Psychological therapies may be evaluated using research methods designed for drugs rather than talking treatments - potentially limiting patient choice and shaping mental health services in the wrong way - according to a new academic analysis from The University of Manchester.
Based on analysing existing research, the article argues that Randomised Controlled Trials (RCTs), widely seen as the gold standard of medical evidence, may be a poor fit for assessing therapy.
Rather than presenting new experimental findings, the editorial brings together and critiques the current evidence base, making the case that talking therapies are often personal, flexible, relationship-based and evolve over time - characteristics that are difficult to capture in standard trial designs.
The analysis suggests that relying too heavily on RCTs may favour short, standardised therapies such as Cognitive Behavioural Therapy (CBT), while overlooking other approaches that may benefit many patients.
Key arguments
Psychological therapy is often assessed using research models originally developed for medicines
Unlike drugs, therapy is typically personalised, open-ended and shaped by the therapist-client relationship
Heavy reliance on RCT evidence may narrow treatment options in public healthcare and insurance systems
Short-term symptom measures may miss broader outcomes such as improved relationships, stability and self-understanding
A broader evidence base is needed, including real-world outcomes and patient experience
Why this matters
RCTs are commonly used to decide which treatments receive funding, policy backing and public provision.
The editorial argues that this has had significant consequences in mental healthcare, where therapies with strong trial evidence - particularly CBT - have become dominant in many systems.
In England's NHS Talking Therapies programme, only a small minority of high-intensity practitioners offer non-CBT approaches, according to evidence discussed in the article.
Why therapy is different from drugs
The article highlights key differences between medicines and psychological therapies.
While drugs can typically be standardised and tested under controlled conditions, therapy is more fluid. Sessions evolve in response to the individual, progress can be non-linear, and meaningful change may include outcomes that are harder to quantify.
What current evidence may miss
The editorial argues that many therapy studies are relatively short, often capturing early symptom improvement but missing longer-term change, setbacks or deeper psychological development.
It also suggests that recovery is often more complex than a single score or endpoint, and may involve ongoing change over time.
To illustrate this, the author uses a fictional case study showing how different therapy models can lead to very different long-term outcomes, even where short-term symptom improvement appears similar.
Towards a broader understanding of evidence
Rather than rejecting trials altogether, the editorial calls for a more pluralistic approach to evidence, including:
Real-world service data from routine NHS and community settings
Research into how therapy works, including the role of relationships and context
Greater emphasis on patient perspectives and lived experience
What the researcher said
"We have become so used to treating Randomised Controlled Trials as the gold standard that we rarely stop to ask whether they are the right tool for every intervention," said Dr Sahanika Ratnayake, philosopher of psychiatry at The University of Manchester.
"This editorial is not based on a new experiment, but on analysing the strengths and limits of the existing evidence base. Therapy is not a pill - it is a human relationship shaped by trust, timing, context and individual need.
Why it matters now
Demand for mental health support is rising, while services face pressure over waiting times, staffing and budgets.
The article argues that policymakers need evidence systems that better reflect how therapy is actually practised, and what patients themselves value from care.
Publication details
The editorial was published in BJPsych.
DOI: https://DOI.org/10.1192/bjp.2026.10674
