Psychotherapy Effective Where Medication Fails, Study Finds

Researchers find psychotherapy is an effective treatment for patients with anxiety and depression who do not benefit from psychiatric drugs.


A recently published meta-analysis in the Clinical Psychology Review evaluates the effect of using psychotherapy for treatment non-response, defined as the failure of primary treatment to achieve its goal (e.g., symptom reduction). Researchers at the University of Basel in Switzerland, led by Andrew Gloster, reviewed published randomized controlled trials (RCTs) of psychotherapy for non-responsive patients with mood and anxiety disorders. Results indicate psychotherapy to be an effective treatment in reducing symptoms and improving well-being for these patients.

The researchers report that “the proportion of non-responders ranges between 30% and 40%. . .  In addition to continued suffering associated with the disorder itself, non-responding patients have decreased quality of life and increased mortality and suicide rates. For example, treatment non-responding anxiety patients experience a disproportionate burden of illness and have the highest rate of suicidal attempts than any other disorders.”

When people endorse more depressive and anxiety symptoms or greater severity of symptoms, clinical practice guidelines often encourage psychiatric drugs over psychotherapy. This is despite evidence that medications do not lead to better outcomes with these groups than psychotherapy.

In fact, many patients often prefer psychotherapy over biomedical treatments. In many cases, patients are not even given the option of psychotherapy and are prescribed medications, even when they have a higher risk.

Adding to the systemic devaluation of psychological treatments is that psychotherapy is often reimbursed at lower rates than medications. When looking into the funding of research on psychiatric treatments, industry bias significantly favors pharmacotherapy over psychotherapy when compared to non-industry funded trials.

In a health care system that primarily uses pharmacotherapy as a first-line treatment for many mental health issues, this study demonstrates the usefulness of psychological interventions to help patients see improvement.

Treatment non-response is an important issue in mental health care that often gets overlooked. However, challenges to studying treatment non-response include the lack of consensus around what ‘response’ or ‘non-response’ actually means. Additionally, studies tend to primarily focus on symptom reduction as an outcome and rarely include well-being and functioning measures, which are somewhat distinct from symptomology.

Guidelines available on secondary treatment options tend to be pharmacological – usually directing clinicians to rule out misdiagnosis, increase the dosage of current medication, or switch to a different medication class. Although psychotherapy may be suggested as an additive treatment, the authors claim a more empirical basis is needed for using psychotherapy as a response option for treatment non-response.

The authors of this study set out “to determine whether psychotherapy is a viable option to treatment non-response, identify any promising signals, while simultaneously identifying further research needs.”

This meta-analysis was the first to explore the efficacy of psychotherapy for improving outcomes for treatment non-responders. For the main research question, eighteen RCTs were included with patients with primary mood and/or anxiety disorders assigned to psychotherapy after initial treatment non-response (either by itself or combined with medication).

Research team members extracted primary (symptomology) and secondary (well-being/functioning) outcome measures from the individual studies while assessing their overall quality and risk of bias. Statistical analyses compared treatment groups with controls and tested the influence of moderators (e.g., types of therapy) and other predictors (e.g., length of treatment) on study outcomes.

Out of the 401 published studies examining outcomes of further treatment on non-responders (either psychotherapy or pharmacology), researchers found that only 20% examined psychotherapy as an option. In contrast, 80% used exclusively pharmacological/medical interventions. For those that included psychotherapy, these studies included a range of therapy types (e.g., cognitive-behavioral, psychodynamic and interpersonal, care management) and some that also used medications in conjunction with psychotherapy.

The researchers found an overall moderate to large impact of psychotherapy on treatment non-response on both symptomology and the quality of life/functioning. However, the latter was measured in less than 50% of the trials included in this study. Analyses on the quality of studies showed a wide variation in the authors’ perspective and a significant publication bias favoring studies with positive results on symptom outcomes.

No significant differences were found in outcomes between anxiety or mood disorders, psychotherapy types, or the type of previous treatment patients came into the trial (medication only, psychotherapy, combined). The researchers did not find any significant predictors (e.g., demographics, length of study treatment) on overall trial outcomes for psychotherapy.

However, lower quality publications tended to report larger effect sizes of treatment groups. Also, the longer a patient was in their original treatment, the larger effect was of the new treatment (i.e., psychotherapy).

Psychotherapy positively affected patients’ quality of life/functioning, with no publication bias found for these outcomes. However, treatment of patients with primary mood disorders had smaller effects on these well-being outcomes than those with anxiety disorders. Also, the length of study treatment did not seem to predict better outcomes.

Considering these findings, Gloster and colleagues conclude that psychotherapy is an effective intervention for cases of treatment non-response. However, they note the rarity of psychotherapy trials (compared to medication trials) and suggest more good quality research on psychotherapy for treatment non-responders.

They suggest the scarcity of such research could result from less funding for psychotherapy trials than pharmacological approaches. It also reflects typical clinical practice in how general practitioners address mental health issues with increasingly more psychotropic medication that doesn’t include psychotherapy.

Given the increased risk for functional disability, findings suggesting the positive effect of psychotherapy on functioning and quality of life are particularly important. The authors also bring up that many psychotherapies do not primarily prioritize symptom reduction but instead focus on how patients relate to their symptoms and live with more meaning. To this end, they recommend future research include measures that go beyond symptomology. To compare studies with each other in the future, the authors would like to see researchers report their definition of ‘non-response’ in a more homogeneous way.

They speculate that because there did not seem to be a particular quality about the treatments or the patients that stood out, just the act of switching treatment might have functioned to re-instill hope. Additionally, the previous treatment patients experience could help prime them for change in the follow-up psychotherapy. Thus, research into previous treatments would be important in understanding how different treatment types interact. Of course, other aspects of treatment influence patient outcomes not measured in these clinical trials that could be further explored, such as the superiority of sequencing different treatments (e.g., CBT to psychodynamic), common factors, therapist qualities, etc.

“Given that no therapy works for all patients, it is likely that treatment options will need to include options for individualization, similar to personalized medicine. This will require more information about what works, how, for whom following initial treatment non-response.”

While finding recommendations for personalizing treatments is complicated, we might consider what we think of as best practice more critically, actively taking patient preferences into account and integrating such research findings with clinical practice.



Gloster, A. T., Rinner, M. T. B., Ioannou, M., Villanueva, J., Block, V. J., Ferrari, G., Benoy, C., Bader, K., & Karekla, M. (2020). Treating treatment non-responders: A meta-analysis of randomized controlled psychotherapy trials. Clinical Psychology Review, 75, 101810. (Link)

Previous articleCharity: Hundreds of Thousands With Mental Health Conditions Being Chained
Next articlePodcast of the Ottawa Institute of CBT Interviews Robert Whitaker
Javier Rizo
Javier Rizo is a graduate student-trainee in the Clinical Psychology PhD program at UMass Boston. His current area of research is qualitative psychotherapy research, with a primary interest in promoting human rights-based framework in psychiatry through the education and training of mental health clinicians and researchers. Javier is committed to building a social justice psychiatry, working to incorporate humanistic, interdisciplinary and critical perspectives on mental health, with particular interest in the role of healers and common factors models of psychotherapy.


  1. I would urge people to read the critics of the psychotherapy research literature – William Epstein’s two books in particular – Psychotherapy and the social clinic in the united stated soothing fictions and his older book the illusion of psychotherapy, anything by David Smail and maybe the Therapy Industry by Paul Maloney.

    Report comment

  2. ” For example, treatment non-responding anxiety patients experience a disproportionate burden of illness and have the highest rate of suicidal attempts than any other disorders.”

    And so this “phenomena” is discovered “post treatment”? The highest rates of suicide are after “treatment”, not after “non response” and is the big “secret” that psychiatry keeps misrepresenting by calling the “treatment” damaged as “non responders”.
    You bet people “respond” to “treatment” and many do so by death. So in fact it is a lie that there are non responsive “patients”. And it would be an impossibility scientifically speaking, since EVERY SINGLE drug will have a “response”.
    The industry however would rather not admit the responses. And continue to call damage and death as “No response”

    Report comment

  3. I wonder if there are any studies or data on the topics therapy focuses on. How much of therapy consists of, “you need to be addicted to these drugs because you are biologically defective.” How much of therapy consist of trying to cope with drug effects such as amotivation, and obesity? How much is about coping with stigma and discrimination which occurs largely because psychiatry lied about how the people they label have broken brains and are so dangerous they need less rights than a criminal?

    Maybe that is why studies find therapy has little benefit while studies like this find some benefit when the focus isn’t on addicting people to drugs and addressing harms causes by psychiatry.

    Report comment

    • “How much of therapy consists of, ‘you need to be addicted to these drugs because you are biologically defective.'” 100%, based upon my experience, but that was all BS.

      “How much of therapy consist of trying to cope with drug effects such as amotivation, and obesity?” 0%, based upon my experience. But this is likely true for many, since none of the psychologists or therapists are medically trained, thus they know less than zero about the common adverse and withdrawal effects of the psych drugs. And, they have all actually been misinformed about such, since their DSM “bible” does misinform people about the effects of the psych drugs.

      “How much is about coping with stigma and discrimination which occurs largely because psychiatry lied about how the people they label have broken brains and are so dangerous they need less rights than a criminal?” Again, 100%, based upon my experience. But at least we now know those are psychiatric and psychological lies, since the vast majority of those labeled with the DSM disorders are actually victims of crimes, NOT dangerous criminals themselves.

      Report comment

  4. I’d be pretty f-ing insulted if I were called a “treatment non-responding anxiety patient.” Isn’t it clear how dehumanizing this whole idea is?

    I also wonder that the model here is to only provide therapy to those whose drug “treatment” seems to have “failed.” Why not start by talking to them before you decide to mess with their brain matter?

    Report comment