Transdiagnostic psychiatry goes above and beyond classification

For the last decade or so I have been involved in developing the science and practice of psychological interventions that apply across psychiatric disorders1, 2. These developments, known collectively as the transdiagnostic approach, have recently been challenged in this journal within a systematic review3. The review extracted research studies that used the term “transdiagnostic” in their title to include a heterogeneous mix of methodologies and samples. The authors report that few studies met the “Mansell criteria”4 for transdiagnostic research in psychiatry. In particular, the studies were critiqued for their limited use of standardized diagnostic interviews, and the lack of any alternative classification system. Treatment studies in the review generally found that the outcomes of transdiagnostic and disorder‐specific interventions were equivalent.

Each of the above points were presented as shortcomings of the transdiagnostic approach. I will explain here the conceptual foundations of the transdiagnostic approach in more depth to challenge that conclusion.

The “Mansell criteria” were initially developed by A. Harvey and colleagues1 to organize the existing research literature on cognitive and behavioural processes across psychiatric disorders. At the time, that review provided evidence that twelve different processes were shared across multiple (at least four) disorders. In other words, the transdiagnostic basis of psychological processes across psychopathology was already established.

The literature that is relevant to the transdiagnostic approach goes well beyond the articles that use the word “transdiagnostic” . For example, there is a large, replicated literature on “p” ,   the general psychopathology factor, which rarely uses the term “transdiagnostic”5. These studies show that a single factor underlying the diverse symptoms of psychiatric disorders can be identified and predicts a range of medical, health and socioeconomic outcomes. In addition, one could mention the human connectome research: large‐scale studies of brain networks have identified the same disrupted neural pathways across different psychiatric disorders. Most recently, a study of 402 patients with a range of affective and psychotic disorders, matched with 608 healthy controls, identified a single network (across the frontoparietal regions) that was shared across disorders, and its level of disruption scaled with severity6.

Earlier critiques of current classification systems have typically attempted to replace them with a new classification system, such as a dimensional system. Yet, the aim of the transdiagnostic approach is different. It is to identify, utilize and test a general theory of psychopathology4. This involves trying to understand the shared, overarching processes that cut across the classification system. This scientific approach is analogous to understanding evolution by natural selection as the mechanism of change that accounts for variation in all the living organisms that are classified7. Transdiagnostic interventions then aim to harness a general, neurally mediated, change process, regardless of psychiatric diagnosis. Furthermore, most transdiagnostic approaches posit a mechanism that is on a continuum with the general population, so the strict delineation between a clinical diagnosis and a sub‐clinical issue is less critical to this field of research1.

The most commonly assessed impact of transdiagnostic interventions is still symptom reduction. Yet, symptom relief is only one possible variable to compare and evaluate treatments. Other valuable variables include efficiency, cost‐effectiveness, accessibility, and reduction in patient‐reported distress. Patients, public, clinicians, service providers and policy makers need to be consulted to determine what is valued. One consequence of this broader perspective is that showing equivalent symptom reduction to a disorder‐specific intervention is a particularly positive outcome for transdiagnostic treatments, because by definition they have a reduced need for diagnostic assessment and no requirement for training in multiple diagnostic treatment models4. Furthermore, emerging evidence indicates that some transdiagnostic treatments are more efficient, since they may achieve the same reduction in distress through fewer numbers of sessions8.

It is commonly held that randomized controlled trials are the gold standard of treatment evaluation. However, on their own, they do not provide evidence that a psychological therapy works through the mechanisms that it claims. The effect could result from the expectation of the therapy working (placebo effect), or through simply talking to a professional. Again, if we follow the successful examples of other sciences, such as chemistry, physics and engineering, the most robust test of a theory is to build and assess a working model of a process9. This tradition started with Galileo, continued with prototyping in machine design, and today is typically carried out within computer simulations. If the model behaves the same way as the real system under natural conditions, then the theory informing the model must be correct. There is no a priori reason why this should not apply as well to human behaviour as it does to the theory of aerodynamics informing airplane design, for example. Our clinical research team uses Method of Levels (MOL) as a transdiagnostic intervention which we disseminate widely2, 8. This therapy is based on perceptual control theory, a general theory of behaviour drawn from control engineering. Its key principles of control, conflict and reorganization have been assessed through testing computational models against behavioural data9.

In sum, transdiagnostic psychiatry is well established, but to understand its transformative potential requires adopting the appropriate scientific approach. Future reviews need to evaluate a broad literature including general psychopathology and shared neuropsychological pathways, and to separate the evaluation of treatment and process studies. Treatment research needs to consider the multiple perspectives of different stakeholders when determining how to index evidence for the potential benefits of a transdiagnostic approach. Process research, on the other hand, needs to be theory driven, hypothesis‐led, and ideally emulate the model‐testing paradigms of other sciences. A transdiagnostic approach of this kind has the potential to generate a genuine, interdisciplinary, paradigm shift in psychiatry and mental health.