|Home | About | Journals | Submit | Contact Us | Français|
The mood disorders are prevalent and problematic. We review randomized controlled psychotherapy trials to find those that are empirically supported with respect to acute symptom reduction and the prevention of subsequent relapse and recurrence.
We searched the PsycINFO and PubMed databases and the reference sections of chapters and journal articles to identify appropriate articles.
One hundred twenty-five studies were found evaluating treatment efficacy for the various mood disorders. With respect to the treatment of major depressive disorder (MDD), interpersonal psychotherapy (IPT), cognitive behavior therapy (CBT), and behavior therapy (BT) are efficacious and specific and brief dynamic therapy (BDT) and emotion-focused therapy (EFT) are possibly efficacious. CBT is efficacious and specific, mindfulness-based cognitive therapy (MBCT) efficacious, and BDT and EFT possibly efficacious in the prevention of relapse/recurrence following treatment termination and IPT and CBT are each possibly efficacious in the prevention of relapse/recurrence if continued or maintained. IPT is possibly efficacious in the treatment of dysthymic disorder. With respect to bipolar disorder, CBT and family-focused therapy (FFT) are efficacious and interpersonal social rhythm therapy (IPSRT) possibly efficacious as adjuncts to medication in the treatment of depression. Psycho-education (PE) is efficacious in the prevention of mania/hypomania (and possibly depression) and FFT is efficacious and IPSRT and CBT possibly efficacious in preventing bipolar episodes.
The newer psychological interventions are as efficacious as and more enduring than medications in the treatment of MDD and may enhance the efficacy of medications in the treatment of bipolar disorder.
The mood disorders are characterized by episodes of depression or mania and are among the most prevalent of the psychiatric disorders. Major depressive disorder (MDD) and the less severe but more chronic dysthymic disorder (DD) involve depression only, whereas bipolar disorder (BD) requires episodes of mania or hypomania. The mood disorders account for the vast majority of suicides and are a leading cause of disability.
Both psychotherapy and medications are widely used in the treatment of the mood disorders. Historically, the primary focus of treatment development was on symptom reduction, but there has been a growing recognition of the need to develop strategies that prevent subsequent relapse and recurrence. Medication treatment has long been considered the standard of treatment for more severe depression and bipolar disorder and has gained market share relative to psychotherapy in recent years with the advent of less problematic medications. Nonetheless, not everyone responds to medications and there is no evidence that drugs do anything to reduce risk for subsequent symptom return once their use is discontinued. Some patients respond to psychotherapy who are refractory to medications and it has long been claimed that psychotherapy has enduring effects that can reduce subsequent risk in ways that medications cannot.
In this article, we review studies of psychological therapies for the mood disorders in adults to determine which ones are empirically supported using the criteria defined by Chambless and Hollon. According to these criteria, a therapy is considered efficacious and specific if there is evidence from at least two settings that it is superior to a pill or psychological placebo or another bona fide treatment. If there is evidence from two or more settings that the therapy is superior to no treatment it is considered efficacious. If there is support from one or more studies from just a single setting, the therapy is considered possibly efficacious pending replication. We further differentiate between the effects of treatment on the reduction of acute symptoms versus the prevention of subsequent relapse or recurrence and pay particular attention to comparisons to medications. Earlier reviews have applied these criteria and we update those reviews.[8,9]
This approach is similar to the one taken by the FDA in determining when a medication can be marketed in the United States. It puts a premium on well-formed studies in fully clinical populations that speak to the efficacy and specificity of a given intervention and requires direct comparisons to draw inferences regarding the relative efficacy of different interventions. As a consequence, it sometimes leads to different conclusions than are drawn from meta-analyses that tend to include all studies in a literature (regardless of quality) and that estimate differential effect sizes in the absence of direct comparisons between conditions. Meta-analytic reviews often find comparable effect sizes between different psychotherapies even when only some of those interventions would meet FDA criteria for efficacy and specificity. In such instances, we are reluctant to declare a treatment efficacious (much less specific) if we cannot find a single well-formed study that supports that conclusion. To do so would be tantamount to accepting the null hypothesis and interpreting the absence of differences between treatments as evidence of comparable efficacy (or specificity). We highlight this distinction when it occurs in the following review.
Despite our preference for the FDA approach, we do think that meta-analyses have considerable value in highlighting the relative impact of different control conditions and other related factors. For example, Cuijpers and colleagues reported that comparisons between all psychotherapies aggregated versus no treatment controls generated an effect size of d = .88 across the depression literature. This can be translated into a number-needed-to-treat (NNT) of 2.15 and means that one additional patient gets better for just over every two patients treated relative to what would have happened in the absence of treatment. Such comparisons are sufficient to establish efficacy. Non-specific controls (including especially pill-placebos) more than halved the effect size (d = .35), which translates into a considerably larger NNT of 5.15. This means that over twice as many patients would need to be treated to produce one additional positive outcome relative to comparison conditions that mobilize nonspecific factors associated with going into treatment. This is still quite respectable; by way of contrast, antihypertensive medications produce an NNT of 15. Such comparisons are necessary to establish specificity. In brief, it is easier to show that something works than it is to establish that it works for specific reasons that go beyond the mere provision of treatment. Curiously, comparisons to “treatment-as-usual” were associated with an intermediate effect size of d = .52 with a corresponding NNT of 3.50. This likely reflects the fact that what passes as “treatment-as-usual” can be quite heterogeneous across studies (and even across patients within studies), ranging from minimal contact to rather extensive care. It is important to note that all such indices are relative in nature and can only be interpreted in terms of the comparison or control conditions against which they were generated.
The method of this review is very similar to those of two other reviews we carried out to determine which psychological therapies are empirically supported for adults with social phobia or acute stress and posttraumatic stress disorders. We carried out a literature search of the PsycINFO and PubMed databases and the reference sections of chapters and journal articles to identify randomized controlled trials (RCTs) of psychological therapies for the mood disorders through the end of 2009. Trials were included only if adult patients with a diagnosed mood disorder were randomly allocated to different treatments including one or more psychosocial interventions. The method of the intervention had to be clearly described and the articles written in English. Therapeutic approaches evaluated in the trials were classified as dynamic, interpersonal, cognitive-behavioral, behavioral, experiential-humanistic, marital/family, or psycho-educational (see Figure 1). Chambless & Hollon's (1998) criteria were used to draw conclusions about the efficacy of each.
As shown in Table 1, 101 RCTs were identified with respect to MDD. These trials evaluated the efficacy of dynamic psychotherapy (N=17), interpersonal psychotherapy (IPT) (N=19), cognitive-behavior therapy (CBT) (N=64), behavior therapy (BT) (N=22), experiential-humanistic psychotherapy (N=6), and marital/family therapy (N=2).
The dynamic psychotherapies represent the oldest treatments for depression. Early findings were generally unimpressive, although the approach was often included as a comparator by investigators with other allegiances. For example, McLean and Hakstian found brief dynamic therapy less efficacious than their preferred behavioral intervention or medications and Covi and colleagues found dynamic group psychotherapy no better than placebo and less efficacious than medications in one study and less efficacious than CBT (with or without medications) in another. No differences were found relative to social-skills training or self-control therapy in a pair of studies implemented by behaviorally-oriented researchers and few differences were found between psychodynamic interpersonal psychotherapy and CBT in a pair of studies by investigators with little expertise with CBT.[19,20]
Recent studies by investigators with expertise in dynamic psychotherapy have been somewhat more supportive but still less than wholly compelling. A study by Cooper and colleagues in England found that psychodynamic psychotherapy did not differ from CBT or non-directive counseling and that each produced greater change on measures of depression than did routine primary care in the treatment of postpartum depression. A study by Burnand and colleagues in Switzerland found that adding dynamic psychotherapy to medication reduced the proportion of patients who met criteria for MDD following treatment and led to better work adjustment, although there were no differences on measures of depressive symptoms. A study by Salminen and colleagues in Finland on patients with mild-to-moderate MDD in a general practice setting found no differences between short-term dynamic psychotherapy versus fluoxetine antidepressant medication, but the sample was too small to draw firm conclusions. De Jonghe and colleagues in the Netherlands found that adding short-term dynamic psychotherapy increased the proportion of patients responding to medications by virtue of reducing rates of attrition and that patients with personality disorders may have been more likely to respond to combined treatment than to medications alone. A subsequent study by this same group found that antidepressant medications worked more rapidly than short-term dynamic psychotherapy and were superior after eight weeks of treatment. Maina and colleagues in Italy found that brief dynamic therapy was no more efficacious than brief supportive psychotherapy when added to medications at the end of treatment, but that patients continued to improve over a subsequent six-month continuation phase and that patients previously treated with dynamic psychotherapy were less likely to experience a recurrence over a subsequent 48-month treatment-free follow-up.
One of these studies found clear evidence of efficacy relative to routine care and adding psychodynamic psychotherapy enhanced the efficacy of medication on at least some measures in a second, and for at least some patients in a third.[24,25] This is better than it had done in those earlier trials. Perhaps most interesting were the indications that patients treated with brief dynamic psychotherapy plus medications continued to improve after the end of active treatment and were less likely to recur than patients previously treated with supportive psychotherapy plus medications. At the same time, none of these studies found dynamic psychotherapy superior to a nonspecific control or alternative treatment; results were more promising than early trials but hardly impressive.
Dynamic psychotherapy has rarely been tested in the treatment of geriatric depression, but the samples studied have been so small and the quality of the alternative interventions suspect; it is not clear that anything but null findings would have been expected. Gallagher and Thompson found few differences between brief dynamic therapy and either CBT or BT, findings replicated in a second study in which all three active treatments pooled were superior to a wait list control. Treatment gains produced by either CBT or BT were better maintained than those produced by dynamic therapy in the first study but not in the second. A third study by this group found that short-term caregivers did better in brief psychodynamic psychotherapy than they did in CBT, whereas long-term caregivers showed the opposite pattern. Conversely, Steuer and colleagues found CBT superior to dynamic psychotherapy delivered in a groups. There is simply little in this literature to warrant a designation of efficacious or specific.
On the whole, although there is still not compelling evidence speaking to the efficacy of dynamic psychotherapy, it would be premature to conclude that it is not efficacious solely on the basis of early trials by advocates of other approaches. More recent studies by investigators who have an investment in and expertise with the approach do offer some limited support and meta-analyses that aggregate across studies regardless of quality find it no less efficacious than alternative types of psychotherapies.10,11 Although none of the studies in the literature provide strong support for the approach relative to either medications or alternative psychotherapies, one study does suggest an advantage over routine primary care and two others suggest that it can enhance the efficacy of medications on at least some measures and for at least some patients.[24,25] Yet another recent study suggests that its effect may build over time and protect against subsequent recurrence. It seems fair to say that dynamic psychotherapy is possibly efficacious with respect to acute response and the prevention of subsequent relapse/recurrence.
Interpersonal psychotherapy (IPT) springs from dynamic roots, but draws on attachment theory and theoretical revisions that focus on interpersonal relationships. It is more structured than dynamic psychotherapy (but less so than cognitive or behavioral approaches) and focuses on current interpersonal difficulties rather than childhood recollections or the therapeutic relationship.
IPT has fared well in a series of controlled trials in fully clinical populations. Klerman and colleagues found that patients treated to remission with the combination of IPT and medications were no more likely to relapse if continued on IPT alone than if continued on medications and patients treated with IPT showed a greater (if somewhat delayed) improvement in interpersonal functioning than did patients treated with medications alone. In a subsequent study, Weissman and colleagues found that outpatients treated with IPT did as well as patients treated with medications and better than patients provided with treatment-on-demand in terms of symptom reduction and that patients treated with combined treatment did better still. Drugs produced more rapid change, but IPT again had a delayed effect on interpersonal functioning. This study speaks to the efficacy of IPT in the reduction of acute symptoms.
IPT also fared well in the placebo-controlled NIMH Treatment of Depression Collaborative Research Project (TDCRP), one of the largest and most influential studies of its time. Among more severely depressed patients, both IPT and drugs outperformed pill-placebo, whereas CBT did not; no differences were evident among less severely depressed patients or in the sample as a whole. Once again, drugs produced faster change than IPT, which showed more change later in treatment. IPT also reduced depressive symptoms and improved social adjustment in women suffering from postpartum depression over wait list in one study and was superior to didactic parent education in a sample of pregnant women with MDD in another. This last study and the TDCRP suggest that IPT is efficacious and specific in the treatment of MDD.
Subsequent studies have not been as supportive. A study conducted in New Zealand found that IPT was less efficacious than CT for patients with more severe depression or perhaps personality disorders. Similarly, a recent Canadian trial found IPT less efficacious than medications. Internal analyses indicated that IPT did particularly poorly with patients high on self-criticism. Although efficacious and specific according to Chambless and Hollon's (1998) criteria, findings with respect to IPT are no longer as consistent as they once were when only advocates conducted trials on the approach.
Studies in special populations also are worthy of note. IPT was as efficacious as drugs (imipramine) plus supportive therapy and more efficacious than either CT or supportive psychotherapy alone in the treatment of depression in a sample of HIV-positive patients; this study would speak to both efficacy and specificity except that only about half the sample met criteria for MDD. Bolton and colleagues found that indigenous nonprofessionals in rural Uganda could be trained to provide group IPT to fellow villagers that reduced rates of depression and improved functioning and that these differences were sustained across a six month follow-up. IPT was as efficacious as medications (if somewhat slower acting) and more efficacious than treatment-as-usual in one study in a primary care setting, although training physicians to provide IPT-based education did little to enhance response to medication in a small general practice sample in another. Adding IPT enhanced response to medication in an inpatient sample, including patients with chronic depression, and there were indications that these differences extended beyond the end of treatment. Finally, a version of the approach adapted for depressed mothers of offspring with psychiatric disorders (IPT-MOMS) was more efficacious than treatment-as-usual in reducing depression in both the mothers and their offspring. On the other hand, Reynolds and colleagues found that IPT did not differ from pill-placebo and was less efficacious than medication in reducing acute distress in a “young” geriatric sample (aged 50 and above) with a history of recent bereavement and no better than usual care with respect to rates of remission or measures of symptom change in another study on a geriatric primary care sample aged 55 and over, although it did reduce the proportion of patients who continued to meet criteria for depression at posttreatment.
Frank and colleagues found monthly maintenance IPT superior to withdrawal onto pill-placebo in a sample of recurrent patients treated to recovery with combined treatment, but maintenance medication (imipramine) was more efficacious still and combined treatment did nothing to improve on medications alone. Maintenance IPT was most efficacious when the sessions maintained a high level of interpersonal focus suggesting the importance of quality of implementation. When this design was replicated in that same setting in a “young” geriatric sample (mainly 60 to 75 years of age), both maintenance IPT and maintenance medications were superior to pill-placebo, with combined treatment best of all. These studies suggest that maintenance IPT is possibly efficacious for the prevention of recurrence, although a subsequent replication found maintenance IPT no more efficacious than pill-placebo and less efficacious than maintenance medication in the treatment of depression in an older geriatric sample aged 70 and above. IPT was protective of cognitively impaired unmedicated elders.
Although negative findings do exist,[46-48] IPT appears to be efficacious and specific in the reduction of acute distress[42,45] and may forestall both relapse and recurrence so long as it is continued or maintained, although perhaps not so well as medications.[36,59,61] In some studies, combined treatment improved on drugs alone,[38,54] although that was not always the case. There also were indications that IPT has a delayed effect on interpersonal skills and relationship quality that builds over time.[37,40] This represents a specific benefit of IPT and may enhance its value as an adjunct to medications. It also appears to be efficacious in the treatment of perinatal depression.[44,45] This is important since pregnant and lactating women may have special reasons to prefer not to be on medication. Recent trials by investigators outside of the core IPT group have not been as uniformly supportive as earlier trials by advocates for the intervention, but the efficacy of the approach appears to be well established when implemented by well-trained therapists.
The cognitive behavioral therapies (CBT), of which cognitive therapy (CT) is the most widely practiced variant, assume that negative beliefs and maladaptive information processing contribute to the onset and maintenance of depression. These interventions seek to produce change by teaching patients to evaluate the accuracy of their beliefs (or the relation between their thoughts and feelings in the newer meditation-based approaches), often by using their own behaviors to test their beliefs. CBT has been tested extensively and typically found to be superior to minimal treatment controls and at least as efficacious as other empirically supported interventions. Nonetheless, questions remain as to how it compares to drugs in the treatment of severe depression.
Early studies suggested that CT might be superior to drugs, but often implemented medications in a less than adequate fashion.[66,67] The same appeared to be the case in a later trial that found both CT and relaxation training (included as a nonspecific control) superior to tricyclic ADM in a very small sample with an uncharacteristically poor response to medication. These studies could be taken as support for the specific efficacy of CT since even inadequately implemented medication conditions should have controlled for nonspecific factors, but we are not prepared to go so far. Subsequent studies suggested that CT and drugs are comparable in efficacy when each is adequately implemented[69,70] and an even more recent study suggests that the same may be true for rational emotive behavior therapy (REBT), with either type of psychotherapy more cost-effective. As previously described, CT was less efficacious than either IPT or medications and no more efficacious than pill-placebo in the treatment of severe depression in the TDCRP, the largest and best controlled study of its time,[41,42] but response to treatment varied across sites and CT did as well as medication in the site with more experienced cognitive therapists. DeRubeis and colleagues reanalyzed individual response data on severely depressed patients from the studies just cited and found no differences between CT and drugs across the pooled samples. However, we are reluctant to base a claim of efficacy solely on equivalence to an established standard.
A more recent trial by Jarrett and colleagues found CT as efficacious as medications and superior to pill-placebo in patients with atypical depression and a subsequent study by DeRubeis and colleagues essentially replicated these findings among patients with more severe depressions. These trials are important because they demonstrate that CT can do as well as medications in fully clinical samples that respond to medications. The fact that CT was superior to pill-placebo in each speaks to both efficacy and specificity. An even more recent trial from Iran found CBT superior to medications and both superior to a no treatment control in a sample of depressed women with fertility problems.
However, the efficacy of CT may be moderated both by patient characteristics and therapist experience. In the study by DeRubeis and colleagues, patients with Axis II disorders did better on medications than they did in CT, whereas patients free from such disorders showed the opposite pattern. Moreover, CT was less efficacious than medications at the site with less experienced cognitive therapists (see also the study by Bright and colleagues). This is reminiscent of earlier findings from the TDCRP and consistent with the poor showing by somewhat less experienced cognitive therapists with more severe and complicated patients in a placebo-controlled comparison to medication or behavioral activation described in a subsequent section. Similarly, Bagby and colleagues found that patients with higher neuroticism scores did better on medications than they did in CBT in a study that otherwise found no main effect for treatment. On the other hand, a recent trial from New Zealand found that CT was more efficacious than IPT among patients with more severe depression or Axis II disorders. Although the therapists in that trial were all experienced, it is not clear just how expert they were with either treatment. These findings suggest that CT's efficacy with more severe and complicated patients may vary across studies and depend in part on therapist experience.
Another recent study found CT as efficacious as drugs in recurrent patients and adding CT typically enhanced the efficacy of medication treatment in inpatient samples.[84,85] Studies in primary care settings have found that adding CT typically enhances the efficacy of usual care[86,87] and did so in one study over and above the benefits provided by a contact control, although that has not always been the case. CT was as efficacious as medications and superior to community referral in a sample of mostly low-income minority women with MDD and its effects extended across a one-year follow-up. CBT was superior to treatment-as-usual (TAU) among severely depressed outpatients. In general, these findings are consistent with the notion that CBT is efficacious (if not necessarily specific) in the treatment of MDD.
Patients treated to remission with CT are less likely to relapse following treatment termination than patients treated to remission with drugs alone;[93-95] the magnitude of this effect is at least as great as keeping patients on continuation medication and superior to placebo withdrawal.[97,98] Only the TDCRP failed to find an enduring effect for prior CT. Moreover, these effects may extend to the prevention of recurrence, although comparisons to placebo controls typically do not extend beyond the period of risk for relapse.[97,98] These studies indicate that CT has an enduring effect that is both efficacious and specific in the prevention of relapse and efficacious with respect to recurrence.
Studies also have shown that CBT can be added following initial medication treatment to prevent subsequent symptom return and that this enduring effect can last for up to several years.[100-103] Providing group CBT to remitted patients reduced risk for subsequent relapse or recurrence among patients with more prior episodes and a similar moderated effect was found for acute CBT followed by brief psychoeducation (but not psychoeducation alone) for patients with four or more prior episodes. An earlier trial by this latter group found no differences between a depression recurrence prevention program with or without CBT relative to usual care in a general practice sample. The only studies that failed to find an enduring effect for CBT provided following remission compared it to continuation medication.[107,108]
Teasdale and colleagues have shown in two studies that training in mindfulness-based cognitive therapy (MBCT) can reduce risk for relapse or recurrence in patients initially treated with medications.[109,110] MBCT had its strongest preventive effects on patients with three or more prior episodes, a pattern of moderation that suggests that it may work through different mechanisms than standard CBT. A subsequent trial found MBCT more effective than maintenance medication in reducing residual depressive symptoms and improving quality of life; 75% of the MBCT patients in that trial were able to discontinue medications. Differences in rates of relapse/recurrence favored MBCT but were not significant. Given that multiple sites were involved in one of the trials, we consider MBCT to be efficacious in the prevention of relapse/recurrence.
Combined treatment typically retains the specific benefits of either modality alone (more rapid or robust change for drugs versus more enduring change for CT), but differences in acute response relative to either monotherapy were not believed to be all that large.[112,113] However, more recent studies suggest a larger incremental effect; d=.25 relative to psychotherapy alone with an NNT = 7.14 and d=.31 relative to medications alone with an NNT of 5.75. It was a recent trial by Keller and colleagues that renewed interest in combined treatment. In that study, the combination of drugs (nefazodone) and a novel cognitive behavioral analysis system for psychotherapy (CBASP) targeted at interpersonal change and incorporating dynamic elements was more efficacious than either monotherapy in patients with chronic depression. Drugs worked best early on, while CBASP worked better late, and combined treatment retained the temporal advantages of each. This study suggests that CBASP is possibly efficacious in the treatment of chronic depression and there was no indication that meeting criteria for a personality disorder did anything to moderate the effects of treatment. Continuing CBASP after recovery reduced risk for recurrence, as does continuing CT with respect to relapse and recurrence. These trials suggested that maintenance CBASP is possibly efficacious in the prevention of recurrence and that continuation CT is possibly efficacious in the prevention of relapse and recurrence. However, a subsequent effort at replication by several of the same investigators found that augmentation with CBASP was no more efficacious than individualized pharmacotherapy alone in chronic patients who failed to respond to three months of initial medication treatment.
CBT was superior to sleep hygiene with respect to remission in both depression and insomnia when each was added to medication in the treatment of patients diagnosed with both MDD and insomnia and superior to treatment-as-usual (often involving antidepressant medications) in the treatment of depression following coronary bypass surgery (especially among the two-thirds of the sample that met criteria for MDD). Dozois and colleagues found that adding CT did little to enhance the efficacy of algorithm-driven pharmacotherapy in a small sample but did produce greater change in underlying cognitive structure. Thompson and colleagues found that combined treatment was superior to medication alone in a geriatric sample with CBT alone intermediate and different from neither. Laidlaw and colleagues found that geriatric patients treated with CBT alone were less likely to meet criterion for MDD than patients treated with treatment-as-usual including medication in a general practice setting.
Cognitive therapy did not differ from medication as a second-step treatment for patients who did not respond to citalopram in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) project, although augmentation with medication resulted in significantly more rapid remission than augmentation with CT. Rohan found CT comparable to light therapy and both superior to a wait list control in a sample of patients with seasonal affective disorder. Strauman and colleagues found no overall differences between CT and their preferred self-system therapy (SST), although the latter was superior to CT for some patients. Two studies have found that computer-assisted CT was as efficacious as therapist-administered CT with both superior to a wait list control.[129,130] Finally, as previously noted, a recent trial found REBT comparable to either CT or fluoxetine in the treatment of MDD and superior to continuation medication at a cross-sectional six month follow-up. We are reluctant to categorize REBT as possibly efficacious with respect to acute response on the basis of null findings in a single trial or with respect to the prevention of relapse on the basis of cross-sectional assessment (since patients had ample time to relapse and subsequently remit in the interim) but note that the findings for REBT in this study were promising and merit further consideration.
On the whole, it appears that CBT (and especially CT) is as efficacious and specific as medications in the treatment of MDD,[75,76] although therapist competence may be an important moderating factor with more severe or complicated patients.[42,81] Early indications of general superiority to medications[66-68] or any specific inferiority among more severely depressed patients (the latter from the TDCRP) have not held up in subsequent trials in which each modality was adequately implemented.[69-71,75,76, 83] There are consistent indications that CT has an enduring effect that protects against subsequent relapse and possibly recurrence regardless of when it is applied[93-98,100-105] and indications that the same might be the case for MBCT.[109-111] This is especially important given the recurrent nature of depression and the fact that medications appear to have no lasting effect following treatment termination. Continuation/maintenance CT has been found to reduce risk for relapse/recurrence in MDD and CBASP has been found to reduce acute distress and subsequent recurrence when maintained in chronic MDD in single studies and can be said to be possibly efficacious.
Behavioral interventions include contextual approaches based on functional analyses (contingency management and behavioral activation), social skills training, self-control therapy, problem-solving therapy, and behavioral marital therapy. Although these approaches have not been tested as extensively as CBT, they have generally done well in controlled trials. BT typically has been found to be superior to minimal treatment and at least as efficacious as other interventions, but studies in fully clinical populations have been few and comparison treatments sometimes suspect.
As previously described, McLean and Hakstian found a modest advantage for contingency management relative to drugs alone or brief dynamic psychotherapy, but dosages were low and the dynamic intervention questionable. Hersen and colleagues found no differences between social skills training (SST) with or without medications and either amitriptyline alone or brief dynamic psychotherapy when each was adequately implemented. Kornblith and colleagues found no differences between self-control therapy (SCT) and dynamic therapy, whereas the addition of self-control therapy (SCT) enhanced response relative to usual care in a day-treatment program.
Nezu and colleagues found problem-solving therapy (PST) superior to nonspecific or wait list controls in two studies with recruited adults,[132,133] as did Arean and colleagues in a geriatric sample. Mynors-Wallis and colleagues found PST comparable to drugs and superior to placebo in one study in a general practice sample and comparable to medications in another. A large multicenter randomized trial by Dowrick and colleagues found PST superior to an assessment only control in reducing levels of depression in participants across five European countries. These studies suggest in aggregate that PST is efficacious and possibly specific in the treatment of MDD. The fact that only one used a pill-placebo control in a bona-fide clinical sample reduces our confidence in this conclusion somewhat since nonspecific psychological controls are hard to implement in a convincing fashion and recruited samples reduce generalizability.
Behavior marital therapy (BMT) was as efficacious as CBT and superior to a wait list control in the treatment of depression in couples with marital distress.[138,139] A second study found BMT as efficacious as CBT in reducing depression for women with marital distress, but less efficacious than CBT for women without marital problems[140,141] and a third found no differences between the two on measures of depression. These studies fall short of suggesting that BMT is possibly efficacious in the treatment of MDD since the first recruited patients with either major depression or dysthymia and the others had small sample sizes that limited the conclusions one could draw on the basis of null findings. Nonetheless, BMT was more efficacious than CT in reducing marital distress in all three studies. A recent trial found coping-oriented couples therapy (COCT) that included many elements of behavioral marital therapy comparable to either CT or IPT in terms of the reduction of depressive symptoms but no better with respect to the resolution of marital distress. However, this study suffered from a small sample size.
Despite these early successes, interest in BT stagnated before Jacobson and colleagues found that the behavioral activation component of CT produced as much change during acute treatment as the full treatment package, with no differences in rates of subsequent relapse. These findings were so unexpected that they led Jacobson and colleagues to conduct a placebo-controlled trial to compare a more comprehensive contextual version of behavioral activation (BA) against both CT alone and medications. In that trial, Dimidjian and colleagues found that BA and medications were comparably efficacious and each was superior to CT or pill-placebo in the treatment of more severely depressed patients. Moreover, Dobson and colleagues found that among remitted patients, those previously treated with BA were no more likely to relapse following treatment termination than patients previously treated with CT or than patients kept on continuation medications and showed a marginal advantage relative to medication withdrawal with respect to both relapse and recurrence. Had the effect been fully significant or the sample larger we would have been tempted to suggest that BA has an enduring effect with respect to the prevention of subsequent relapse or recurrence, but we are not prepared to go quite that far as yet. Nonetheless, these findings are promising in that respect and deserve to be followed up in future studies.
The fact that prior exposure to CT showed an enduring effect in Dobson and colleagues relative to withdrawal onto a pill-placebo for previously medicated patients goes along with findings from Hollon and colleagues to establish the specific efficacy of prior CT with respect to the prevention of subsequent relapse.[97,98] CT's poor performance with more severe and complicated patients was reminiscent of earlier findings from the TDCRP and may reflect the difficulties inherent when less experienced cognitive therapists try to implement a complicated treatment in a time-limited fashion. That being said it is worth noting that BA encountered no such difficulties. Given that BA appears to be less complex and easier to learn than CT, this study has generated renewed interest in behavior therapy as a cost-effective alternative to medication in the treatment of MDD and another recent trial found behavioral activation superior to supportive psychotherapy in an inpatient sample. Combined with the results from the earlier study by McLean and Hakstian, these studies indicate that behavior activation is efficacious and specific in the treatment of MDD.
On the whole, these studies suggest that BT is efficacious and specific in the treatment of MDD. The evidence is most compelling for the contextual approaches (BA and contingency management) that have been tested against other efficacious interventions in fully clinical samples. [14,81,147] PST also met our criteria for being efficacious and specific, although the supporting studies come from recruited volunteers[132,134] or general practice samples. BMT falls short of being possibly efficacious with respect to depression (although it is clearly efficacious for marital distress) since outcomes are mixed at best and the samples too small for equivalence to support efficacy.[138-143]
Although widely practiced in clinical settings, experiential-humanistic psychotherapy has been tested only rarely in the treatment of depression but has shown promise in recent trials. Beutler and colleagues found modest main effects for CBT and better response among resistant patients to a self-directed control but little specific benefit for a focused expressive psychotherapy based on Gestalt principles. Greenberg and Watson found no difference between client-centered psychotherapy or a process-experiential therapy (PET) that incorporated gestalt principles on measures of depression, although PET produced greater change on measures of interpersonal processes and self-esteem. Watson and colleagues similarly found no differences between PET and CT (adequacy unknown) on measures of depression, although PET again produced greater change on self-reports of interpersonal problems. That same research group found emotion-focused therapy (EFT) superior to client-centered therapy (CCT) and that responders to EFT were less likely to relapse over the next 18 months than responders to CCT. This is the most promising trial in a rather limited literature and suggests that experiential therapy might be possibly efficacious with respect to both acute response and subsequent prevention.
Castonguay and colleagues added humanistic and interpersonal strategies to standard CT to repair ruptures in the working alliance and found this integrative approach superior to a wait list control in one study  and superior to standard CT in a second. While the approach is not wholly experiential in nature and the studies relied on relatively inexperienced graduate student therapists, these findings do suggest that humanistic and interpersonal strategies might be a useful adjunct to CT.
Marital and family problems are common in depression and may contribute to its etiology and complicate its treatment. Nonetheless, traditional marital and family therapies have been little studied in the treatment of depression. Friedman found that drugs were better than marital therapy in the reduction of acute distress, whereas marital therapy produced greater changes in the quality of relationships. Clarkin and colleagues found that adding psycho-educational family therapy designed to reduce criticism of the patient enhanced response to standard inpatient treatment (including medication) among female patients (males actually did worse), but that only patients with bipolar disorders retained those gains. Traditional marital and family therapy may have a role to play in the treatment of MDD, but the studies are too few and the findings too mixed to draw firm conclusions.
IPT, CBT (especially CT), and two different variants of BT (BA and maybe PST) are as efficacious as medications (even among more severely depressed patients) and specific in the treatment of MDD, although those findings may be conditioned upon therapist experience and patient characteristics (at least for CT). Results for dynamic and experiential-humanistic interventions are mixed (with later studies conducted by advocates rising to the level of possible efficacy) and marital and family therapies rarely tested. There are indications that IPT (and possibly BMT) may improve the quality of interpersonal relationships and that CT and MBCT (and possibly BA) have enduring effects (specific for CT) that last beyond the end of treatment and may extend to the prevention of recurrence. This suggests a greater breadth and stability of response than found for drugs and makes these interventions attractive alternatives or additions to medications. CBASP is possibly efficacious as an adjunct to medication in the treatment of chronic depression and preventive so long as it is maintained and continuation CT also appears to be possibly efficacious in the prevention of relapse and subsequent recurrence.
As shown in Table 2, nine RCTs were identified that satisfied our inclusion criteria with respect to dysthymic disorder (DD). One evaluated dynamic psychotherapy and included a humanistic comparison condition, four evaluated IPT, and two each evaluated CBT and PST respectively. All but two involved comparisons to antidepressant medications.
A recent trial found both brief dynamic therapy (BDT) and brief supportive psychotherapy (BSP) superior to a wait list control over nine months of acute treatment and BDT superior to BSP across a six-month follow-up in a sample of patients with less severe depression. Only about a third of the sample met criteria for DD, so as to preclude drawing conclusions about the efficacy of BDT for this disorder.
IPT has been modified for DD by conceptualizing lifelong traits as chronic but treatable states. Results have been mixed. Adding IPT produced a small but nonsignificant advantage relative to medications alone in the treatment of a small sample of dysthymic patients; most also met criteria for MDD. A subsequent study found that IPT was less efficacious than drugs alone and did nothing to enhance efficacy when added in combination. A third study found IPT less efficacious than medications and no more efficacious than a brief supportive control, although there was turnover from more to less experienced IPT therapists over the course of the study. A fourth study found IPT more efficacious than brief supportive psychotherapy in a sample of dysthymic patients with secondary substance abuse or dependence. This last study is sufficient to meet criteria for possible efficacy, but the bulk of this literature suggests that IPT is less efficacious than and adds little to drugs in the treatment of DD.
CBT has been adapted in recent years to deal with chronic problems like dysthymia and underlying personality disorders, but has yet to be formally tested. Similarly, recent work by McCullough developing CBASP for patients with chronic depression will likely have relevance for patients with dysthymia as well. These developments are noteworthy because conventional CBT did little to enhance the efficacy of medications and did not separate from placebo in one study with a large sample and did not differ from medications in another smaller trial.
Two studies with largely overlapping protocols have explored the effectiveness of problem-solving therapy (PST) in the treatment of dysthymia. In each, six sessions of PST was compared to antidepressant medication in the context of a pill-placebo control in samples that included patients who met criteria for either dysthymia or minor depression. Barrett and colleagues found no differences between the three conditions in reducing depressive symptoms although both PST and medication treatment produced greater rates of response than did pill-placebo in an adult sample under the age of 60. Williams and colleagues found medications but not PST superior to pill-placebo in a geriatric sample over the age of 60. Neither study provides much support for PST in primary care, although session number was quite low (only six sessions).
Efforts to modify existing interventions for use with dysthymia are relatively recent and show mixed results at best. One would think that the same interventions that are efficacious for MDD also would prove to be of use with dysthymia, but results to date have not been encouraging, although IPT may be considered possibly efficacious for the treatment of DD. It should be noted that few of the trials testing IPT or CBT were conducted by investigators expert in those approaches (Markowitz being the sole exception) and that the studies involving PST compared a very limited number of sessions to full-dose medication. Whether subsequent studies will provide greater support for those interventions remains to be seen, but neither the studies that have been conducted to date nor the results they obtained have been all that impressive.
Whereas the psychosocial interventions represent a viable alternative to drugs in the treatment of MDD, they have only recently begun to be explored as adjuncts to medication in bipolar patients. There is little evidence that the psychosocial interventions can forestall the onset of manic episodes (with the possible exception of psychoeducation intended to enhance medication adherence) and mood stabilizers like lithium and the anticonvulsants remain the standard of treatment. Nonetheless, in the face of growing evidence that drugs are less than wholly adequate, there has been increased interest in adapting psychosocial interventions to the treatment of bipolar disorder. These efforts have focused on the newer interpersonal, cognitive-behavioral, and family-focused interventions, as well as psychoeducation. As shown in Table 3, fifteen RCTs satisfied our inclusion criteria with respect to BD, including three for psychoeducation (PE), two for IPT, eight for CBT, three for family-focused therapy (FFT). These studies are sometimes hard to categorize, since patients may or may not be in episode at the time of randomization (making it difficult to determine whether the trial is focused on acute treatment or subsequent prevention) and cumulative episode onset often is analyzed over the course of several years (making it hard to separate relapse from recurrence). Moreover, only some of the studies report separate analyses as a function of polarity and talk instead of bipolar episodes (making it hard to tell whether treatment effects pertain to mania or depression). With those caveats in mind we review the relevant studies.
Adding simple psychoeducation (PE) in which patients are trained to identify prodromes and to seek prompt treatment was found to reduce risk for relapse/recurrence with respect to mania but not depression over routine care in a sample of patients who had relapsed within the previous year. A subsequent study in a sample of euthymic patients past the period of risk for relapse essentially replicated these findings relative to recurrence with respect to mania and hypomania and also found evidence of prevention with respect to depression relative to an unstructured nonspecific support group and a subsequent reanalysis found that these results held even among the small subset of bipolar II patients across a five-year follow-up. Providing PE to caregivers of euthymic patients with BD who had been euthymic for the last three months resulted in lower rates of relapse/recurrence with respect to mania and hypomania, but not depression. These studies suggest that PE is efficacious as an adjunct to medications for the prevention of subsequent episodes of mania and hypomania in BD. Since patients in all three studies were euthymic at the time of entry it is not possible to tell whether PE had an impact on the reduction of acute distress. Moreover, since not all patients had been euthymic long enough (at least six months) at study entry to be past the point of risk for relapse and since the analyses typically aggregated cumulative episodes across time (the study by Colom and colleagues being the exception in both respects), it was difficult to distinguish between relapse and recurrence in these trials. Therefore, we are inclined to conclude that PE is efficacious in the prevention of undifferentiated relapse/recurrence with respect to mania/hypomania in the treatment of BD (with the study by Colom and colleagues suggesting that this effect might be specific relative to controls and fully evident with respect to recurrence)[170,171,173] and possibly efficacious in the prevention of recurrence in depression (the latter based solely on Colom and colleagues).
Frank and colleagues modified IPT to serve as an adjunct to drugs in the treatment of BD based on the notion that social interactions (social Zeitgebers) provide order in daily life and to help maintain affective stability. Activity scheduling and efforts to regularize sleep were added to conventional IPT to produce a hybrid intervention called interpersonal social rhythms therapy (IPSRT). Frank and colleagues found no differences between IPSRT and intensive clinical management (ICM) in the time it took to stabilize symptoms (patients could be depressed, manic, or mixed), but did find that patients who received IPSRT during acute treatment went longer without another episode of mood disturbance during a subsequent two-year maintenance phase, regardless of whether they were kept on IPSRT or switched to ICM. Additionally, more regularized social rhythms over acute treatment predicted lower risk for recurrence during maintenance, suggestive of mediation. Given that half of the recurrences involved manic (31%) or mixed episodes (19%) it would appear that any preventive effect for acute phase IPSRT was not limited to depression. Two things are important to note. First, despite its name, ICM was not all that different from the standard clinical management typically provided as part of pharmacotherapy in the unipolar literature. Although implemented by the same therapists as IPSRT, ICM involved only about half as much contact time and focused largely on issues related to medication management. Therefore, we do not consider it sufficient to speak to specificity. Second, although patients in this trial were said to enter maintenance treatment and to be at risk for recurrence after four weeks of stabilization, the convention in the literature is to refer to the first several months of treatment following remission as continuation treatment (rather than maintenance) and to view patients as being at risk for relapse (the return of the treated episode) rather than recurrence (the onset of a wholly new episode). We therefore are prepared to say that IPSRT is possibly efficacious in the prevention of relapse/recurrence with respect to bipolar episodes, but cannot differentiate between relapse and recurrence or mania and depression in this study.
A subsequent effectiveness study compared up to 30 sessions of IPSRT (or CBT or FFT) across 9 months to an abbreviated collaborative care (three sessions across six weeks) in a sample of medicated bipolar patients (I or II) who all met criteria for current major depressive episode as part of the Systematic Treatment Enhancement Program for Bipolar Disorder (STEP-BD). The three intensive psychotherapies pooled were superior to collaborative care in terms of both time to recovery and proportion of patients recovered and there were no differences among the intensive treatments. Although not reported in the published article, the authors did find that both IPSRT and FFT (but not CBT) were superior to collaborative care in post hoc analyses that did not control for multiple comparisons (Miklowitz, personal communication, December 2009). Given the brevity of the collaborative care condition, it is not clear that this study can be used to speak to specificity, since it provided only minimal control for nonspecific factors, but it is interesting that the results observed were achieved with relatively minimal training and low-intensity supervision (that is why the trial is classified as an effectiveness and not an efficacy study). On the whole, we are prepared to say that IPSRT is possibly efficacious as an adjunct to medication in the acute treatment of depression in BD.
As for IPT, there is growing interest in adapting CBT for the treatment of bipolar disorder. One early study found that CBT could be used to enhance adherence to medication and reduce rates of hospitalization but did not find a difference with respect to the frequency of affective episodes. More recent work also has focused on regularizing everyday routines and coping with negative life events. In an early pilot study, adding CT to treatment-as-usual (which in most cases involved mood stabilizers and psychiatric support) improved global functioning and reduced depressive symptoms in a group of patients with bipolar disorder heterogeneous with respect to current affective state. The addition of CT plus emotive techniques to mood stabilizer medication reduced levels of depression and delayed depressive relapse at the level of a nonsignificant trend in an Australian study and adding CBT to PE in bipolar patients currently stabilized on medications resulted in reduced symptom levels, 50% fewer days spent depressed, and fewer medication increases in a Canadian study. All three studies speak to the efficacy (but not the specificity) of CBT with respect to the reduction of depressive symptoms; although suggestive, none provides sufficient data to speak directly to the prevention of subsequent relapse. As previously noted, CBT (unlike IPSRT and FFT) did not differ from a less intensive collaborative care in the treatment of depression in bipolar patients stabilized on medications in the STEP-BD project.
In another early pilot trial, Lam and colleagues found that adding six months of CT to ongoing medication treatment reduced hospitalizations and the occurrence of bipolar episodes across the course of a year and improved residual functioning in a sample of bipolar I patients not currently in acute episode. A subsequent study in a larger sample found that adding CT to medications both reduced residual depression across the course of treatment and essentially replicated the findings from the pilot study just described with respect to the prevention of bipolar relapse (in this larger sample differences were significant with respect to both depressive and manic episodes) and hospitalizations. Although gains associated with CT were maintained across a subsequent 18-month follow-up (for depressive episodes but not for mania), there was no indication that prior CT had any additional preventive effect on subsequent recurrence. Given that both studies were done by the same group, they suggest that CT is possibly efficacious as an adjunct to medications in prevention of relapse/recurrence for bipolar episodes, but that those effects were limited with respect to mania (relapse only) and recurrence (differences with respect to depression were maintained but not enhanced across the extended follow-up suggesting that most of the effect was concentrated during the period of risk for relapse). It should be noted that what differentiated this study from others in the literature was that separate analyses were conducted with respect to both relapse and recurrence and both mania and depression; it is not clear that those other trials would have found a separate effect for recurrence over and above what they found for relapse had separate analyses been conducted.
However, in the largest trial to date in this literature, Scott and colleagues found that adding CBT provided no additional benefit over treatment-as-usual including medication in a sample of bipolar patients heterogeneous with respect to whether they were currently euthymic or in episode other than mania. Post hoc analyses did find a moderating effect for number of prior episodes, with CBT doing better than treatment-as-usual for patients with fewer prior episodes and worse for patients with more, although no specific tests of significance were conducted within those patient subgroups. Unlike the earlier trials by Lam and colleagues, which were largely restricted to euthymic patients currently stabilized on medications, about a third of the patients in Scott et al. were in episode at the time of randomization. This might have contributed to the differences in findings between the trials, something that Scott and colleagues could have addressed by conducting secondary analyses restricted to patients similar to those found in the earlier Lam studies. It is also possible that treatment was more competently implemented in those earlier trials, as therapists in the study by Scott and colleagues were unable to get through the full course of CBT with about 40% of the patients. As was the case for both IPT and CBT in the treatment of MDD, the literature points to efficacy in the acute treatment[180-182,184] and prevention of bipolar disorder (particularly with respect to depressive symptoms),[183,184] but a major study exists with findings to the contrary.
Traditional family therapy has not fared well as an adjunct to medication in the treatment of bipolar disorder. However, one of the more promising innovations in the treatment of bipolar disorder has been the adaptation of a family-focused therapy (FFT) originally developed to reduce the high levels of expressed emotion (criticism) in families of schizophrenic patients. As previously noted, the multi-center STEP-BD program found that FFT was superior to a less intensive collaborative care control in the treatment of depression in medicated bipolar patients.
A nine-month course of FFT has been shown to reduce depressive symptoms and risk for relapse through twelve months relative to a less intensive crisis management intervention in a sample of medicated bipolar patients recruited shortly after an illness episode (predominantly manic or mixed). The benefits of FFT extended across a second follow-up year to the prevention of recurrence with respect to both depression and mania, with the latter mediated by enhanced compliance with the medication regime. A subsequent study found that a similar nine-month course of FFT did not differ from individually-focused treatment matched for frequency and duration of contact (but not session length) across a one-year medication treatment phase but did reduce rates of recurrence and hospitalization across a subsequent one-year post-treatment follow-up. The fact that differences favoring FFT did not emerge until after the end of active treatment suggests that nonspecific factors may be sufficient to protect against relapse and enhance medication compliance during active treatment but that the specific benefits provided by a skills-training approach like FFT do not become evident until after the end of treatment. Both studies speak to the efficacy of FFT (as does STEP-BD) and the second speaks in part to its specificity. FFT can be said to be efficacious in the reduction of depressive symptoms[177,189] and efficacious (and enduring) in the prevention of subsequent relapse/recurrence in BD.[189-191]
Several of the newer psychosocial interventions including PE, IPSRT, CBT, and FFT appear to be useful adjuncts to medication in the treatment of BD. All focus on maintaining regular schedules and all seek to reduce interpersonal conflicts that can trigger episode onset. Controlled trials are still few in number and the evidence for preventive effects with respect to mania or hypomania is most clear for PE, but it does appear that the others each enhance the efficacy of drugs in the treatment of depressive symptoms and the prevention of relapse and recurrence. Given the chronic and episodic nature of BD and the disability that it can cause, any such indications are most welcome.
As shown in Table 4, IPT,[41,42,45,49] CBT (especially CT),[75,76] and at least two variants of BT (BA[14,81,147] and maybe PST[132,134,131]) meet the Chambless and Hollon criteria for being efficacious and specific in the treatment of MDD. In most instances these conclusions are based on placebo-controlled comparisons to medication or other bona fide therapies and not simply comparisons to nonspecific psychological control conditions that lack credibility to both patients and therapists. All appear to be comparable to drugs in the hands of experienced therapists and to enhance the efficacy of medications when added in combination (including CBASP for chronic depression). There was little support for more traditional dynamic, experiential-humanistic, or marital and family approaches in the treatment of MDD in the older literature, but more recent studies by investigators expert in the respective approaches have been more encouraging with respect to the short-term dynamic and experiential-humanistic psychotherapies. CBT[96-98] and MBCT[109-111] appear to have enduring effects that prevent subsequent relapse and possibly recurrence following treatment termination (with CBT specific)[97,98] and the same may be true for BDT and EFT although the evidence is scant for each. In a chronic recurrent disorder in which current practice now calls for keeping patients on medications indefinitely, this is a boon that could lead to major cost savings. Early trials provide little support for the efficacy of psychotherapy in the treatment of DD, although a recent study suggests that IPT is possibly efficacious. As adjuncts to medication in the treatment of BD, CBT[180-182,184] and FFT[177,189,190] are efficacious with respect to the reduction of depressive symptoms (with IPSRT possibly efficacious) and FFT efficacious with respect to the prevention of subsequent relapse and perhaps recurrence[189-191] (with IPSRT and CBT[183-185] possibly efficacious), with PE efficacious in the prevention of mania/hypomania[170,171,173] and possibly depression.
Our findings are consistent with and build upon those of earlier reviews that used similar criteria. DeRubeis & Crits-Christoph determined that CBT was efficacious and specific in the treatment of MDD and that both IPT and BA were efficacious and PST possibly efficacious. Numerous studies regarding MDD have been published since the time of their review and they covered neither dysthymia nor bipolar disorder. Chambless and Ollendick updated that review and included bipolar patients, but did not include dysthymia. Roth & Fonagy included all the mood disorders but did not differentiate between treatments that were efficacious and specific from those that were simply efficacious. They found evidence for the efficacy of both IPT and CBT for MDD and limited support for short-term psychodynamic psychotherapy. They also concluded that psychotherapy was less efficacious than medications alone in the treatment of dysthymia and noted some support for the efficacy of IPSRT, CBT, FFT, and PE as adjuncts to medication in bipolar disorder. None of these reviews provided separate classifications with respect to acute response and prevention of relapse and recurrence. Our conclusions also are largely consistent with those reached by the National Institute for Clinical Excellence (NICE) in England, although we are somewhat more positive about the dynamic and experiential/humanistic approaches based on our reading of recent trials.
The bulk of the RCTs of psychological interventions for mood disorders included in this review evaluated the efficacy of newer interpersonal or cognitive behavioral treatments. We still know little regarding the efficacy of more traditional dynamic or experiential-humanistic approaches and neither was always implemented adequately in those few studies that have been done. Absence of empirical support for these approaches does not necessarily mean that they offer no benefit to patients suffering from mood disorders, but further research is required to determine whether or not they do. 
It is especially important that these studies be done by research groups that include experts in the implementation of those interventions and that is starting to be the case. Quality of implementation of the respective interventions is perhaps the biggest source of variance with respect to outcomes in the literature and a major reason why we prefer the minimum number of positive studies approach adopted by the FDA to more conventional meta-analyses to summarize the literature. Those interventions like IPT or CBT that have been more extensively tested can survive the occasional study in which they are less than adequately implemented, but leaving research solely in the hands of advocates of other approaches has been a major problem for the more traditional dynamic and experiential-humanistic interventions. Absence of evidence is not necessarily evidence of absence with regard to efficacy, but absence of evidence will likely lead to absence of inclusion in any third-party or single-payer health care systems. RCTs are far from perfect and care must be used in interpreting their results (especially with respect to the adequacy of implementation of the respective interventions), but they represent the “gold standard” for drawing causal inferences regarding treatment efficacy. Advocates for the more traditional interventions avoid their use at the peril of their preferred interventions.
Finally, two recent meta-analytic reviews suggest that severity moderates the effects of both psychotherapy and medication treatment in a surprisingly similar fashion among non-bipolar patients. In placebo-controlled medication trials, “true” drug effects (drug-placebo differences) were negligible among patients with less severe depressions (with effect sizes less than d=.20 and NNTs greater than 10) but grew in magnitude as severity increased (with effect sizes greater than d=.50 and NNTs less than 4). Given that over half the patients who meet criteria for MDD fall on the “less severe” end of the severity continuum, this suggests that many patients for whom medications are prescribed derive no real “pharmacological” benefit from taking them. That is, when such patients improve (as they often do), they do so largely for nonspecific psychological reasons.
What is even more surprising is that the same appears to be the case for psychological treatments. When different types of psychotherapies prove to be superior to nonspecific controls (whether pill-placebos or nonspecific psychotherapy controls) such differences are only apparent among more severely depressed patients. Among randomized controlled trials with nonspecific controls that have tested for moderation, MDD patients with less severe depression show effect sizes that are small at best (d=.22; NNT = 8.06) whereas MDD patients with more severe depressions show effect sizes that are considerably larger (d=.63; NNT = 2.91). What this suggests is that nonspecific processes may be sufficient to produce change among patients with less severe depressions but that specific mechanisms that go beyond the simple provision of contact with a therapist and the expectation of change may be required for patients with more severe depressions.
These findings may have implications for mental health policy. It may be the case that many patients with less severe (or chronic) depressions can be adequately treated by less expert therapists using generic psychological interventions (as opposed to medications with their attendant side effects), whereas treatments with specific effects (whether pharmacological or psychotherapeutic) conducted by more expert therapists may be required for patients with more severe (or chronic) depressions or bipolar disorder. Moreover, given that CBT and possibly BA have enduring effects and IPT greater breadth of effect not found for medications and that neither appears to depend upon patient severity or chronicity, it may be wise to choose one of these approaches as the first line of treatment (rather than medications) as is currently being done in England.
This work was supported by Grant 1003551 to Myrna M. Weissman from the Josiah Macy Jr. Foundation and Independent Scientist Award MH01697 to Steven D. Hollon from the National Institute of Mental Health.