Although cognitive therapy (CT)1 is the best-studied form of psychotherapy, its effectivenes compared with antidepressant medication remains controversial. Over the years, there has been some variability in the results of randomized controlled trials and other types of clinical trials, as well as meta-analyses. A review of the findings of the STAR*D trial may help shed more light on this important treatment issue.2Historical clinical studies
Efforts to compare the efficacy of CT with that of antidepressant medication date to a 1977 study by Rush and colleagues.3 In randomized controlled trials, CT was generally shown to be as effective as—if not superior to—antidepressant medications. This is also true in a number of studies using other designs and comparison groups (eg, wait lists, placebo, and other psychotherapies). Several meta-analytical reviews also have reported CT as superior to no treatment (ie, untreated control groups), wait list, pharmacotherapy with tricyclic antidepressants, and other therapies.4
However, the results of the influential NIMH Treatment of Depression Collaborative Research Program (TDCRP)5 initially appeared to be inconsistent with these findings. This early (1989) but large trial used a randomized, controlled study design at 3 sites. The researchers found that CT was as effective as the tricyclic antidepressant imipramine(Drug information on imipramine) in the full study sample. However, neither CT nor imipramine was significantly more effective than the control condition—supportive clinical management and pill placebo. Furthermore, in the more severely ill patients and in patients with greater functional impairment, CT appeared to be less effective than imipramine. As a result, the appropriateness of CT for severe depression was questioned. In addition, the adequacy of CT provided by the therapists in the TDCRP trial has been questioned,6 as has the role of the therapeutic alliance in the outcomes of this study.7
More recently, DeRubeis and colleagues8 conducted another NIMH-sponsored controlled 3-site clinical trial to attempt to resolve these issues. The investigators randomized 240 patients with moderate to severe depression to CT, paroxetine(Drug information on paroxetine), or pill placebo. Overall, CT was shown to be as effective as paroxetine in the treatment of moderate to severe depression when CT was provided by highly experienced cognitive therapists. As in the TDCRP study, there was a significant difference in results among the study sites.9CT and the STAR*D findings
The recent, multisite, NIMH-sponsored STAR*D study compared CT with second-step medications in patients who did not achieve sufficient benefit from, or could not tolerate, an adequate trial of the SSRI, citalopram(Drug information on citalopram).1 STAR*D aimed for symptomatic remission, not just response, since remission has been shown to result in better daily functioning and a better long-term prognosis. STAR*D used an equipoise-stratified randomization that enabled participants, in consultation with their physicians, to eliminate the possibility of being randomized to a condition they found unacceptable. This was done to best represent clinical practice. Unfortunately, as it turned out, few patients accepted all treatments, making patient acceptance an important limitation of this study. It is interesting to note that CT acceptance was primarily associated with sociodemographic characteristics (eg, higher level of education or family history of depression were indicative of increased likelihood for CT acceptance).10
One second-step strategy compared switching from citalopram to an alternative antidepressant or CT. There was no statistical difference in the intent- to-treat remission rate among those switched to the SSRI, sertraline(Drug information on sertraline) (18%); the dopamine(Drug information on dopamine)-norepinephrine reuptake inhibitor, bupropion (21%); or the serotonin-norepinephrine reuptake inhibitor, venlafaxine (25%). Outcomes in patients who were switched to CT (n = 36; remission rate, 25%) did not significantly differ from those whose medications were switched (n = 86; remission rate, 27.9%), except for lower reported adverse effects in the CT group.
The other second-step strategy involved augmenting citalopram with CT, bupropion, or the serotonin partial agonist, buspirone(Drug information on buspirone) (both antidepressant treatments resulted in intent-to-treat remission rates of 30%). Patients in whom citalopram was augmented with CT (n = 65; remission rate, 23.1%) did not significantly differ in outcome from those in whom citalopram was augmented with a second medication (n = 117; remission rate, 33.3%). The patients in whom citalopram was augmented with another medication generally had a more rapid response and remission than those in the CT augmentation group (40.1 ± 25.8 days vs 55.3 ± 31.2 days).2
Despite its size, a major limitation of the STAR*D study was the lack of statistical power to detect moderate between-group differences. There were also some limitations associated with the psychotherapy comparison, including:
- All patients agreed to a pharmacotherapy trial with multiple steps, which may have resulted in a population more accepting of medication therapy.
- A smaller number of people than expected accepted a CT condition. Because only 26% of patients were willing to accept CT assignment, it is difficult to make comparisons.
- Medications were dosed aggressively,11 and the fidelity of patients to CT was not independently validated. Thus, adherence and compliance may not have been optimal.
In the study, patients who were switched to CT, or for whom CT was an augmentation treatment, were not found to have statistically different response and remission rates from those who received medications. For those patients who continued on citalopram, medication augmentation produced a significantly more rapid remission than did augmentation with CT; and both treatments were equally well tolerated.
There was a trend and a numerical advantage favoring medication augmentation—which might have reached clinical significance if there had been a larger sample size. Patients who discontinued citalopram showed no significant differences in outcome. However, those who switched to a different antidepressant experienced significantly more adverse effects than those who switched to CT alone. It is important to note that the STAR*D study population included very ill patients, with 75% reporting recurrent depressive episodes and 27% reporting chronic depression. Furthermore, many patients had comorbid Axis I or medical disorders.
Thus, the STAR*D study demonstrated that CT was no less effective than antidepressant medications in patients who switched treatments and that there was improved tolerability with CT compared with the medications. Those who augmented citalopram treatment with antidepressant medications versus CT demonstrated a numerical advantage in remission rate that did not reach statistical significance but may have been significant with a larger sample size.12Clinical implications and lessons learned
What clinical recommendations can we draw from the STAR*D study and related literature on CT and medication treatment? The evidence for the equal efficacy of CT versus wait list control, medications, and other psychotherapies as a first-step treatment is well established4,9 and justifies the use of CT as a first-step treatment for many patients with mild to moderate depression. However, there appear to be significant site differences in most studies despite rigorous attempts to ensure adequate CT implementation.5,13 This may result from varying therapist skill levels. Successful CT outcomes require adequate adherence, just as medication efficacy requires adequate adherence. Therefore, successful CT implementation may require explicit levels of skill training and oversight, such as ongoing supervision, to resolve obstacles and prevent therapeutic drift. This may limit the efficacy of CT as a first-step treatment in many community settings if systematically trained therapists are not available or access to supervision (eg, from supervisors accredited by the Academy of Cognitive Therapy) is limited.
The STAR*D study employed close monitoring of care progression with the use of the measurement-based care model11 and monthly group supervision with the national CT director for the study.14 Clinical supervision focused on case conceptualization and the provision of explicit guidance for optimizing technical CT interventions. Thus, although there were varying degrees of CT experience among the therapists, CT may have enjoyed favorable outcomes as a result of the monitoring.
New technologies for CT dissemination and the provision of case supervision with supervisory feedback for CT therapists may offer a way to provide adequate treatment in a manner that is accessible and acceptable to many patients. Studies are under way that examine Web-based CT provision. One such preliminary study by Wright and colleagues15 demonstrated efficacy and acceptability of computer-based CT treatment.
The STAR*D data also show that CT is a reasonable second-step treatment for those who do not achieve adequate benefit from, or are intolerant to, an initial trial of antidepressant medication. Indeed, the data show that CT monotherapy is a good and reasonable option for patients who are willing to switch from an antidepressant to CT. For those patients who obtained some benefit but not symptom remission from an initial trial of an antidepressant, augmentation with medication produced a more rapid remission and a relative numerical advantage over CT. CT augmentation could then be considered as a next-step treatment after a partially successful pharmacotherapy augmentation treatment strategy has been implemented and further treatment is indicated.
Finally, the STAR*D study reported disappointing relapse rates for all treatments, highlighting the overall lack of durability of remission in patients with moderate to severe depression. This suggests that there is a need for innovative combined and/or sequential psychotherapy and medication treatment strategies to optimally treat the greatest number of patients with depression.13,16 In the meantime, we must disseminate the message that major depression is a serious illness requiring long-term aggressive treatment with the best evidence-based treatments we have to offer. Beyond acute-phase care, we must develop programs for the prevention of relapse and maintenance of remission to help our patients function well and sustain a high quality of life.