Blog

Munder, T., Fluckiger, C., Leichsenring, F, Abbass, A.A., Hilsenroth, M.J., … Wampold, B.E. (2018). Is psychotherapy effective? A re-analysis of treatments for depression. Epidemiology and Psychiatric Sciences, 1-7.

Based on a deeply flawed review in 1952, Hans Eysenck declared that psychotherapy was no more effective than custodial care for treating mental disorders. Later, he qualified this by stating that behaviour therapy was effective and other forms of psychotherapy were not. These statements touched off decades of angst and debate in the psychotherapy community, and also resulted in a great deal of research about psychotherapy’s effectiveness. By the 1970s the new research technique of meta-analysis was developed and was applied to psychotherapy research. In their seminal meta analysis of controlled studies, Smith and Glass found that psychotherapy was useful and with large effects compared to no treatment. And yet the debate continues. In 2018, Cuijpers argued that waitlist control groups (i.e., a common control condition in psychotherapy studies in which patients receive no treatment) are an inappropriate comparison leading to exaggerated estimates of the effects of psychotherapy. Recently, Munder and colleagues argued that waitlist controls are a way of estimating the natural course of the disorder (what would happen with no treatment) plus the effect of expecting to receive treatment (client expectations of receiving treatment tend to have a positive impact on symptoms). In fact, research shows that pre- to post-study effect sizes for the waiting period is approximately g = .40, or a medium effect. In other words, waiting for therapy in a study results in a moderate proportion of individuals getting better on their own without treatment. Therefore, Munder and colleagues argued that comparing psychotherapy to a waitlist control is appropriate and may be a conservative estimate of psychotherapy’s effects (i.e., psychotherapy has to outperform the effects of clients expecting treatment to help them). In their meta analysis, Munder and colleagues re-analysed 71 studies of psychotherapy for depression compared to a waitlist control condition. They found that the effect size in favour of psychotherapy was g = 0.75 (SE = 0.09) indicating a moderate to large effect. Psychotherapy was also more effective than care as usual (i.e., compared to another intervention that was not psychotherapy), g = 0.31 (SE = 0.11). There were no differences between types of psychotherapy (CBT, IPT, PDT, etc.) for depression outcomes.

Practice Implications

Despite various attempts during the history of psychotherapy to downplay or disparage its efficacy, research continues to show that psychotherapy is in fact effective. The average effect size compared to the natural history of depression is moderate to large (and that is likely an under-estimate). Again, there is no evidence that one type of psychotherapy is superior to another for treating depression. It is time for the field to move beyond questions of efficacy of psychotherapy and of the relative efficacy of different treatments, and look to understanding therapist interpersonal stances, client characteristics, and relationship factors that may improve outcomes from psychotherapy.

Razza, L. B., Moffa, A. H., Moreno, M. L., Carvalho, A. F., Padberg, F., Fregni, F., & Brunoni, A. R. (2018). A systematic review and meta-analysis on placebo response to repetitive transcranial magnetic stimulation for depression trials. Progress in Neuro-Psychopharmacology & Biological Psychiatry, 81, 105-113.

Transcranial magnetic stimulation (TMS) is a new treatment for depression thought to modulate brain activity through electromagnetic pulses delivered by a coil placed over the patient’s scalp. A meta analysis shows that TMS may be effective in treating depressive disorders when compared to a placebo control, although only 18.6% of those receiving TMS were no longer depressed at the end of treatment. The placebo control condition usually involves a sham version of TMS in which the coil is placed over the scalp but no magnetic stimulation is applied. In antidepressant trials, the placebo response is quite high such that approximately 40% of patients respond to the placebo condition (in antidepressant trials, the placebo condition includes an identical pill that is inert). In this meta analysis, Razza and colleagues assess the placebo response in TMS. They included only double blind randomized controlled trials (i.e., trials in which both the patient and physician were not aware if the treatment was real or a sham). The authors estimated the placebo response based on pre- to post-sham TMS scores of common measures of depression. The meta analysis included 61 studies of over 1300 patients. The main result showed that sham response was large (g = 0.80; 95%CI = 0.65–0.95). Trials including patients with only one episode of depression or who were not treatment resistant (g =0.67, 95%CI = 0.06–1.28, p= 0.03) had higher placebo responses than those trials in which patients previously had two or more failed antidepressant treatments (g = 0.5, 95%CI = 0.03–0.99, p = 0.048).

Practice Implications

The results of this meta analysis demonstrates a high placebo response in trials testing TMS. This is similar to the high level of placebo response commonly seen in patients in antidepressant medication trials. It appears that psychological factors like attention, instillation of hope, patient expectations of receiving benefit, and perhaps working alliance may account for an important portion of why pharmacological and other medical interventions appear to work for those with depressive disorders. This is particularly true for patients who are receiving treatment for the first time or for whom previous medical treatment was successful.

Gueorguieva, R., Chekroud, A.M., & Krystal, J.H. (2017). Trajectories of relapse in randomised, placebo-controlled trials of treatment discontinuation in major depressive disorder: An individual patient-level data meta-analysis. Lancet Psychiatry.

Individuals with a history of depression who get better have a 30% to 50% chance of relapse in the first year. That is, major depression tends to take a recurrent course, so that about a third to half of patients who initially improve will then experience a re-emergence of symptoms. In this meta-analysis, Gueorguieva and colleagues looked at whether they could identify classes of patients who respond differently to antidepressant medications depending on whether they discontinued or continued with the medications after symptoms improved. The meta-analysis included over 1,400 patients from four studies of duloxetine or fluoxetine (i.e., Cymbalta or Prozac) who participated in a discontinuation trial. A discontinuation trial design involves randomly assigning patients who respond positively to the medication either (1) to stay on the effective medication or (2) to discontinue the treatment and receive a placebo. Such a design gives us an estimate of the advantage of maintenance versus discontinuation of medications to reduce relapse of depression in the longer term. Gueorguieva and colleagues found that 33% of those in the medication continuation condition relapsed (i.e., 33% those who responded well to the initial trial of medications and who then continued with medications had a recurrence of depressive symptoms). By contrast, 46% of those in the placebo/medication discontinuation condition relapsed (i.e., 46% of those who responded well to the initial trial of medications and who then received a placebo had a recurrence of depressive symptoms). In other words, continuation of antidepressant medications resulted in a small 13% reduction in relapse rates compared to continuation with a placebo.

Practice Implications

This meta analysis indicates that continuing with antidepressant medications after depressive symptoms remit provides only a modest level of protection against a relapse of depression. Thus continuation with antidepressants after symptoms improve may not be worth it for patients who struggle with medication side effects and complications, or who cannot afford continuation of the medications. There is growing evidence that psychotherapy is effective for preventing relapse, likely because psychotherapy teaches patients ways of coping and interacting with others that allows them to manage life stresses more effectively after the treatment is over.

Johnsen, T. J., & Friborg, O. (2015, May 11). The effects of cognitive behavioral therapy as an anti-depressive treatment is falling: A meta-analysis. Psychological Bulletin. Advance online publication. http://dx.doi.org/10.1037/bul0000015

Depression is a highly debilitating disorder and ranked third in terms of disease burden in the world. Cognitive behavioral therapy (CBT) is an effective treatment for depression that was introduced over 40 years ago. In part, CBT sees depression as caused by maladaptive thoughts that maintain emotional distress and dysfunctional behavior. Reducing depression is achieved by eliminating the impact of or chancing maladaptive thoughts. CBT is the most researched psychological treatment for depression, and the research goes back several decades. A number of technical variations and new additions have been made over the years to CBT to improve patient outcomes. The volume of research and its history provides a unique opportunity to assess time trends in the effects of CBT. In this meta analysis, Johnsen and Friborg asked: “have the effects of CBT changed over time”? They also looked at whether client factors (e.g., demographics, symptom severity), therapist factors (e.g., age, experience, training), common factors (e.g., therapeutic alliance, client expectancies), and technique factors (e.g., fidelity to a treatment manual) can explain these trends. Johnsen and Friborg reported on 70 studies of 2,426 patients conducted from 1977 to 2014. Males accounted for 30.9% of patients, 43% had comorbid psychiatric conditions, and the average patient was at least moderately depressed. The average effect of CBT in reducing depression was large (g = 1.46 after accounting for publication bias). Women had better outcomes, studies with poorer methodological quality showed larger effects, and patients of more experienced therapists had better outcomes. There were too few studies measuring therapeutic alliance to assess the effect of common factors on outcomes. Most interesting was a significant relationship between effect sizes and year of publication. That is, the effects of CBT declined significantly over the years, though the average effect remained large. Surprisingly, there was a steeper decline for studies that used a treatment manual compared to those that did not. No other variables were reliably associated with this decline.

Practice Implications

Women and patients of more experienced therapists appear to benefit most from CBT. Although the effects of CBT declined over time, the treatment remained highly effective. Johnsen and Friborg’s study could not easily explain this decline. The authors suggested that the placebo effect (expectation on the part of patients, researchers, and therapists) is typically stronger for new treatments. However, as time passes the strong initial expectations tend to wane thus reducing the overall effect of the intervention. They also suggested that CBT treatment outcomes may be improved not by technical variations and new additions, but by better ways of integrating common, therapist, and client factors.

Author email: tjj@psykologtromso.no

Rutherford, B.R., Pott, E., Tandler, J.M., Wall, M.M., Roose, S.P., & Lieberman, J.A. (2014). Placebo response in antipsychotic clinical trials: A meta-analysis. JAMA Psychiatry, doi:10.1001/jamapsychiatry.2014.1319.

The placebo response refers to improvements in symptoms among participants in medication trials that cannot be specifically attributed to the active ingredient of the intervention. For this reason, it is common to have a placebo control condition in trials of medications. In these trials, some participants are randomly assigned to the medication condition, and some are randomly assigned to a placebo control condition. Typically, the placebo is a pill that looks exactly like the medication but that has no active ingredient. Both patients and providers are blind or unaware of whether the patient is receiving the active medication or the placebo. The placebo response is usually attributed to a number of sources: (1) the patient’s expectation of receiving benefit, (2) the patient’s contact with a caring provider and the healing effect of factors like therapeutic alliance and provider empathy, (3) statistical and measurement error, and (4) random changes in patient symptoms that are unrelated to the medication or the placebo. The first two sources are psychological factors that are often specifically active and purposefully enhanced in psychotherapies. That is, some psychotherapists actively work to develop an alliance with the patient and to align therapeutic interventions with patient expectations and preferences. (For a broader discussion, see my review of Common Factors in this month’s PPRNet blog.) The placebo response can sometimes be quite powerful such that antidepressant medications, and antipsychotic medications for example, only tend to be modestly superior to placebo. People with schizophrenia have cognitive difficulties that may reduce their expectations of receiving benefits from treatment. These patients also have significant interpersonal difficulties so that their alliance with health care providers may be significantly hampered. For these reasons, it may be possible that the placebo response may play a smaller role in the medical treatment of patients with schizophrenia. Rutherford and colleagues conducted a meta analysis of 105 studies of over 24,000 participants from 1960 to the present. Their goal was to examine if the average drug-placebo difference decreased significantly over time (i.e. across years of publication). They found that the placebo response significantly increased from 1960 to the present. That is, the average placebo patient tended to get worse in the 1960s, but by the 2000s the average placebo participant tended to get better. The effect of this trend was large (r = .52). By contrast, treatment change associated with antipsychotic medications decreased over time, and the effect of this trend was moderate (r = -.26). The authors suggested possible explanations for this trend. The average participant in drug trials in the 1960s was more severely ill than the average patient enrolled in drug trials in the 2000s. It is possible that the placebo response is more powerful in less severely ill individuals. Also, the authors suggested that a number of study design factors (e.g., multi site vs single site trials, financial incentives to recruit more patients may result in less severely ill and younger samples) may also contribute to this trend.

Practice Implications

One of the practical implications of these findings is that drug companies may be less inclined to fund research and development of new medications for mental illnesses if the research is increasingly showing only modest benefits over control conditions. On the other hand, health care workers who provide: support and empathy, a positive therapeutic alliance, positive expectations about benefits of treatment, attention to patient preferences, and a coherent narrative to understand their patient’s illness may help to enhance the effects of interventions including antipsychotic medications. This may be especially true for younger and less severely ill individuals with schizophrenia.