The Sequential Integration of Pharmacotherapy and Psychotherapy in the Treatment of Major Depressive Disorder: A Meta-Analysis of the Sequential Model and a Critical Review of the Literature

Published reports were identified with the use of electronic database searches. Keyword searches were conducted in EMBASE, PubMed, PsycINFO, Web of Science, and the Cochrane Library, from inception of each database through October 2014, combining the following terms: “sequential treatment,” “drugs and psychotherapy,” “combined treatment,” “continuation or maintenance,” “relapse or recurrence and prevention,” “depress* or major depress*,” selecting “adults” and “randomized controlled trials” as additional limits. Reference lists from relevant studies and reviews were examined for further clinical trials not yet identified. Authors of significant articles and other experts in the field were contacted.

Selection of studies was performed independently by two reviewers (J.G. and E.T.). Any disagreement was resolved in a meeting between the reviewers and a senior investigator (G.A.F.). We selected randomized controlled trials examining the efficacy of the sequential use of psychotherapy following response to acute-phase pharmacotherapy in the treatment of adult patients with major depressive disorder. The primary outcome measures were relapse or recurrence rates of depression as defined by study investigators (i.e., reaching a cut-off on any depression symptom rating scale used by authors and/or the occurrence of a defined episode of major depression after remission/recovery in acute-phase treatment) at the longest available follow-up.

We excluded studies if they 1) were not randomized controlled trials, 2) did not contain original data, or 3) did not primarily involve face-to-face delivery of psychotherapy. We also excluded studies in which relapse or recurrence rates were not identified categorically. Additionally, we excluded clinical trials of continuation and maintenance treatments for major depressive disorder in which psychotherapy was also administered during the acute phase, so that continuation-phase treatments matched the modality used during the initial phase.

Other criteria of exclusion were patients younger than 18 years old, studies that exclusively focused on the treatment of patients with bipolar disorder, dysthymic disorder, minor depressive disorder, or seasonal affective disorder, as well as studies including or exclusively focused on patients with predominant anxiety disorders, schizophrenia or other psychotic disorders, comorbid alcohol or substance use disorders, antisocial personality disorder, borderline personality disorder, or active medical illness.

Finally, studies that were judged to be dissimilar from other investigations on the basis of clinical characteristics of the intervention, such as use of ECT in addition to antidepressant treatment (15) or telephone- or Internet-based psychotherapy (16, 17), were excluded, as suggested by Jane-wit et al. (18).

Data were independently extracted by both reviewers with the use of a precoded form. The following data were extracted from studies meeting criteria for inclusion in the meta-analysis: age, gender distribution, methods used to define and diagnose study participants, and other inclusion criteria (i.e., recovered from a depressive episode or in remission); type of psychological intervention or control condition, number of patients randomly assigned to each treatment arm, treatment duration, and assessment times; and methods used to define relapse/recurrence and relapse/recurrence rates. The methodological quality of the included trials was assessed independently by both reviewers based on three basic criteria: random allocation of treatments, blinding of outcome assessment, and handling of attrition.

The primary outcome of the meta-analysis was efficacy of the sequential use of psychotherapy after pharmacotherapy, expressed in relapse or recurrence rates. Therefore, the risk ratio of relapse or recurrence and its standard error were calculated from each study. Examination of the pooled results was performed based on the random-effects model to increase the generalizability of findings, since this model is more conservative then the fixed-effects model. An alpha level of 0.05 was used for hypothesis tests. The number needed to treat was calculated as the reciprocal of the absolute risk difference for relapse or recurrence between patients treated with the sequential approach and the control group.

In addition to point estimates and confidence intervals, the Q statistic was performed to assess heterogeneity between study results. With this statistic, the null hypothesis is tested so that effect sizes from each of the studies were similar enough that a common population effect size could be calculated (19). However, the Q statistic only informs about the presence versus the absence of heterogeneity; it does not report on the extent of such heterogeneity. The I² statistic, which is an indicator of heterogeneity in percentages, was also calculated. A value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity, with 25% as low, 50% as moderate, and 75% as high heterogeneity (20).

The likelihood of significant publication bias was assessed through Begg’s funnel plot (21), and testing for asymmetry was assessed using Egger’s test statistic (22). Sensitivity analyses were implemented in order to estimate the influence of each study by deleting each in turn from the analysis and noting the degree to which the size and significance of the treatment effect changed. Meta-regression was performed to investigate how certain characteristics (i.e., drug continuation during psychotherapy, treatment duration, and time-points to data collection) acted to influence treatment effects. Finally, clinical heterogeneity between studies was explored performing subgroup analyses. All analyses were conducted using the user-written packages for meta-analysis available in Stata 10.1 (Stata Corporation, College Station, Tex.).

The initial search strategies identified 59 articles for potential inclusion in the meta-analysis (Figure 1). Of these, 13 studies (23–35) met criteria for inclusion in the meta-analysis. These studies reported relapse and/or recurrence rates for a total of 1,410 participants (728 patients in a sequential treatment arm and 682 in a control arm). The average age of participants was 45 years (SD=2.60), and 68.5% (range: 49.5%−81.0%) were female. They were judged as fully or partially remitted after acute-phase pharmacotherapy, based on clinical interviewing. Selected characteristics of these included studies are presented in Table 1. All studies involved cognitive-behavioral therapy (CBT) and its modifications. Three studies compared a sequential treatment arm with antidepressant medication and clinical management (26, 28, 30), six with treatment as usual (25, 27, 29, 31, 32, 35), one study with psycho-education and treatment as usual (34), and three with clinical management alone (23, 24, 33). Treatment as usual involved standard care as typically provided by the referring agencies (e.g., family doctors or other sources), with no restriction on the use of pharmacotherapy. Clinical management consisted of monitoring medication administration (including tapering antidepressant drugs), reviewing the patient’s clinical status, and providing the patient with limited support and advice if necessary, whereas specific interventions (e.g., exposure strategies, diary work, cognitive restructuring) were proscribed.

The methodological quality of these clinical trials was high. In all studies, participants were assigned at random to the conditions, and it was reported that assessors were not aware of the treatment to which patients were assigned. Intention-to-treat analyses were performed in nine studies (25–28, 30–34), while all patients were retained in two studies (23, 24), and completers’ data only were reported in two studies (29, 35).

We compared the effects of the sequential use of psychotherapy (either alone or in combination with antidepressant medication) with control conditions (Figure 2). The pooled risk ratio for relapse/recurrence was 0.781 (95% confidence interval [CI]=0.671–0.909; number needed to treat=8) in the random-effects model, suggesting a relative advantage in preventing relapse/recurrence (i.e., lower risk of relapse/recurrence) for the sequential administration of treatments compared with both active and nonactive controls. Heterogeneity across trials was not significant (Q=7.200, df=12, p=0.844). The I² statistic also indicated no significant heterogeneity (I²=0%) among the pooled studies. Visual inspection of both Begg’s funnel plot and Egger’s test (p=0.024) was suggestive for the presence of publication bias. A sensitivity analysis was performed to determine the contribution of each study to the overall effect size, and none of them appeared to markedly influence the observed risk ratio for relapse or recurrence.

Performing meta-regression analyses, we did not find any advantage of continuing medication during psychotherapy versus tapering and discontinuation (coefficient: 0.092; 95% CI=–0.096 to 0.280). We also tested for the duration of treatment, as well as for the length of follow-up, and we did not find significant effects on relapse/recurrence rates among the included studies (coefficient: 0.285; 95% CI=–0.128 to 0.185; coefficient: –0.012; 95% CI=–0.674 to 0.650, respectively).

Analyses were also performed in order to examine separately studies involving continuation of antidepressant drugs during psychotherapy and those with tapering and discontinuation.

Nine clinical trials contributed data for this subgroup analysis (25–29, 31, 32, 34, 35). Data showed a significant difference favoring the use of psychotherapy during continuation of antidepressant medication in reducing rates of relapse/recurrence compared with active control conditions (i.e., continuation of antidepressant medication) or treatment as usual. The pooled risk ratio for relapse was 0.811 (95% CI=0.685–0.961; number needed to treat=10) in the random-effects model. Both Q and I² statistics did not suggest any significant heterogeneity among the studies pooled (Q=4.842, df=8, p=0.774; I²=0%). Both Begg’s funnel plot and Egger’s test (p=0.186) did not indicate the presence of publication bias.

Four studies contributed data (23, 24, 30, 33). Subgroup analysis showed that patients randomly assigned to continuation-phase psychotherapy while antidepressant drugs were discontinued were significantly less likely to experience relapse/recurrence compared with both nonactive (i.e., clinical management) and active control (i.e., continuation of antidepressant medication) conditions. Across the trials, the pooled risk ratio for relapse was 0.674 (95% CI=0.482–0.943; number needed to treat=5) in the random-effects model. The Q statistic was not significant (Q=1.432, df=3, p=0.698), nor was the I² statistic (I²=0%). Visual inspection of both Begg’s funnel plot and Egger’s test (p=0.171) was not suggestive for the presence of publication bias.

The spectrum of severity provides a first helpful indication for selecting treatment options in depression. The magnitude of benefit from antidepressant drugs increases with the severity of depression (41, 42). Antidepressants are unlikely to be better than placebo in mild or minor depression (41, 43). As a result, psychotherapy alone (i.e., CBT, interpersonal psychotherapy and problem solving) appear to be justified for this milder end of the spectrum (36, 44). The potential disadvantages of antidepressant drugs encompass side effects, particularly with long-term treatment (45), potential interactions with medical conditions and drugs (46), and problems upon discontinuation (47, 48). Such disadvantages appear to clearly outweigh the potential benefits in this patient population.

The joint use of psychotherapy and pharmacotherapy in depression was found to yield short-term advantages compared with antidepressant drug treatment alone (49) or psychotherapy alone (50). More limited benefits were reported in terms of relapse prevention (9–13). In a recent meta-analysis (51), the effectiveness of psychological interventions in preventing recurrences was enhanced when a sequential design was used.

Like antidepressants, psychotherapy also appears to entail side effects, as demonstrated in a recent review by Forand et al. (52), and requires additional costs that need to be justified. The joint use of pharmacotherapy and psychotherapy does not necessarily yield a superior effect: meta-analyses display a mixture of outcomes, and large additive effects tend to occur only in particular subgroups (9–13, 50–52). High severity of depression and chronicity appear to be two important clinical characteristics that predict positive response to combined treatment (44, 49, 50). Hollon et al. (53), however, in a large and accurately designed study, found advantages of the addition of cognitive therapy to pharmacotherapy over drug treatment alone only with regard to severity. As Thase (54) pointed out, the cognitive therapy model is conventionally time-limited and may not be suitable for conditions that require long-term treatments. The advantages of a psychotherapeutic strategy specifically developed for chronic depression—cognitive behavioral analysis system of psychotherapy—in yielding remission were demonstrated in two large trials (55, 56).

Simultaneous use of psychotherapy and pharmacotherapy was found to yield lower drop-out rates than monotherapy (49) and may suggest its use in patients who are at risk for poor medication adherence. Interpersonal psychotherapy added to pharmacotherapy has been found to be better than medication alone in maintenance treatment (57), and this can be an additional indication for combined therapy.

The results of this meta-analysis should be interpreted with caution. First, we found evidence of publication bias that might have led to an overestimation of treatment effects. However, because publication bias is only one of the possible reasons for funnel plot asymmetry, this could reflect the tendency for the smaller studies in the meta-analysis to show larger treatment effects. It is also known that research without statistically significant results could take longer to achieve publication than research with significant results, further biasing evidence over time (time lag bias) (58). Nonetheless, it seems very unlikely that negative or inconclusive results regarding the efficacy of psychotherapy compared with pharmacotherapy, treatment as usual, or other control conditions in the residual phase of major depressive disorder would remain unpublished. Furthermore, the research designs that have been used may have produced overly optimistic results. Most of the studies compared the sequential approach with treatment as usual, which varied across studies, and differences might have reflected nonspecific effects and expectations (59). However, highly significant differences were also detected against clinical management that consisted of the same number and duration of sessions as the treatment condition (23, 24). When active control comparison groups were used, significant differences occurred, even though to a lesser degree compared with other investigations (28, 30), and the study (26) that did not report them was characterized by a very short follow-up. Additional limitations are that the sample sizes, the duration of treatments, and the length of follow-up periods varied across the included studies. Nonetheless, meta-regression analyses did not show significant effects on relapse/recurrence rates. No controlled studies, however, have directly compared the use of psychotherapy during antidepressant continuation with tapering and discontinuation. Finally, use of medications in the studies that were included followed a naturalistic pattern, and results might have been different with more aggressive medication dosing aimed to achieving a truly asymptomatic state.

Despite these limitations, our findings lend support to the view that the sequential administration of psychotherapy after response to acute-phase pharmacotherapy, either alone or in combination with antidepressant drugs, may play a role in reducing relapse and recurrence in major depressive disorder.

This can be seen as support for the hypothesis that psychotherapy may generate skills that patients can continue to use after treatment ends to manage their own affective states, reducing internal and external risks for relapse or recurrence. The concept of improvement reflects the distance of the current state of the patient from the pretreatment position (60). Addition of psychotherapy for a patient who has partially responded to the first line of treatment might result in more favorable ground than in an acute situation, after the first treatment has mobilized the patient’s recovery to its very limit. Comparable learning may not take place with pharmacotherapy alone (61). Acute depression may impair learning and processing and retrieval of information (62). Cognitive restructuring and behavioral exposure may thus have more impact when acute depression has abated and may allow detection of under-dosing in drug treatment that the joint use of treatments in the initial phase may mask. In addition, the preventive effects of the sequential strategy appear to be related to abatement of residual symptoms and/or increase in psychological well-being and coping skills (63–66). Residual symptoms are among the most powerful predictors of relapse (65–67). Anxiety was found to have a prominent role in this symptomatology and in many patients is unlikely to be adequately addressed by one course of treatment, whether pharmacotherapy or psychotherapy (68). Psychotherapy may thus improve symptoms that the other type of treatment (pharmacotherapy) was unable to affect. This may be particularly important when treatments modulate cortical-limbic pathways in different ways (69–72).

The application of the psychotherapeutic intervention in the sequential model departs from the traditional treatment strategies in depression (Figure 4). First, it is applied to the residual phase of major depressive disorder according to a longitudinal view of development of disorders that can be subsumed under the staging model (67). Second, the target of psychotherapeutic work is thus no longer predetermined, but varies according to the nature, characteristics, and intensity of residual symptoms (7). Third, all the included studies involved cognitive-behavioral treatments, since we could not find any single randomized controlled trial of the sequential use of other well-established psychotherapies (such as interpersonal psychotherapy) after response to pharmacotherapy. However, modifications of cognitive-behavioral techniques, including mindfulness-based cognitive therapy (73), CBT of residual symptoms (7), and well-being therapy (74), were used in most of the studies.

A sequential strategy may include discontinuation of antidepressant drug treatment or its maintenance, thereby offering the advantage of yielding enduring effects while limiting exposure to drug therapy. Performing subgroup analyses, we found a significant effect favoring the combination of psychotherapy and antidepressant medication during the continuation phase compared with treatment as usual or antidepressant medication alone. Furthermore, the sequential use of psychotherapy after discontinuation of antidepressant medication was found to be significantly more effective in reducing relapse/recurrence compared with control conditions (i.e., clinical management or antidepressant drugs). The evidence also suggests that discontinuation of antidepressant drugs is feasible when psychotherapy is provided and may yield enduring results. Indeed, the number needed to treat showed a greater advantage of the sequential option with drug tapering and discontinuation in preventing relapse or recurrence compared with continuation of antidepressants. This is important in view of the fact that a substantial proportion of patients discontinue antidepressant therapy after responding to the initial acute phase treatment, regardless of the physician’s advice (40).

The sequential model does not appear to be suitable for major depressive disorder of mild severity or minor depression, since the initial treatment with antidepressant drugs may not be warranted. However, when severity is higher, there appear to be several indications. The first indication is provided by recurrent depression. Some individual studies were concerned specifically with recurrent depression (24, 29–35), or benefits were more pronounced in case of recurrent depression (25, 27). There is a tendency to protract drug treatment for long periods of time, with the assumption that it may be protective against relapse. Yet, the evidence becomes questionable in trials of at least 18 months in duration (42, 75). In a meta-analysis, Kaymaz et al. (76) found that antidepressants reduce the relapse risk in the maintenance phase. However, patients with multiple depressive episodes experienced significantly less benefit in relapse prevention during the antidepressant maintenance phase compared with those with a single episode. These findings suggest that drug treatment is less helpful in patients who need more protection.

A second indication is provided by the presence of residual symptoms despite response to pharmacotherapy, which appears to be a common finding (23, 26, 28, 65, 66). Since incomplete recovery from the first lifetime major depressive episode was found to predict a chronic course of illness during a 12-year prospective naturalistic follow-up (77), this approach appears to be particularly indicated whenever substantial residual symptoms are present.

A third important indication occurs when the physician or the patient or both intend to interrupt drug treatment. The findings of our meta-analysis indicate its feasibility and favorable long-term outcome when this event takes place with the close monitoring of the second phase of the sequential treatment, also in view of discontinuation problems that may arise (78).

A variation of the sequential model, where the psychotherapeutic approach starts only if and when a loss of clinical efficacy ensues in patients who were fully remitted and in maintenance drug treatment, has been suggested, but its efficacy rests only on two small pilot studies (79, 80).