better self-reported PTSD outcomes immediately after treatment, but at follow-up the treatment groups no longer differed.

A newer study allocated 57 war veterans with PTSD to MCT, PE or control and followed quality of life using the SF-36. Both MCT and PE produced better results than the control condition immediately after treatment and at the three-month follow-up. No statistically significant difference between MCT and PE was found, however. The published result concerns quality of life and cannot establish which method reduced PTSD symptoms most or entailed the least burden.

A study published in 2024, using data from 2010–2013, compared 32 patients who received MCT with 28 who received treatment as usual: EMDR, PE or trauma-focused CBT. The groups were not randomised. The MCT group had better results on the PTSD and anxiety measures after treatment. According to the study’s criteria for recovery, the proportion was 53 per cent against 29 per cent on the IES-R and 59 per cent against 36 per cent on the PDS. However, the overall difference in the distribution between recovery, improvement, no change and deterioration was not statistically significant. The study had no follow-up and reports no systematic collection of adverse events. It therefore provides no firm answer about comparative risks.

What matters most may not be that MCT ‘won’ these small comparisons. It is that clear improvement seems possible without organising treatment around prolonged, repeated re-experiencing. This does not show that exposure is unnecessary for everyone, but it does show that it is not the only conceivable route to change.

This is promising, but not conclusive proof. The clinical comparison published in 2024 was not randomised, had no follow-up and the comparison group contained several different treatments. The earlier randomised studies were very small and conducted by researchers close to the method. MCT may be a better route for some patients with PTSD. We do not yet know how many, for whom or whether it is better in larger independent trials.

Dropout data do not show that MCT is easier to complete either. In the small randomised comparison of MCT and PE, one person discontinued each treatment. That provides no evidence of a difference between the methods.

A PTSD-specific meta-analysis from 2024 found a very large pre-to-post-treatment change in seven studies with a total of 99 participants. That sounds convincing, but the pooled effect was a within-group comparison, the studies were highly heterogeneous and the analysis was based on participants who completed treatment and provided post-treatment measurements. It also combined full MCT with a study of Attention Training Technique and a study of young people. The result confirms a strong signal, but does not show that MCT is superior to established treatments or produces fewer dropouts and negative effects.

A direct randomised comparison of MCT and EMDR has also been described in a published study protocol. A protocol shows how researchers intend to test a treatment, not what the treatment has achieved. It therefore cannot be used as outcome evidence in either direction. The question of how the methods compare needs to be answered with reported treatment outcomes.

The broader research picture nevertheless makes the method hard to relegate to the margins. An updated meta-analysis across psychiatric conditions found a large effect compared with a waiting list and a smaller but clear advantage over other cognitive behavioural therapies. At the same time, smaller studies produced larger effects, another reason to call for larger independent trials.

ISTDP: what happens to the feeling that was never allowed to exist?

ISTDP enters through a different door. Here, the focus is on feelings aroused in relation to what happened, and on the defences and anxiety activated as those feelings approach awareness.

Trauma does not leave only fear behind. There may be rage towards the person who caused harm, grief over what was lost, guilt at having survived, shame, love and longing for someone who was also dangerous. Such conflicting feelings can be difficult to bear. In ISTDP, the therapist follows moment by moment how a feeling arouses anxiety in the body and how the person protects herself by, for example, shutting down, overanalysing, becoming self-critical or distancing herself.

This too is a form of approaching, but not necessarily a planned retelling of the trauma. One approaches the emotional experience that comes alive in the room. The work is adapted to the person’s anxiety tolerance. If anxiety becomes too high or takes the form of cognitive disorganisation, somatisation or dissociation, the work needs to be graded. Intensity does not mean putting everyone under the same pressure.

The direct evidence for PTSD is even thinner than for MCT. In a naturalistic pilot study, 41 patients with PTSD received ISTDP, an average of 5.5 sessions. Physician visits and costs fell in the year after treatment, and the reduction was still apparent in the three-year data. General psychiatric symptoms decreased with a large within-group effect.

The study had no control group. Only 13 of 41 had sufficient repeated self-reports, and the researchers used no standardised outcome measure of PTSD symptoms. The study therefore cannot establish whether ISTDP reduced PTSD or was more cost-effective than PE or CPT. It shows an interesting signal: treatment was followed by fewer general psychiatric symptoms and lower healthcare use in patients with PTSD and extensive comorbidity. That signal deserves a randomised trial with PTSD-specific measures.

A health-economic scoping review from 2025 found recurring signals of reduced hospital, physician, medication and sickness-absence costs after ISTDP in public healthcare. It strengthens the case for further testing of the method, but is not PTSD-specific and does not turn the pilot study into a cost-effectiveness trial.

ISTDP must not be exempt from the same scrutiny of negative effects as other therapies either. In a qualitative study, eight experienced ISTDP therapists were interviewed about negative treatment effects. They described incorrect assessment and misapplied technique as central risks and highlighted the relationship between negative reactions, patient feedback and dropout. The study cannot say how common this is, but it underlines why the work must be adapted moment by moment to anxiety tolerance and defences, and why deterioration needs to be measured in future trials.

ISTDP also has positive results in, among other things, depression, anxiety, somatic difficulties and complex or treatment-resistant conditions. A meta-analysis found large changes within groups and an advantage over control conditions, but study quality varied considerably. Here too, then, the same gap in knowledge recurs as in the new PTSD review: we see that change can happen, but know less securely which process brings it about.

A person is not a treatment package

The point is not to replace one dogma with another. MCT must not become the new method for everyone, and ISTDP must not be used to press every patient towards stronger feelings. Traditional trauma-focused exposure has a place when it suits the person, the problem and the goal.

For me, the choice of treatment needs to bring research evidence together with an individual formulation and the patient’s own goals and preferences. I want to understand the role of avoidance, rumination, threat monitoring, anxiety and defences in that particular person. These are working hypotheses we need to examine together and reconsider during treatment, not a ready-made answer key for which method the person should receive.

Imagine two people with the same PTSD diagnosis and the same recurring trauma memory. This is an example of how I might formulate clinical questions, not a research-tested rule for choosing a method. For one, the memory is followed by hours of rumination: What did I miss? What does it mean? Could it happen again? The person scans the surroundings, tries to control their thoughts and attempts to fill every gap in the memory. Here I want to examine prolonged thinking and threat monitoring, which are central questions in MCT.

In the other, grief, rage or guilt are aroused while the person experiences strong bodily anxiety, becomes self-critical, shuts down or withdraws. Using the ISTDP model, I can explore how feeling, anxiety and defences relate in the moment. The work needs to be adapted to the person’s ability to stay present in the contact and in the feeling without being overwhelmed. Both people may also need to approach things they avoid. The difference lies in which processes we examine and where the work may begin.

I believe better matching can reduce unnecessary burden and perhaps dropout too. This is a clinical hypothesis, not a demonstrated advantage for MCT or ISTDP in PTSD. To test it, future studies need to compare clearly described ways of adapting treatment with well-conducted established treatment. Patient involvement and preferences need to be taken seriously in both groups.

This is not an argument for mixing techniques without a plan. It is an argument for process-based precision. We need to know what we are trying to change, follow whether that particular thing actually changes and be prepared to think again when the patient is not helped.

A diagnosis describes a pattern. A method offers a map. Neither is the person or the terrain.

Research needs to follow the person between sessions

For me, the review points towards research that needs to move closer to people’s everyday lives. I want studies that follow feelings, metacognitions, attention, avoidance, defences, functioning and quality of life between sessions. Deterioration and adverse events need to be reported as systematically as improvement. The treatments also need testing in ordinary healthcare settings, with room to examine patient preferences and the costs of different care pathways. That is my proposal for what the next step needs to include.

Above all, research needs to dare to ask the question that comes before the method’s name: what is it that needs to change here?

That is where psychotherapy begins. Not in the idea that one method suits everyone, but in a joint attempt to understand what holds this particular person in place and what might set her in motion.