A DILEMMA‐FOCUSED INTERVENTION FOR DEPRESSION: A MULTICENTER, RANDOMIZED CONTROLLED TRIAL WITH A 3‐MONTH FOLLOW‐UP

Abstract

Background

Since long ago it has been asserted that internal conflicts are relevant to the understanding and treatment of mental disorders, but little research has been conducted to support the claim. The aim of this study was to test the differential efficacy of group cognitive behavioral therapy (CBT) plus an intervention focused on the dilemma(s) detected for each patient versus group individual CBT plus individual CBT for treating depression. A comparative controlled trial with a 3‐month follow‐up was conducted.

Methods

One hundred twenty‐eight adults meeting criteria for MDD and/or dysthymia, presenting at least one cognitive conflict (implicative dilemma or dilemmatic construct, assessed by the repertory grid technique) and who had completed seven sessions of group CBT were randomly assigned to eight sessions of individual manualized CBT or dilemma‐focused therapy (DFT). The Beck Depression Inventory‐II was administered at baseline, at the end of therapy and after 3 months’ follow‐up.

Results

Multilevel mixed effects modeling yielded no significant differences between CBT and DFT with the intention‐to‐treat sample. Equivalent effect sizes, remission, and response rates were found with completers as well. In combination with group CBT, both individual CBT and DFT significantly reduced depressive symptoms.

Conclusions

Both conditions obtained comparable results to those in the literature. Thus, the superiority of the adjunctive DFT was not demonstrated. Working with dilemmas can be seen as a promising additional target in the psychotherapy of depression, but further research is still required.

INTRODUCTION

Depression is a recurrent and debilitating mental health disorder. Projections for 2030 consider that depression will be a major contributor to the burden of disease and a leading cause of disability worldwide.1 In addition to pharmacotherapy, psychological interventions are considered effective in the treatment of depressive disorders;2, 3 in particular, cognitive behavioral therapy (CBT) has accumulated a significant amount of favorable research evidence.4, 5 Despite the proven efficacy of CBT, however, difficulties such as resistance, relapse, and recurrence may occur in the therapy process and limit it as a fully satisfactory treatment for depression. It has also been acknowledged that the cognitive model of depression needs further elaboration, in part to account for these clinical difficulties. Expanding conceptualization might provide new cognitive treatment targets and techniques which, in turn, might increase treatment efficacy.

The awareness that people with depression may have cognitive conflicts that hinder their change process may help to enrich the conceptualization and practice of CBT. Cognitive or internal conflicts have been proposed in several theories for the understanding both of symptoms and of patients’ resistance to change. Personal construct theory (PCT),6 a precursor of the cognitive approach, provides a conceptual and methodological framework for the study of this notion. PCT conceptualizes conflicts as implicative dilemmas in which the desired change (e.g., becoming happy) implies an undesired change in another aspect of the self (e.g., being selfish). Methodologically, PCT uses the repertory grid technique (RGT) that allows an assessment of the self‐concept and its cognitive structure using the patient's own views (personal constructs); therefore, it is well suited for detecting conflicts involving the current and the ideal self.

The operationalization of implicative dilemma by means of the RGT was proposed by Feixas and Saúl7 and includes the development of a specialized software to analyze repertory grid data and to detect implicative dilemmas.8 To establish the presence of an implicative dilemma, in the first step discrepant constructs (those in which the desired self is different from the current self) and congruent constructs (in which the desired self is similar to the ideal self) are detected. Then, the correlations between discrepant and congruent constructs are analyzed in pairs; an implicative dilemma is identified whenever the desirable pole of the discrepant construct correlates (r ≥ .35) with the undesired pole of the congruent construct. Therefore, a desirable change in one construct implies an undesirable change in the other construct. Another form of cognitive conflict reflected in the RGT is dilemmatic construct, in which the “ideal self” is rated at the midpoint. This rating could imply that both poles of the construct comprise undesirable characteristics for the person.

Using the definition of cognitive conflicts as implicative dilemmas, several studies have found this notion clinically relevant to the understanding of psychological distress in a variety of different disorders.9 Specifically, some authors indicate that the presence of implicative dilemmas was greater in people with depression than in control groups.10, 11 Moreover, depressed patients with implicative dilemmas presented lower levels of global functioning and more frequent history of suicide attempts.10 The relevance of these dilemmas for the understanding of depressive symptoms and the assumption that they may hamper the recovery process led us to design a specific intervention to resolve the conflicts evidenced by implicative dilemmas. As a result, we created a manual of dilemma‐focused therapy (DFT)12, 13, 14 for use as an adjunct of a broader psychotherapy approach. The presence of implicative dilemmas was not necessarily considered as the cause of depression or as the only factor to be targeted in the therapy process, but it was regarded as a cognitive process that impedes change and recovery.

OBJECTIVES AND HYPOTHESES

The aim of the current study was to compare the efficacy of a combined therapy (group CBT plus individual DFT) with CBT (group plus individual) in the treatment of depression (MDD and dysthymia). To do so, a comparative controlled trial with a 3‐month follow‐up was conducted (a 1‐year follow‐up is in progress). Two specific objectives were considered: (1) to determine if there was a significant reduction of symptoms in both therapies, and (2) to compare the difference between the two treatment conditions. We hypothesized that (1) there would be a significant reduction of symptoms over time in both types of treatment, but that (2) the treatment condition including a specific module focused on the dilemmas identified for each patient would prove more effective, thus increasing the efficacy of CBT for depression.

METHODS

This trial was conducted from November 2011 to December 2014 in Barcelona, Spain, at several primary and mental health centers (CSMA Nou Barris Nord, Associació Catalana de Teràpies Cognitives, Hospital de Mataró, and others belonging to Parc de Salut Mar and Fundació Sant Pere Claver). In total, 22 groups were formed with between four and nine patients in each group. Patients were treated at their usual health centers in order to emphasize the naturalistic condition of the study and to maximize external validity.

PARTICIPANTS

Patients

Patients were recruited by advertisements and by referrals from the above‐mentioned centers. Eligibility criteria were (1) age range between 18 and 70 years old; (2) score of 20 or above on the BDI‐II; (3) meeting diagnostic criteria for major depressive disorder or dysthymia according to the Diagnostic and Statistical Manual of Mental Disorders, 4th edition, text revision (DSM‐IV‐TR),15 and having at least one cognitive conflict detected by the application of the RGT. Exclusion criteria were (1) psychotic symptoms, manic or hypomanic episodes in the past, substance abuse, organic brain dysfunction, acute suicidal ideation, and mental retardation; (2) already receiving psychological treatment; (3) substantial visual, hearing, and cognitive deficits; and (4) insufficient linguistic competence to communicate in Spanish or Catalan.

THERAPISTS

All therapists (N = 44) were recent graduates in psychology with some initial training and experience obtained in master postgraduate courses. The CBT group part of the treatment was conducted by two therapists who also conducted the individual part, in which patients were randomly assigned to receive either CBT or DFT. Therapists were selected for each treatment on the basis of their previous experience in CBT or DFT, and some additional training in the use of each manual was provided for each condition by the senior members of the team (A.B. and E.G. for CBT, and G.F. and V.C. for DFT). Twenty‐two therapists (females = 17) conducted CBT and 22 therapists (females = 12) conducted DFT. All of them were closely supervised before and after every session by senior therapists with more than 5 years’ clinical experience. To this aim, two small supervision groups (two therapists and the supervisor), one for each orientation, held 1‐hr weekly meetings for each therapy group. Therapists did not participate in the initial evaluation of the patients or in any subsequent assessments.

TREATMENTS

After initial assessment, treatment consisted in 16 psychotherapy sessions delivered in two stages. The first stage involved seven 2‐hr sessions of CBT group therapy and the second stage entailed eight sessions of individual therapy in CBT or DFT format according to randomized allocation. After the individual phase, one final group session served to close the treatment.

GROUP CBT

For this part of the treatment, a specific manual16 was created for the study based on Beck's cognitive therapy manual17 and other publications explaining CBT procedures in detail.18 The manual was highly specific with regard to the tasks for each of the seven sessions and ways to handle potential difficulties including the management of suicide risk.

Individual CBT

Patients allocated to this condition received CBT following the guidelines described in a complementary manual19 also created for this trial. It was borne in mind that patients had already participated in the group phase, and issues related to relapse prevention were left mainly for the final group session.

Individual DFT

Individual DFT was applied following a specific manual designed for the study.20, 21 Broadly speaking, this type of therapy begins with the analysis of the demand and of how it is related to the cognitive conflicts identified by the RGT. So, the objective was to reframe the problematic situation or symptoms in terms of one or more dilemmas. Once patient and therapist agreed on a dilemma to serve as the focus for the therapy, its implications were explored across a range of interpersonal situations and the patient's past history. The final sessions centered on the resolution of the dilemma and the formulation of future prospects of a life without it. DFT techniques are based on PCT22 and also include an adaptation of the two‐chair dialogue. Interestingly, the dilemmas found with the RGT were seldom presented by patients as an issue for which they are seeking help (e.g., a dilemma about whether to look for a different job or not). Rather, implicative dilemmas appeared in the construal of the self and of others and were quite different from practical dilemmas such as the one just mentioned. Consequently, the procedures included in the DFT manual had little or no resemblance to the practical problem‐solving techniques commonly used in CBT.

TREATMENT ADHERENCE

A 36‐item scale (with 18 items for each modality) was created to assess treatment adherence in the individual phase of the study. Two graduate students blinded to the treatment conditions and trained to use the scale reliably rated audiotapes of 10 sessions of DFT (k = .64) and 10 of CBT (k = .56).

RANDOMIZATION

Permuted block randomization was applied through the use of an online application.23 It was carried out by a staff member at the Department of Personality, Assessment and Treatment of the University of Barcelona, completely oblivious to the study and blind to treatment conditions. Randomization was carried out after the completion of the CBT group phase, and so therapists had no knowledge of patient allocation until they called them for individual therapy.

INSTRUMENTS

1. Structured clinical interview for DSM‐IV axis I disorders (SCID‐I)24 for the diagnosis of mental disorders and collection of sociodemographic data, psychotropic drug consumption, and treatments received. The Spanish version of this interview was used for the present study.25

2. Beck depression inventory‐second edition (BDI‐II),26 a 21‐item self‐report used to assess severity of symptoms of depression in the last 2 weeks, including the day when the test is taken. This measure has excellent internal consistency and convergent validity27 and has been translated and validated for the Spanish population.28 As the primary outcome measure of this study, it was used to determine response (decrease from pre‐ to posttreatment of ≥ 47%)29 and remission rates (posttreatment BDI‐II total score ≤ 12).29

3. Clinical outcomes in routine evaluation outcome measure (CORE‐OM)30 is a 34‐item self‐report questionnaire for the assessment of subjective well‐being, symptoms or problems, life functioning, and risk. It has good psychometric properties and it has been validated for the Spanish population.31 It was considered as a secondary outcome measure. A short form of this questionnaire (CORE‐SFB) was used to monitor patients’ progress, but these data will be reported separately.

4. The Hamilton rating scale for depression (HAM‐D)32 is a clinician‐rated 17‐item instrument. It is the most widely used outcome measure in clinical trials for depression, and also has good psychometric properties. For the present study, the Spanish version of the instrument was administered.33 This scale was used in order to provide convergent validity for the results obtained with the BDI‐II.

5. RGT34, 35 for assessing the presence of cognitive conflicts. This semistructured interview is designed to elicit personal constructs and their assessment as they are applied to a set of significant others.

Assessments

At the initial assessment (T1) the nature of the study was explained. If the patient agreed to participate and gave written consent, the assessment proceeded. BDI‐II was applied to confirm the clinically significant presence of depressive symptoms (a score above 19). Patients meeting this criterion were evaluated using the SCID‐I24 interview to establish the diagnosis of major depressive disorder or dysthymic disorder. CORE‐OM and HAM‐D were also applied. A second assessment session was scheduled for the patients selected in this phase in which RGT was administered. The resulting data were analyzed using GRIDCOR 4.0 software. Patients who presented at least one form of cognitive conflict (implicative dilemma or a dilemmatic construct) were included in the treatment phase (the study protocol36 contains detailed information on this process). Once the treatment was completed, a second assessment was carried out (T2), and a follow‐up assessment (T3) was conducted 3 months after the end of therapy.

ETHICS

The study protocol was approved by the Bioethics Committee of the University of Barcelona (ref. IRB0003099) and by the ethical committees of the centers taking part in the study. All the participants were informed of the implications of the study and signed an informed consent document before enrolling.

STATISTICS

Efficacy Analysis

Following the intention‐to‐treat (ITT) principle, all randomly assigned patients were included into the outcome analysis and a multilevel mixed effects model was applied to deal with missing values. Symptom reduction over time (pretreatment, posttreatment, and follow‐up) was compared between CBT and DFT. The analysis was conducted separately for the BDI‐II (primary outcome measure) and the CORE‐OM (secondary) using a two‐level model. At level 1 (within‐subject), individual slope and intercept of the outcome variable were modeled for each participant over time. At level 2 (between‐subject), parameters (intercepts and slopes) from level 1 were modeled using treatment allocation to test whether the differences in the rate of change were conditioned by type of treatment (CBT or CBT + DFT). The same analysis was conducted for completers.

Cohen's d ^[37] was calculated to express the effect size within group from pre‐ to posttreatment and from posttreatment to follow‐up. Between‐treatment effect sizes (d) were also calculated for posttreatment and follow‐up outcome measures. Effect sizes were calculated for both ITT and completers. To address the problem of missing data, the last observation carried forward (LOCF) method was used. In addition, differences in response and remission rates at the end of the treatment and after 3 months’ follow‐up between groups were assessed using the Fisher's exact test. Tests were considered significant at P < .05 (two‐sided). Analysis of the data, descriptive statistics, and the multilevel effects model were performed with R.38

RESULTS

PATIENT FLOW AND CHARACTERISTICS

The flow of patients over the entire study is displayed in Fig. 1. In total, 315 patients were assessed for eligibility. Of those, 108 did not meet the inclusion criteria and 10 declined to participate from the outset. Of those eligible (n = 197), 56 left the study for various reasons (scheduling conflict, refusal to participate, loss of contact, participation in another psychotherapy process). In all, 141 patients began CBT group therapy and 128 completed this part of the treatment. These 128 patients were randomly assigned to individual therapy in CBT (n = 63) or DFT (n = 65) conditions. The dropout percentages were evenly distributed across CBT (15.87%) and DFT (18.46%). Baseline characteristics are presented in Table 1. There were no significant differences in patient characteristics in either treatment condition.

Figure 1.

CONSORT Diagram of participants in the present study.

Table 1.

Baseline, demographic, and clinical characteristics of participants

	Group therapy (N = 141)		CBT (n = 63)		CBT + DFT (n = 65)		Statistics
Characteristic	M (SD)	n (%)	M (SD)	n (%)	M (SD)	n (%)	t	x 2	P	95% CI
Age (in years)	49.28 (10.98)		50.06 (11.03)		48.37 (11.22)		.86		.39	[−2.20, 5.58]
Gender								.094	.83
Female		107 (75.9)		48 (76.2)		51 (78.5)
Male		34 (24.1)		15 (23.8)		14 (21.5)
Marital status								1.88	.59
Single		21 (14.9)		9 (14.3)		9 (13.8)
Married		73 (51.8)		31 (49.2)		38 (58.5)
Divorced		34 (26.6)		20 (31.7)		14 (21.5)
Widowed		8 (5.7)		3 (4.8)		4 (6.2)
BDI‐II	36.55 (9.56)		37.05 (10.07)		36.31 (9.16)		.43		.66	[−2.62, 4.11]
CORE‐OM	2.13 (0.54)		2.12 (.57)		2.13 (.49)		−.09		.92	[−1.96, 0.17]
HAM‐D	(n = 113)		(n = 40)		(n = 43)
	19.18 (6.33)		19.90 (6.40)		18.54 (6.54)		1.06		.28	[−1.17, 3.89]
GAF	57.26 (7.11)		56.68 (7.10)		57.77 (7.35)		−.83		.41	[−3.67, 1.49]
Diagnosis								1.39	.49
MDD		65 (46.1)		31 (49.2)		27 (41.5)
MDD‐R		56 (46.1)		29 (49.2)		32 (49.2)
Dysthymia		11 (7.8)		3 (4.8)		6 (9.2)
Depressive episodes	2.04 (1.36)		2.02 (1.37)		2.15 (1.40)		−.51		.61	[−0.62, 0 .36]
Previous suicide attempts
Yes		29 (22.7)		15 (23.8)		14 (21.5)		.09	.83
No		99 (77.3)		48 (76.2)		51 (78.5)
Psychiatric medication								1.4	.32
Yes		102 (72.3)		48 (76.2)		42 (66.7)
No		39 (27.7)		15 (23.8)		21 (33.3)
Chronicitya	11.35 (11.69)			11.50 (11.58)		11.33 (11.76)	0.08		.93	[−4.01, 4.36]

Note: Two‐tailed hypothesis.

^aChronicity, number of years from the first depressive episode; CBT, cognitive behavioral therapy; DFT, dilemma focused therapy; CI, confidence interval.

EFFICACY

ITT Analysis

Comparison of the rates of change between treatments was computed with multilevel modeling, including all randomized participants (N = 128). In reference to the first hypothesis, which predicted a significant reduction of symptoms for all participants over time, the results showed a statistically significant reduction in both symptom measures: BDI‐II (β = −6.06, t ₍₁₂₇₎ = −9.85, P < .001) and CORE‐OM (β = −0.24, t ₍₁₂₇₎ = −7.51, P < .001). With regard to the second hypothesis, which predicted a larger decrease in symptoms over time in the CBT + DFT condition (than in CBT alone), the results indicated that treatment allocation was not significantly related to change in either the BDI‐II (β = −1.03, t ₍₁₂₇₎ = −0.56, P = .57) or the CORE‐OM (β = 0.01, t ₍₁₂₇₎ = 0.19, P = .84). The effect size for adding treatment modality to the model is negligible for both measures (R ² = .002 for BDI‐II and R ² = 0 for CORE‐OM). Therefore, there is no support for claiming a differential effect for the therapy condition.

Completers

Analyzing only the data of those participants who completed the treatment (n = 106), similar results were obtained. There was a statistically significant decrease on symptom measures across time for both conditions: BDI‐II (β = − 8.27, t ₍₁₀₅₎ = −9.33, P < .001) and CORE‐OM (β = −0.32, t ₍₁₀₅₎ = −7.42, P < .001). The addition of the linear parameter accounted 44.56% of the variance in the BDI‐II and for 35% of the variance in the CORE‐OM. The decrease in symptoms was not related to treatment allocation BDI‐II (β = −0.62, t ₍₁₀₅₎ = −0.31, P = .75) and CORE‐OM (β = 0.03, t ₍₁₀₅₎ = 0.33, P = .73). The effect size for the addition of the treatment condition was very small (R ² = .006 for BDI‐II and R ² = 0 for CORE‐OM).

EFFECT SIZES

Means, standard deviations, effect sizes, and 95% confidence intervals for the effect sizes (computed for ITT and completers samples at the end of treatment and at 3‐month follow‐up) are displayed in Table 2 for each treatment condition. The statistics for HAM‐D are not reported since it was only applied to 78 patients at pre‐ and posttreatment. Nevertheless, HAM‐D correlated significantly at pretreatment with BDI‐II (r = .56) and CORE‐OM (r = .58) and even presented a high correlation at posttreatment (BDI‐II, r = .83; CORE‐OM, r = .82).

Table 2.

Outcome data at all assessment points and effect sizes with 95% confidence intervals within and between groups

	Treatment condition
		CBT		CBT + DFT		Between group
Measure	Time	Mean (SD)	ES; Cohen's d (95% CI)	Mean (SD)	ES; Cohen's d (95% CI)	ES, Cohen's d (95% CI)
Intention‐to‐treat sample (LOCF, N = 128)
BDI‐II	Pretreatment	37.04 (10.07)		36.31 (9.15)
	Posttreatment	23.71 (15.22)	1.02 [0.71, 1.35]	21.60 (15.50)	1.13 [0.83, 1.46]	0.16 [−0.13, 0.43]
	Three‐month follow‐up	23.88 (15.95)	−0.01 [−0.21, 0.19]	23 (15.04)	−0.09 [−0.23, 0.05]	0.05 [−0.28, 0.4]
CORE‐OM	Pretreatment	2.12 (0.57)		2.13 (0.49)
	Posttreatment	1.53 (0.82)	0.81 (0.63, 1.33)	1.50 (0.76)	0.96 (0.65, 1.28)	0.03 (‐0.31, 0.38)
	Three‐month follow‐up	1.60 (0.81)	−0.08 [−0.24 to 0.07]	1.62 (0.77)	−0.16 [−0.32, 0.004]	−0.03 (−0.38, 0.31)
Completers sample (n = 106)
BDI‐II	Pretreatment	36.94 (9.56)		37.22 (8.96)
	Posttreatment	(n = 53)		(n = 53)
		21.41 (14.65)	1.23 [0.86, 1.63]	19.60 (15.49)	1.37 [1.01,1.75]	0.119 [−0.26,0.50]
	Three‐month follow‐up	(n = 43)		(n = 50)
		21.79 (15.86)	−0.05 [−0.34,0.24]	21.80 (15.61)	−0.12 [−0.31, 0.05]	0.031 [−0.37,0.43]
CORE‐OM	Pretreatment	2.08(0.56)		2.17 (0.50)
	Posttreatment	(n = 53)		(n = 53)
		1.41 (0.78)	0.97 [0.63,1.33]	1.41 (0.77)	1.14 [0.79,1.53]	0.002 [−0.37,0.38]
	Three‐month follow‐up	(n = 43)		(n = 50)
		1.45(0.76)	−0.12 [−0.37,0.11]	1.58 (0.81)	−0.21 [0.42, 0.01]	−0.12 [−0.53,0.28]

Note: ES, effect size; SD, standard deviation; BDI‐II, Beck Depression Inventory‐II; CORE‐OM, clinical outcomes in routine evaluation outcome measure; CI, confidence interval; CBT, cognitive behavioral therapy; DFT, dilemma‐focused therapy; LOCF, last observation carried forward.

Within effect sizes were measured from pre‐ to posttreatment and from posttreatment to 3‐month follow‐up.

Response and Remission

Based on data from patients who completed posttreatment assessment (n = 106), 50.9% (n = 54) responded to treatment and 36.8% (n = 39) met criteria for remission. Although the response rate of the CBT + DFT condition was higher than that of the CBT alone, this difference was not statistically significant (CBT = 43.4% [n = 23], CBT+ DFT = 58.5% [n = 31]; $x^2\left(1\right)=2.41,$ P = .17). Nor did remission rates (CBT = 34.0% [n = 18], CBT + DFT = 39.6% [n = 21]; $x^2\left(1\right)=.36,$ P = .68) differ between treatment conditions. Considering the data of patients who completed the 3‐month follow‐up assessment (n = 93), 50.5% (n = 47) responded to treatment and 37.6% (n = 35) reached remission criteria. At follow‐up, response (CBT = 48.8% [n = 21], CBT + DFT = 52% [n = 26]; $x^2\left(1\right)=$0.09, P = .83) and remission rates (CBT = 34.9% [n = 15], CBT + DFT = 40% [n = 20]; x ²(1) = 0.25, P = .67) did not differ between treatment conditions.

Assessment of Treatment Adherence

Ratings of the adherence scale suggested therapists adhered closely to the respective treatments. In the DFT condition, therapists scored significantly more on DFT items than in the CBT condition (U = .00; P < .001) and comparable levels of adherence were found for the CBT condition (U = .00; P < .001).

DISCUSSION

The current study aimed to assess the differential efficacy between (a) CBT, combining group and individual formats, and (b) group CBT plus an individual dilemma‐focused intervention for depression in adults. The results indicated that both types of treatment significantly reduced the level of depressive symptoms (BDI‐II) and psychological distress (CORE‐OM), although we found no evidence of differential effects between them.

In the ITT analysis, the effect sizes from pre‐ to posttreatment measured with the BDI‐II for both conditions were comparable to those reported in published RCTs using CBT as compiled by Schindler et al.39 (mean effect size for ITT, d = 1.14). This point is worth noting since treatment was relatively short (16 sessions), with a combined group‐individual format, and therapists were relatively inexperienced. There was a higher (nonsignificant) benefit in symptom reduction in the CBT + DFT condition, but this small difference disappeared at 3‐month follow‐up. We found a similar landscape in relation to response and remission rates that were comparable to other studies; the differences favored the condition with DFT at posttreatment, but were not statistically significant and disappeared almost entirely at 3‐month follow‐up.

There may be a variety of reasons for the lack of significant differences between treatment conditions. The most obvious one is that the inclusion of DFT did not really improve the outcome of CBT for depression. In any case, this result does not invalidate the usefulness of the DFT intervention since it proved to be at least as efficacious as CBT, the most prestigious psychological treatment for depression. Both approaches tackle cognitive issues, such as the meaning patients ascribe to events, and both explore alternative ways of perceiving them. However, CBT includes some ingredients such as encouraging patients to engage in pleasurable activities and challenging cognitive errors or distortions in the patient's thinking, which are virtually absent in DFT. The latter intervention proposes that patients face a dilemma in which both sides have both positive and negative implications for the subjects. The focus is, then, on assisting patients to resolve their dilemmas. Further studies might explore which patients benefit most from this differential emphasis on dilemmas.

Another important point is that DFT is not presented as an alternative stand‐alone treatment for depression, but as a putative adjunct to already existing therapies such as CBT. Therefore, in this study all patients had a common CBT group intervention that included half of the sessions so the differential treatment conditions affected only the other half. In these conditions, much larger differential effects would have been needed for the results to yield statistically significant differences.

Now that we know that DFT can be an efficacious ingredient in psychotherapy for depression, we can begin to explore which patients might benefit more (perhaps in areas not measured in this study) from working with their, often implicit dilemmas. Conceivably, those who tend to react more negatively to direct prescriptions or to the questioning of their thoughts and beliefs might accept dilemma‐focused work more easily and engage with it more deeply.

STRENGTHS AND LIMITATIONS

The present study has several strengths. First, the broad range of the sample's characteristics supports the generalizability of the outcomes. Patients were recruited by various types of professionals (GPs, psychiatrists, clinical psychologists) but also using direct advertisements; therefore, our sample includes patients with different levels of depression. Second, the use of manualized treatments favors replicability. Third, researchers’ allegiance was minimized; although the first author and V.C. wrote the DFT manual the second and third authors (A.B. and E.G.) are well known and experienced CBT trainers and therapists. Fourth, this is the first RCT to test the efficacy of a dilemma‐focused intervention for depression. Finally, the relatively high level of efficacy found for DFT as an intervention targeting cognitive conflicts can be taken as additional evidence in support of previous studies13, 14 that have shown the relevance of these conflicts for depression. This study, then, highlights the relevant role of dilemmas in the conceptualization of depressive suffering, and of the difficulties of treating depressive symptoms in psychotherapy, and provides some novel techniques for dealing with them.

The study also has limitations. The treatment was conducted by novice therapists; however, all of them received intensive supervision by experienced psychotherapists. Second, it was not possible to prevent patients from seeking additional treatment during follow‐up. Third, although the HAM‐D was included in the protocol as a clinician‐rated secondary outcome measure, due to momentary lack of training of the evaluators during the first months of its implementation it could only be applied reliably to 61% of the sample. However, the majority of the sample was also assessed with this instrument and the resulting data correlated strongly with the self‐report assessment using BDI‐II.

CONCLUSIONS

There is no significant evidence that the addition of DFT to CBT treatment increases efficacy in treating depressed patients in a combined (group + individual) format. Around 37% of the patients achieved remission at posttreatment and gains were maintained at 3‐month follow‐up. However, this study may provide some support for the use of DFT as an individual adjunct therapy to CBT for patients who might find this approach more acceptable. Further research is needed to determine the clinical usefulness of its addition to extant psychotherapies for depression.