Prevention of Recurrence After Recovery From a Major Depressive Episode With Antidepressant Medication Alone or in Combination With Cognitive Behavioral Therapy: Phase 2 of a 2-Phase Randomized Clinical Trial | Depressive Disorders | JAMA Psychiatry | JAMA Network
[Skip to Navigation]
Sign In
Figure 1.  CONSORT Diagram
CONSORT Diagram
Figure 2.  Estimated Recurrence Following Phase 1 Conditions: ADM Monotherapy or Combination Therapy
Estimated Recurrence Following Phase 1 Conditions: ADM Monotherapy or Combination Therapy

The x-axis represents time, in days, from random assignment to be maintained on or withdrawn from medications. The y-axis represents the estimated recurrence rates across the 3-year maintenance period. ADM, antidepressant medication.

Figure 3.  Estimated Sustained Recovery as a Function of Phase 1 and Phase 2 Condition
Estimated Sustained Recovery as a Function of Phase 1 and Phase 2 Condition

The x-axis represents time, in days, from random assignment to be maintained on or withdrawn from medications. The y-axis represents the estimated sustained recovery rates across the 3-year maintenance period. The beginning points for the CT and ADM conditions from phase 1 reflect the percentage of patients with these conditions who met recovery criteria during phase 1. Sustained recovery was estimated as a function of phase 1 and phase 2 treatment conditions. ADM + CT, combination antidepressant medication and cognitive behavior therapy during phase 1; ADM Only, antidepressant medication alone during phase 1.

Table 1.  Baseline Characteristics of Phase 2 Participants
Baseline Characteristics of Phase 2 Participants
Table 2.  Recurrence, Survival, and Sustained Recovery Rates in Phase 2 Treatment Conditionsa
Recurrence, Survival, and Sustained Recovery Rates in Phase 2 Treatment Conditionsa
Limit 200 characters
Limit 25 characters
Conflicts of Interest Disclosure

Identify all potential conflicts of interest that might be relevant to your comment.

Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.

Err on the side of full disclosure.

If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.

Not all submitted comments are published. Please see our commenting policy for details.

Limit 140 characters
Limit 3600 characters or approximately 600 words
    1 Comment for this article
    EXPAND ALL
    Confounding recurrent MDD episode with ADM withdrawal symptoms
    Justine Lalonde, MD | Private practice
    The tapering schedule in phase 2 for the 3 ADMs +/- lithium was 4 weeks or longer. I could not find the average discontinuation time and % beyond 4 weeks. The modified criteria of 3 weeks, instead of 2 weeks, of meeting LIFE 5 or 6 level for a recurrence of MDD criteria for the first 8 weeks to mitigate withdrawal symptoms for participants who had been on ADM for a min. 9months -2 years appears inadequate.

    I submit that the short tapering schedule and the modified recurrence criteria are surely capturing withdrawal experiences which go beyond physical symptoms,
    and include symptoms of depression, anxiety, insomnia and neurocognitive difficulties. It is also interesting that the KM curves show a differential rate in the first 6 months of the phase 2 but then appear to have similarly recurrence rates thereafter. Such an analysis of curves could be of interest.

    I am convinced that confounding recurrence of MDD episodes instead of ADM withdrawal symptoms is likely, and certainly cannot be ruled-out, and should be stressed in the discussion and conclusion, especially in light of CBT vs ADM trials in the past have shown an advantage for CBT inrecurrence which is duly highlighted in the paper.
    CONFLICT OF INTEREST: None Reported
    READ MORE
    Original Investigation
    December 4, 2019

    Prevention of Recurrence After Recovery From a Major Depressive Episode With Antidepressant Medication Alone or in Combination With Cognitive Behavioral Therapy: Phase 2 of a 2-Phase Randomized Clinical Trial

    Author Affiliations
    • 1Department of Psychology, University of Pennsylvania, Philadelphia
    • 2Department of Psychiatry, Rush University, Chicago, Illinois
    • 3Department of Psychiatry, Vanderbilt University, Nashville, Tennessee
    • 4Department of Psychiatry, University of Alabama at Birmingham, Birmingham
    • 5Department of Psychiatry, University of Pennsylvania, Philadelphia
    • 6Department of Psychiatry, University of New Mexico, Albuquerque
    • 7Department of Mathematics and Applied Statistics, West Chester University, West Chester, Pennsylvania
    • 8Department of Psychology, Vanderbilt University, Nashville, Tennessee
    JAMA Psychiatry. 2020;77(3):237-245. doi:10.1001/jamapsychiatry.2019.3900
    Key Points

    Question  What are the effects of combining cognitive behavioral therapy with antidepressant medications on the prevention of depressive recurrence when antidepressant medications are withdrawn or maintained after recovery in patients with major depressive disorder?

    Findings  In this phase 2 randomized clinical trial of 292 adult patients with major depressive disorder who recovered from a chronic or recurrent major depressive episode, withdrawal of antidepressant medication treatment was associated with higher rates of recurrence compared with maintenance of antidepressant medication treatment regardless of whether patients achieved recovery with or without acute cognitive behavioral therapy treatment.

    Meaning  Maintenance of antidepressant medication treatment was associated with a reduced risk of depressive recurrence, but previous treatment with cognitive behavioral therapy was not; whether cognitive behavioral therapy has a similar protective effect or whether adding antidepressant medications to cognitive behavioral therapy treatment interferes with any such protective effect remains unclear.

    Abstract

    Importance  Antidepressant medication (ADM) maintenance treatment is associated with the prevention of depressive recurrence in patients with major depressive disorder (MDD), but whether cognitive behavioral therapy (CBT) treatment is associated with recurrence prevention remains unclear.

    Objective  To determine the effects of combining CBT with ADM on the prevention of depressive recurrence when ADMs are withdrawn or maintained after recovery in patients with MDD.

    Design, Setting, and Participants  A total of 292 adult outpatients with chronic or recurrent MDD who participated in the second phase of a 2-phase trial. Participants had recovered in the first phase of the trial receiving ADM, either alone or in combination with CBT. The trial was conducted in research clinics in 3 university medical centers in the United States. Patients in phase 2 were randomized to receive maintenance of or withdrawal from ADM and were followed up for 3 years. The first and last patients entered phase 2 in August 2003 and October 2009, respectively. The last patient completed phase 2 in August 2012. Data were analyzed from December 2013 to December 2018.

    Interventions  Maintenance of or withdrawal from treatment with ADM.

    Main Outcomes and Measures  Recurrence of an MDD episode using longitudinal interval follow-up evaluations; sustained recovery across both phases.

    Results  A total of 292 participants (171 women, 121 men; mean [SD] age 45.1 [12.9] years) were included in analyses of depressive recurrence. Maintenance ADM yielded lower rates of recurrence compared with ADM withdrawal regardless of whether patients had achieved recovery in phase 1 with ADM alone (48.5% vs 74.8%; z = −3.16; P = .002; number needed to treat [NNT], 2.8; 95% CI, 1.8-7.0) or ADM plus CBT (48.5% vs 76.7%; z = −3.49; P < .001; NNT, 2.7; 95% CI, 1.9-5.9). Sustained recovery rates differed as a function of phase 2 condition, with maintenance ADM superior to ADM withdrawal (z = 2.90; P = .004; OR, 2.54; 95% CI, 1.37-4.84; NNT, 2.3; 95% CI, 1.5-6.4). Phase 1 condition was not associated with differential rates of sustained recovery (ADM alone vs ADM plus CBT; z = 0.22; P = .83; OR, 1.08; 95% CI, 0.52-2.11; NNT, 26.0; 95% CI, number needed to harm 3.2 to NNT 2.8), nor was there a significant interaction of phase 1 condition and phase 2 condition (z = 0.30; P = .77; OR, 1.14; 95% CI, 0.49-2.88).

    Conclusions and Relevance  Maintenance ADM treatment, but not previous exposure to CBT, was associated with reduced rates of depressive recurrence. In previous studies, when CBT has been provided without ADM, CBT has shown a preventive effect on depressive relapse. Whether CBT also has a preventive effect on depressive recurrence, or if adding ADM interferes with any such preventive effect, remains unclear.

    Trial Registration  ClinicalTrial.gov identifier: NCT00057577

    Introduction

    Research has indicated that antidepressant medications and cognitive behavioral therapy (CBT) have similar efficacies on acute outcomes (response or remission) in the treatment of patients with major depressive disorder (MDD),1 and the combination of treatment with antidepressant medications and CBT is associated with better acute outcomes than either treatment modality alone.2-4 A substantial proportion of patients with MDD experience a chronic or recurring course; thus, a crucial feature of a treatment modality or a combination of modalities is the ability to protect patients from relapse or recurrence. Acute treatment with antidepressant medications has not been associated with the prevention of recurrence or relapse, but continuing or maintaining antidepressant medication regimens for patients has been associated with forestalling symptom return.5,6 Cognitive behavioral therapy, when delivered alone, appears to be associated with relapse prevention7 (ie, the return of symptoms associated with a treated episode), but less is known about the association between CBT treatment and recurrence prevention (ie, the onset of wholly new episodes), whether CBT treatment is provided alone or as part of combination therapy.

    Quiz Ref IDThis article describes the long-term outcomes of a 2-phase study of treatments for patients with MDD. In phase 1, patients were randomized to treatment with antidepressant medication monotherapy or a combination therapy of antidepressant medication plus CBT, and treatment with the combination therapy was associated with a higher rate of recovery.8 Phase 2 comprised a 3-year follow-up of patients who had achieved recovery in phase 1. Patients who achieved recovery with monotherapy treatment were randomized to receive either maintenance or withdrawal of treatment with antidepressant medications, as were patients who achieved recovery with combination therapy treatment. Our primary analyses compared rates of recurrence as a function of phase 2 assignment (medication maintained vs medication withdrawn) in 2 patient samples, one in which recovery was achieved with monotherapy treatment (monotherapy maintained vs monotherapy withdrawn) and the other in which recovery was achieved with combination therapy treatment (combination therapy maintained vs combination therapy withdrawn).

    The recurrence rates in the combination of phase 1 and phase 2 treatment conditions could be compared with each other, but valid causal inferences regarding the prevention of recurrence achieved by combination therapy treatment compared with monotherapy treatment could not be drawn from these comparisons. This study was unable to establish causal inferences because the populations in the monotherapy and combination therapy groups differed, as was evident from the differential recovery rates observed in phase 1.8 However, a sustained recovery system of measurement,9,10 which combines outcomes across the 2 phases, could be applied and compared across the 4 combinations of treatment conditions (monotherapy maintained, monotherapy withdrawn, combination therapy maintained, and combination therapy withdrawn). Such a measure reflects the ability of a treatment regimen to get patients well and keep them well.

    Methods
    Patients

    For the primary analyses of phase 2 data, the sample comprised a subset of patients diagnosed with either chronic (episode duration ≥2 years) or recurrent (an episode in the past 3 years even if only a second episode) MDD using the DSM-IV; these patients were previously randomized in phase 1 of the clinical trial, which compared patients who received monotherapy treatment (n = 225) with patients who received combination therapy treatment (n = 227).8 Other inclusion and exclusion criteria were described in the phase 1 report.8 The characteristics of patients who participated in phase 2 are reported in Table 1. Data were analyzed from December 2013 to December 2018.

    The study was conducted at outpatient research clinics at the University of Pennsylvania in Philadelphia, Rush Medical Center in Chicago, Illinois, and Vanderbilt University in Nashville, Tennessee. The clinical trial protocol is available in Supplement 1. The institutional review boards at the respective institutions approved the clinical trial protocol, and the study implementation was monitored by an independent data safety monitoring board. Written informed consent was received before any research activity began and was again obtained at the second randomization. This study followed the Consolidated Standards of Reporting Trials (CONSORT) reporting guideline.

    Procedures

    Quiz Ref IDFigure 1 depicts the study design and patient flow. The procedures used in phase 1 of the clinical trial were described in a previous article.8 A total of 318 of 452 participants (70%) from phase 1 were eligible to participate in phase 2 because they met the criteria for recovery before the maximum allowable time (3.5 years). Of those, 292 patients provided consent to be randomized to either maintain treatment with antidepressant medications or to be slowly withdrawn from medications. Of the 170 patients who met recovery criteria in the combination therapy group, 155 (91%) provided consent to participate in phase 2; 85 patients were assigned to the maintenance group (combination therapy maintained) and 70 to the withdrawal group (combination therapy withdrawn). Of the 148 patients who met recovery criteria in the monotherapy group, 137 (93%) consented to participate in phase 2; 68 patients were assigned to the maintenance group (monotherapy maintained) and 69 to the withdrawal group (monotherapy withdrawn). Thus, 160 patients (72 in combination therapy and 88 in monotherapy) who were assigned to phase 1 treatment conditions did not receive assignments to phase 2 treatment conditions because they either did not achieve recovery or achieved recovery but did not consent to phase 2 assignment.

    Random assignment to treatment was implemented using an adaptive (urn) randomization procedure, which can accommodate a larger number of factors than stratified randomization.11 The factors included in the adaptive randomization algorithm for phase 2 assignment were current medication (serotonin norepinephrine reuptake inhibitor vs selective serotonin reuptake inhibitor), sex (male vs female), number of episodes (≤2 vs ≥3), personality disorder (absent vs present), and recovery status (complete vs partial). After the last patient was randomized, the balance of each structure was assessed across intervention assignment, indicating balance was achieved for each. The project coordinator at each site was able to access a patient’s assignment only after the patient met recovery criteria and provided informed consent. Intake for phase 1 occurred from September 13, 2002, through February 22, 2006. The first patient entered phase 2 on August 13, 2003, and the last patient entered phase 2 on October 19, 2009. The last patient completed phase 2 on August 19, 2012. Data were analyzed from December 2013 to December 2018.

    Patients who had received combination therapy treatment during phase 1 ended their course of CBT treatment when phase 2 began. The mean (SD) number of CBT sessions received by patients in the combination therapy group during phase 1 was 33.3 (22.8). The mean (SD) length of time patients received combination therapy or monotherapy treatment before reassignment to phase 2 was 80.3 (40.0) weeks. Patients assigned to receive medication maintenance (monotherapy maintained or combination therapy maintained) met with their pharmacotherapist every 12 weeks; adjustment or augmentation of the medication regimen was permitted. A detailed account of the medications used will be reported separately. Patients assigned to medication withdrawal (monotherapy withdrawn or combination therapy withdrawn) were tapered from their regimen over a 4-week period, or longer if clinically indicated. Because medication withdrawal was not accompanied by the use of placebo, phase 1 assignments were not blinded for patients or pharmacotherapists.

    Outcome Assessment

    Interviewers who were blinded to the patients’ treatment conditions assessed patient status using the Longitudinal Interval Follow-up Evaluation (LIFE) every 4 weeks for the first 12 weeks of phase 2 and every 12 weeks thereafter. The LIFE tool provides weekly retrospective assessments of each patient’s psychiatric status rating on a scale of 1 to 6, with scores of 5 or 6 indicating the patient met the DSM-IV symptom criteria for MDD that week. Recurrence was defined as LIFE ratings of 5 or 6 for 2 consecutive weeks at any time after the first 8 weeks of phase 2. To ensure we did not misconstrue the symptoms of medication discontinuation as a recurrence, 3 consecutive weeks with a LIFE rating of 5 or 6 were required for a classification of recurrence during the first 8 weeks. The LIFE interviews were conducted until a patient withdrew from the study, experienced a documented recurrence, or completed 3 years without recurrence. Serious adverse events were reported to the respective institutional review boards and the data safety monitoring board as they occurred.

    Statistical Analyses
    Recurrence

    The fact that only patients who met recovery criteria in phase 1 were eligible to participate in phase 2 created the possibility that the proportions of patients entering phase 2 would differ between the combination therapy and monotherapy treatment conditions, which they did (75.2% of the combination therapy group vs 65.6% of the monotherapy group entered phase 2).8 Even if the proportions had not differed, the phase 1 treatment conditions could have acted as a differential sieve that reflected differential risk if the 2 samples entering phase 2 differed on characteristics that are associated with recurrence risk.12

    Therefore, analyses of recurrence rate as a function of phase 2 assignment (maintained vs withdrawn) were conducted separately for the monotherapy maintained vs monotherapy withdrawn groups and for the combination therapy maintained vs combination therapy withdrawn groups. Cox proportional hazards regression models were used to test for differences in recurrence rates and to provide estimates of the likelihood of recurrence over the 3-year follow-up. The main effects of the phase 2 treatment condition, site, and condition-by-site interaction were modeled, with the interaction term retained in the final models only when significant. Estimates of recurrence were calculated using the averages-of-survival-curves approach.13 To characterize the clinical significance of the findings, we computed the number needed to treat. All statistical analyses were performed with R software, version 3.3.3 (GNU Project) using the survival package.14

    Sustained Recovery

    Because both recovery and recurrence are important for long-term outcomes, we calculated estimates of sustained recovery for the monotherapy maintained, monotherapy withdrawn, combination therapy maintained, and combination therapy withdrawn groups by prorating the estimates of nonrecurrence by the respective rates of recovery in phase 1. A patient was considered to have experienced sustained recovery if he or she recovered in phase 1 and completed phase 2 (3 years of follow-up) without a recurrence.

    To account for all 452 patients initially assigned to treatment in phase 1 and to include them in the sustained recovery analysis, we needed to include the 160 patients who did not receive a phase 2 assignment. We estimated sustained recovery rates in each of the 4 treatment condition combinations as if these patients, within each phase 1 treatment condition, were assigned equally to the phase 2 treatment conditions. Because none of these patients provided data during phase 2, the assignment of specific patients to the phase 2 treatment conditions was not a factor in the sustained recovery estimates. Patients who did not recover were represented only in the respective denominators, whereas those who recovered but were not assigned to a phase 2 treatment condition were prorated with respect to sustained recovery estimates.

    Estimates of the number of patients who achieved sustained recovery in each of the phase 1 and phase 2 treatment combinations were calculated for the initially randomized 452 patients by rounding the product of the sustained recovery rate and the total overall assignment to each treatment condition (actual assignments plus imputed assignments). The use of these values, in which sustained recovery for each observation was either a 0 or a 1, allowed for estimates of the main effects of phase 1 assignment, phase 2 assignment, and their interaction using logistic regression analysis.

    To detect an advantage of 20% in sustained recovery in the combination therapy vs monotherapy group comparison (relative to a 25% rate), we had approximately 0.99 power using a 2-tailed analysis with a significance of α = .05. The smallest advantage that could be detected (relative to a 25% rate) with a power of 0.80 was 12.2%. For the comparison of greatest interest (combination therapy withdrawn vs monotherapy withdrawn), the power to detect a 20% advantage (relative to a 15% rate) was 0.94. The smallest advantage that could be detected (relative to a 15% rate) with a power of 0.80 was 15.5%.

    In phase 1 of the clinical trial, the higher recovery rate observed in the combination therapy compared with the monotherapy group was moderated by chronicity; patients with nonchronic MDD had a higher rate of recovery with combination therapy treatment vs monotherapy treatment (80.1% vs 61.9%, respectively), but this higher rate was not observed in patients with chronic MDD (72.1% vs 66.3%, respectively). Further, the advantage of combination therapy compared with monotherapy treatment within the nonchronic group was moderated by severity such that the patients with severe nonchronic MDD experienced an even greater benefit from receiving combination therapy compared with monotherapy treatment. We therefore conducted exploratory analyses to test whether severity and chronicity also moderated sustained recovery.

    Results
    Recurrence

    Quiz Ref IDThe outcomes of 292 adult outpatients (171 women and 121 men; mean [SD] age, 45.1 [12.9] years) were included in the analyses of depressive recurrence. Among patients who had achieved recovery in the monotherapy group (n = 137), recurrence rates were lower in those randomized to the monotherapy maintained compared with the monotherapy withdrawn group (48.5% vs 74.8%, respectively; z = −3.16; P = .002; hazard ratio [HR], 0.47; 95% CI, 0.29-0.75; number needed to treat [NNT], 2.8; 95% CI, 1.8-7.0). Among patients who had achieved recovery in the combination therapy group (n = 155), recurrence rates were similar, with lower rates in the combination therapy maintained compared with the combination therapy withdrawn group (48.5% vs 76.7%, respectively; z = −3.49; P < .001; HR, 0.46; 95% CI, 0.30-0.71; NNT, 2.7; 95% CI, 1.9-5.9).

    Figure 2 shows the 2 pairs of survival curves corresponding to these analyses. Condition-by-site interactions were not observed in either model, so only the main effect of site was included as a covariate.

    Sustained Recovery

    Sustained recovery curves (Figure 3) were produced by prorating the values in the survival curves of the monotherapy maintained, monotherapy withdrawn, combination therapy maintained, and combination therapy withdrawn groups by recovery rates in the respective phase 1 treatment conditions (monotherapy or combination therapy). Note that the curves start at the level determined by the respective recovery rates at the end of phase 1. Estimates of sustained recovery at the end of phase 2 are shown in Table 2 along with the values used to calculate the estimates.

    Quiz Ref IDThe outcomes of 266 women and 186 men (mean [SD] age, 43.2 [13.1] years) were included in the analyses of sustained recovery, which considered the differential rates of recovery observed in phase 1. After inclusion of all patients assigned to phase 1 treatment conditions, the main effect of the phase 2 treatment condition (maintained vs withdrawn) on sustained recovery was significant (z = 2.90; P = .004; odds ratio [OR], 2.54; 95% CI, 1.37-4.84; NNT, 2.3; 95% CI, 1.5-6.4), which echoed the advantage of medication maintained compared with medication withdrawn in each of the analyses of recurrence rates. The main effect of the phase 1 treatment condition (monotherapy vs combination therapy) was not significant (z = 0.22; P = .83; OR, 1.08; 95% CI, 0.52-2.11; NNT, 26.0; 95% CI, number needed to harm 3.215 to NNT 2.8), nor was the interaction between the phase 1 and phase 2 treatment conditions (z = 0.30; P = .77; OR, 1.14; 95% CI, 0.49-2.88).

    In the nonchronic subgroup, sustained recovery differed as a function of the phase 2 treatment condition, with medication maintained indicating better results than medication withdrawn (z = 3.00; P = .003; OR, 3.49; 95% CI, 1.58-8.15; NNT, 1.8). Neither the phase 1 treatment condition (combination therapy vs monotherapy; z = 0.75; P = .45; OR, 1.40; 95% CI, 0.58-3.44; NNT, 6.0) nor the interaction of the phase 1 and phase 2 treatment conditions was significant (z = 0.27; P = .79; OR, 0.85; 95% CI, 0.29-2.62). In the severe nonchronic subgroup, sustained recovery differed as a function of the phase 2 treatment condition (z = 2.03; P = .04; OR, 3.18; 95% CI, 1.07-10.31; NNT, 1.9). Neither the effects of the phase 1 treatment condition (combination therapy vs monotherapy; z = 0.41; P = .68; OR, 1.30; 95% CI, 0.37-4.58; NNT, 7.7) nor the interaction of the phase 1 and phase 2 treatment conditions (z = 0.09; P = .91; OR, 1.09; 95% CI, 0.23-5.28) was significant.

    No statistically significant differences were observed in the number of serious adverse events across the 4 treatment conditions (17 events for monotherapy maintained, 20 for monotherapy withdrawn, 19 for combination therapy maintained, and 16 for combination therapy withdrawn; F3288 = 0.605; P = .61).

    Discussion

    Among patients with chronic or recurrent MDD who achieved recovery with antidepressant medication treatment or a combination of CBT and medication treatments, recurrence rates were substantially lower in patients who were maintained on medication treatment compared with those who were withdrawn from medication treatment. This finding is in line with a large body of previous evidence.5,6 Further, the difference was observed regardless of whether recovery was achieved with antidepressant monotherapy or combination therapy treatment. No evidence was found that the provision of CBT during acute/continuation treatment provided protection against subsequent recurrence. An analysis of sustained recovery rates, which integrated outcomes from both the acute/continuation and maintenance phases of the study, also did not reveal that a significant advantage was provided by the addition of CBT treatment in the acute/continuation phase. If anything, the initial advantage of the combination therapy treatment that was observed in phase 1 appeared to decrease during the phase 2 follow-up period. Even in the subgroup of patients with severe nonchronic MDD, who were observed to have benefitted from the addition of CBT treatment in phase 1, the added value of CBT treatment did not appear to be sustained across the 3-year follow-up period.

    The fact that we did not detect an association between CBT treatment and recurrence prevention in this study might appear to be at odds with data from 2 previous clinical trials in which patients who received previous CBT treatment experienced enduring recurrence prevention compared with recovered patients who were withdrawn from monotherapy treatment.9,16 Neither of these clinical trials, however, combined treatment with CBT and medication therapies during the acute or continuation phases. When CBT treatment has been delivered in the context of combination therapy, findings regarding the prevention of relapse or recurrence have been mixed. One clinical trial reported an enduring effect of previous CBT treatment,17 a second was inconclusive,18 and a third found no enduring effect.19 Although the failure to find an effect in this study does not preclude the possibility that an effect could be found in future clinical trials, the addition of the present findings does cast further doubt on the ability of acute CBT treatment to produce an enduring effect when delivered in the context of combination therapy.

    A similar pattern has been observed in the treatment of patients with anxiety such that the combination of certain anxiolytic medications with CBT during the acute treatment phase attenuates the long-term positive effects of CBT treatment.20,21 In a study by Barlow et al,22 patients with panic disorder who achieved remission with imipramine treatment were twice as likely to relapse after treatment termination than patients who achieved remission with CBT treatment; patients who achieved remission with a combination of CBT and imipramine treatments had the same elevated risk of relapse as those who received imipramine treatment alone, whereas patients who achieved remission with CBT treatment and placebo were no more likely to relapse after treatment termination than patients who achieved remission with CBT treatment alone. In effect, the receipt of antidepressant medication interfered with the enduring effect of CBT treatment, whereas simply believing that one was receiving a medication did not. In the current study, we did not include a CBT-alone or a CBT-plus-placebo treatment condition. Had we done so, we would have been able to address directly the question of whether adding monotherapy treatment interferes with any enduring effect that CBT treatment may have with respect to recurrence and, if so, whether this enduring effect is due to psychological or pharmacological reasons. In the Barlow et al22 clinical trial, the enduring effect was clearly for pharmacological reasons.

    Other researchers have found recurrence-prevention benefits of psychotherapy that is received after initial recovery with antidepressant monotherapy treatment.23-25 However, those studies differed from the present study in the following ways: (1) the psychotherapy provided was focused on the prevention of future depressive relapse or recurrence, (2) the psychotherapy was provided after initial confirmed recovery with acute antidepressant medication treatment, and (3) the patients were all discontinued from antidepressant medication treatment after recovery. Thus, with the exception of a normal tapering window, patients did not receive a combination therapy regimen of antidepressant medication and psychotherapy treatments over an extended period.

    Limitations

    Our overall study design lacked a CBT-alone treatment condition in phase 1, which limited the ability to compare our findings with studies in which patients receiving CBT treatment alone experienced a relapse or recurrence prevention effect. Our findings can, at most, be generalized to populations of patients with chronic or recurrent MDD and not to patients experiencing a first depressive episode, for whom the prevention of recurrence would be of great interest.

    The sample sizes for the primary comparisons of sustained recovery rates had sufficient power to detect differences between treatments of approximately 10% to 15%, but the study did not have sufficient power to detect smaller differences.

    Participants were randomized to their phase 2 treatment condition only after recovering in phase 1 and consenting to participate in phase 2. One reason we opted for a sequential randomization procedure was to avoid making an assignment that could be implemented as many as 180 weeks after the initial assignment. However, this procedure could have resulted in differences among the phase 2 samples that may have affected estimates of sustained recovery.26

    Conclusions

    Quiz Ref IDThe combination of psychotherapy and antidepressant medication treatments is an evidence-based recommendation for optimizing efficacy in the acute treatment of patients with MDD.2-4 Findings from phase 1 of the present clinical trial are in line with this recommendation, as treatment with a combination of CBT and antidepressant medications was associated with a higher rate of recovery than treatment with antidepressant medications alone, an advantage that was especially evident in the subset of patients with nonchronic but more severe MDD. However, the findings from the 3-year follow-up reported herein raise questions about whether adding antidepressant medication therapy during acute and continuation treatment interferes with the enduring effect of CBT treatment. Patients with chronic and recurrent MDD who achieved recovery with antidepressant medication treatment benefited from medication maintenance, although only an estimated one-third of patients achieved sustained recovery, and indications exist to suggest that combining monotherapy with CBT may interfere with any enduring effect CBT treatment may have.

    Back to top
    Article Information

    Accepted for Publication: September 12, 2019.

    Published Online: December 4, 2019. doi:10.1001/jamapsychiatry.2019.3900

    Correction: This article was corrected on January 29, 2020, to fix errors in the curves of Figure 2 and to revise the subtitle and Abstract, as well as titles and captions to Figures 2 and 3.

    Corresponding Author: Robert J. DeRubeis, PhD, Department of Psychology, University of Pennsylvania, 425 S University Ave, Philadelphia, PA 19104 (derubeis@psych.upenn.edu).

    Author Contributions: Drs DeRubeis and Hollon had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

    Concept and design: DeRubeis, Zajecka, Shelton, Amsterdam, Fawcett, Hollon.

    Acquisition, analysis, or interpretation of data: DeRubeis, Shelton, Amsterdam, Fawcett, Xu, Young, Gallop, Hollon.

    Drafting of the manuscript: DeRubeis, Zajecka, Xu, Gallop.

    Critical revision of the manuscript for important intellectual content: DeRubeis, Zajecka, Shelton, Amsterdam, Fawcett, Xu, Young, Hollon.

    Statistical analysis: DeRubeis, Xu, Gallop.

    Obtained funding: DeRubeis, Amsterdam, Fawcett, Hollon.

    Administrative, technical, or material support: Zajecka, Shelton, Amsterdam, Fawcett, Xu, Young, Hollon.

    Supervision: DeRubeis, Amsterdam, Fawcett, Young, Hollon.

    Conflict of Interest Disclosures: Dr DeRubeis reported receiving grants from the National Institute of Mental Health during the conduct of the study. Dr Zajecka reported receiving grants from the National Institute of Mental Health during the conduct of the study and grants from Actavis Generics, Alkermes, Allergan, AstraZeneca, Axsome Therapeutics, the Cheryl T. Herman Foundation, Cyberonics, ElMindA, Forest Pharmaceuticals, Hoffman-La Roche, Janssen Pharmaceuticals, Johnson & Johnson, Lundbeck, Naurex, Neuralstem, Novartis, Otsuka, Sage Therapeutics, Taisho, and Takeda outside the submitted work. Dr Shelton reported receiving grants from the National Institute of Mental Health during the conduct of the study and grants from Avanir Pharmaceuticals, Intra-cellular Therapies, Myriad Genetics, NeuroRx, Novartis, Otsuka, the Patient-Centered Outcomes Research Institute, Sage Therapeutics, and Genomind; grants and personal fees from Acadia Pharmaceuticals, Allergan, Cerecor, Janssen Pharmaceuticals, and Takeda; and personal fees from Lundbeck outside the submitted work. Dr Fawcett reported receiving grants from the National Institute of Mental Health during the conduct of the study. No other disclosures were reported.

    Funding/Support: This study was supported by grants MH60713 and MH01697 (K02) (Dr Hollon), MH60998 (Dr DeRubeis), and MH060768 (Drs Fawcett and Zajecka) from the National Institute of Mental Health. Wyeth Pharmaceuticals provided venlafaxine, and Pfizer provided sertraline for the clinical trial.

    Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

    Data Sharing Statement: See Supplement 2.

    Additional Contributions: Brent Freeman, BA, and Bernadette Kooi, MS, of the University of Pennsylvania; Debra Klbecka, RN, and Matthew Marasco, BA, of Rush University; and Margaret L. Lovett, MEd, of Vanderbilt University served as the study coordinators. Giampaolo Gallo, MD, Moira Molloy, MSN, Bobbie Posmontier, PhD, Nancy Rutherford, MSN, Irene Soeller, CRNP, Jeffrey Staab, MD, and Jay D. Amsterdam, MD, of the University of Pennsylvania; Jagannath Devulapally, MD, Corey Goldstein, MD, Ian Mackey, MD, William Miles, MD, Raj Tummale, MD, and John Zajecka, MD, of Rush University; and Virginia Gardner, MSN, Jennifer Scroggie, MSN, Sandra Seidel, MSN, and Vatsal Thakkar, MD, of Vanderbilt University served as study pharmacotherapists. Julie Jacobs, PhD, Cory P. Newman, PhD, and Rita Ryan, PhD, of the University of Pennsylvania; David C. Clark, PhD, Kristen Flynn, PhD, John Larson, MD, Patricia Meaden, PhD, Chad Owen, PsyD, and Paula R. Young, PhD, of Rush University; and Laurel L. Brown, PhD, Kirsten Haman, PhD, Karl N. Jannasch, PhD, Sandra Seidel, MSN, and Dorothy D. Tucker, PhD, of Vanderbilt University served as the cognitive therapists. Kirsten L. Haman, PhD, oversaw the training of the clinical interviewers. All of the aforementioned contributors were reimbursed for their participation. William T. McKinney, MD, of Northwestern University (emeritus); Irene Elkin, PhD, of the University of Chicago (emerita); Robert Gibbons, PhD, and Mark Siegler, MD, of the University of Chicago; and Burt Jensen of the Federal Bureau of Investigation (retired) served on the independent data safety monitoring board and received honoraria for their participation. DatStat developed the data entry and patient management system used to conduct the study. Paul W. Andrews, PhD, of McMaster University and Lois A. Gelfand, PhD, of the University of Pennsylvania provided helpful comments on an earlier draft of this article.

    Additional Information: Drs DeRubeis, Fawcett, and Hollon were the principal investigators; Drs Amsterdam, Zajecka, and Shelton were the coprincipal investigators. Drs DeRubeis, Hollon, and Young oversaw the implementation of computed tomography at the University of Pennsylvania, Vanderbilt University, and Rush University, respectively. Dr Fawcett oversaw the implementation of pharmacotherapy across the study; Drs Amsterdam, Zajecka, and Shelton supervised the implementation of pharmacotherapy at the respective sites.

    References
    1.
    Cuijpers  P, Berking  M, Andersson  G, Quigley  L, Kleiboer  A, Dobson  KS.  A meta-analysis of cognitive-behavioural therapy for adult depression, alone and in comparison with other treatments.  Can J Psychiatry. 2013;58(7):376-385. doi:10.1177/070674371305800702PubMedGoogle Scholar
    2.
    de Maat  SM, Dekker  J, Schoevers  RA, de Jonghe  F.  Relative efficacy of psychotherapy and combined therapy in the treatment of depression: a meta-analysis.  Eur Psychiatry. 2007;22(1):1-8. doi:10.1016/j.eurpsy.2006.10.008PubMedGoogle Scholar
    3.
    Cuijpers  P, van Straten  A, Warmerdam  L, Andersson  G.  Psychotherapy versus the combination of psychotherapy and pharmacotherapy in the treatment of depression: a meta-analysis.  Depress Anxiety. 2009;26(3):279-288. doi:10.1002/da.20519PubMedGoogle Scholar
    4.
    Cuijpers  P, Sijbrandij  M, Koole  SL, Andersson  G, Beekman  AT, Reynolds  CF  III.  Adding psychotherapy to antidepressant medication in depression and anxiety disorders: a meta-analysis.  World Psychiatry. 2014;13(1):56-67. doi:10.1002/wps.20089PubMedGoogle Scholar
    5.
    Glue  P, Donovan  MR, Kolluri  S, Emir  B.  Meta-analysis of relapse prevention antidepressant trials in depressive disorders.  Aust N Z J Psychiatry. 2010;44(8):697-705. doi:10.3109/00048671003705441PubMedGoogle Scholar
    6.
    Reid  S, Barbui  C.  Long term treatment of depression with selective serotonin reuptake inhibitors and newer antidepressants.  BMJ. 2010;340:c1468. doi:10.1136/bmj.c1468PubMedGoogle Scholar
    7.
    Cuijpers  P, Hollon  SD, van Straten  A, Bockting  C, Berking  M, Andersson  G.  Does cognitive behaviour therapy have an enduring effect that is superior to keeping patients on continuation pharmacotherapy? a meta-analysis.  BMJ Open. 2013;3(4):e002542. doi:10.1136/bmjopen-2012-002542PubMedGoogle Scholar
    8.
    Hollon  SD, DeRubeis  RJ, Fawcett  J,  et al.  Effect of cognitive therapy with antidepressant medications vs antidepressants alone on the rate of recovery in major depressive disorder: a randomized clinical trial  [retracted in JAMA Psychiatry. 2016;73(6):639-640].  JAMA Psychiatry. 2014;71(10):1157-1164. doi:10.1001/jamapsychiatry.2014.1054PubMedGoogle Scholar
    9.
    Hollon  SD, DeRubeis  RJ, Shelton  RC,  et al.  Prevention of relapse following cognitive therapy vs medications in moderate to severe depression.  Arch Gen Psychiatry. 2005;62(4):417-422. doi:10.1001/archpsyc.62.4.417PubMedGoogle Scholar
    10.
    Fournier  JC, DeRubeis  RJ, Shelton  RC, Gallop  R, Amsterdam  JD, Hollon  SD.  Antidepressant medications v. cognitive therapy in people with depression with or without personality disorder.  Br J Psychiatry. 2008;192(2):124-129. doi:10.1192/bjp.bp.107.037234PubMedGoogle Scholar
    11.
    Wei  LJ, Lachin  JM.  Properties of the urn randomization in clinical trials.  Control Clin Trials. 1988;9(4):345-364. doi:10.1016/0197-2456(88)90048-7PubMedGoogle Scholar
    12.
    Klein  DF.  Preventing hung juries about therapy studies.  J Consult Clin Psychol. 1996;64(1):81-87. doi:10.1037/0022-006X.64.1.81PubMedGoogle Scholar
    13.
    Therneau  TM, Crowson  CS, Atkinson  EJ. Adjusted survival curves. https://cran.r-project.org/web/packages/survival/vignettes/adjcurve.pdf. Published January 2015. Accessed May 7, 2019.
    14.
    Therneau  TM. Survival analysis in S version 2.38. https://CRAN.R-project.org/package=survival. Published 2015. Accessed May 7, 2019.
    15.
    Altman  DG.  Confidence intervals for the number needed to treat.  BMJ. 1998;317(7168):1309-1312. doi:10.1136/bmj.318.7200.1764cPubMedGoogle Scholar
    16.
    Dobson  KS, Hollon  SD, Dimidjian  S,  et al.  Randomized trial of behavioral activation, cognitive therapy, and antidepressant medication in the prevention of relapse and recurrence in major depression.  J Consult Clin Psychol. 2008;76(3):468-477. doi:10.1037/0022-006X.76.3.468PubMedGoogle Scholar
    17.
    Evans  MD, Hollon  SD, DeRubeis  RJ,  et al.  Differential relapse following cognitive therapy and pharmacotherapy for depression.  Arch Gen Psychiatry. 1992;49(10):802-808. doi:10.1001/archpsyc.1992.01820100046009PubMedGoogle Scholar
    18.
    Blackburn  IM, Eunson  KM, Bishop  S.  A two-year naturalistic follow-up of depressed patients treated with cognitive therapy, pharmacotherapy and a combination of both.  J Affect Disord. 1986;10(1):67-75. doi:10.1016/0165-0327(86)90050-9PubMedGoogle Scholar
    19.
    Simons  AD, Murphy  GE, Levine  JL, Wetzel  RD.  Cognitive therapy and pharmacotherapy for depression: sustained improvement over one year.  Arch Gen Psychiatry. 1986;43(1):43-48. doi:10.1001/archpsyc.1986.01800010045006PubMedGoogle Scholar
    20.
    Otto  MW, McHugh  RK, Kantak  KM.  Combined pharmacotherapy and cognitive-behavioral therapy for anxiety disorders: medication effects, glucocorticoids, and attenuated treatment outcomes.  Clin Psychol (New York). 2010;17(2):91-103. doi:10.1111/j.1468-2850.2010.01198.xPubMedGoogle Scholar
    21.
    Rosen  CS, Greenbaum  MA, Schnurr  PP, Holmes  TH, Brennan  PL, Friedman  MJ.  Do benzodiazepines reduce the effectiveness of exposure therapy for posttraumatic stress disorder?  J Clin Psychiatry. 2013;74(12):1241-1248. doi:10.4088/JCP.13m08592PubMedGoogle Scholar
    22.
    Barlow  DH, Gorman  JM, Shear  MK, Woods  SW.  Cognitive-behavioral therapy, imipramine, or their combination for panic disorder: a randomized controlled trial.  JAMA. 2000;283(19):2529-2536. doi:10.1001/jama.283.19.2529PubMedGoogle Scholar
    23.
    Fava  GA, Ruini  C, Rafanelli  C, Finos  L, Conti  S, Grandi  S.  Six-year outcome of cognitive behavior therapy for prevention of recurrent depression.  Am J Psychiatry. 2004;161(10):1872-1876. doi:10.1176/ajp.161.10.1872PubMedGoogle Scholar
    24.
    Segal  ZV, Bieling  P, Young  T,  et al.  Antidepressant monotherapy vs sequential pharmacotherapy and mindfulness-based cognitive therapy, or placebo, for relapse prophylaxis in recurrent depression.  Arch Gen Psychiatry. 2010;67(12):1256-1264. doi:10.1001/archgenpsychiatry.2010.168PubMedGoogle Scholar
    25.
    Teasdale  JD, Segal  ZV, Williams  JM, Ridgeway  VA, Soulsby  JM, Lau  MA.  Prevention of relapse/recurrence in major depression by mindfulness-based cognitive therapy.  J Consult Clin Psychol. 2000;68(4):615-623. doi:10.1037/0022-006X.68.4.615PubMedGoogle Scholar
    26.
    March  J, Kraemer  HC, Trivedi  M,  et al.  What have we learned about trial design from NIMH-funded pragmatic trials?  Neuropsychopharmacology. 2010;35(13):2491-2501. doi:10.1038/npp.2010.115PubMedGoogle Scholar
    ×