Effectiveness of trauma-focused treatment for adolescents with major depressive disorder.

Background: Major Depressive Disorder (MDD) in adolescence has a high prevalence and risk of disability, but current treatments show limited effectiveness and high drop-out and relapse rates. Although the role of distressing experiences that relate to the development and maintenance of MDD has been recognized for decades, the efficacy of a trauma-focused treatment approach for MDD has hardly been studied. Objective: To determine the effectiveness of eye movement desensitization and reprocessing (EMDR) therapy as a stand-alone intervention in adolescents diagnosed with MDD. We hypothesized that reprocessing core memories related to the onset and maintenance of MDD using EMDR therapy would be associated with a significant decrease in depressive and comorbid symptoms. Method: We recruited 32 adolescents (12-18 years) fulfilling DSM-IV criteria for mild to moderate-severe MDD from an outpatient youth mental health care unit. Treatment consisted of six weekly 60-min individual sessions. Presence or absence of MDD classification (ADIS-C), symptoms of depression (CDI), symptoms of posttraumatic stress (UCLA), anxiety (SCARED), somatic complaints (CSI), and overall social-emotional functioning (SDQ) were assessed pre and post-treatment and 3 months after treatment.
Results: 60.9% of the adolescents completing treatment no longer met DSM-IV criteria for MDD after treatment anymore, and 69.8% at follow-up. Multilevel analyses demonstrated significant posttreatment reductions of depressive symptoms (CDI: Cohen's d = 0.72), comorbid posttraumatic stress, anxiety and somatic complaints, while overall social-emotional functioning improved. These gains were maintained at 3-month follow-up (Cohen's d = 1.11). Severity of posttraumatic stress reactions significantly predicted the posttreatment outcome; however, duration of MDD, number of comorbid disorders, or having a history of emotional abuse, emotional neglect or physical neglect were not predictive for outcome. Conclusions: This is the first study suggesting that EMDR therapy is associated with a significant reduction of depressive symptoms and comorbid psychiatric problems in adolescents with mild to moderate-severe MDD.

Introduction

Major depressive disorder (MDD) is one of the most common psychiatric disorders of childhood and adolescence (Mullen, 2018). It has been estimated that 14% to 25% of adolescents experience at least one episode of a depressive disorder before entering adulthood (Ryan, 2005). MDD is a leading cause of disability in terms of burden of disease, and poor functioning (Smith, 2014; Stikkelbroek, Bodden, Deković, & van Baar, 2013). Furthermore, MDD with adolescent onset has been found to be associated with a range of physical health problems and other mental health disorders in adult life (Thapar, Collishaw, Pine, & Thapar, 2012; Weersing, Jeffreys, Do, Schwartz, & Bolano, 2017) as well as with social problems, legal problems, and elevated suicide risk (Stikkelbroek et al., 2013).

There are several treatments for adolescent MDD, with cognitive behavioural therapy (CBT) being recommended as a psychosocial intervention for both mild and moderate to severe forms of MDD (NICE guideline, 2019). The efficacy of CBT has been established in several reviews and meta-analyses (Compton et al., 2004; Oud et al., 2019; Weisz, McCarty, & Valeri, 2006) and is known to be the intervention with the largest body of evidence. However, the effectiveness of CBT for this target population has been found to be attenuated when compared to active control conditions, and when applied to clinically complicated samples (Lewis et al., 2010; Weersing et al., 2017). For example, in a Dutch multicentre study carried out in specialized mental health institutions for depressed adolescents (12–21 years; Stikkelbroek, 2016) CBT was not found to be more effective than treatment as usual (TAU). In fact, CBT performed worse on both drop-out (CBT 57% vs TAU 41%) and the number of adverse events during treatment (CBT 3 vs TAU 0). The relative poor efficacy of CBT for adolescent MDD is underlined by a mean effect size of 0.29, as computed in a recent meta-analysis of CBT for adolescent MDD (Weisz et al., 2017). Therefore, there is an urgent need to enhance the treatment outcome for depressed adolescents.

Eye movement desensitization and reprocessing (EMDR) therapy (Shapiro, 2017) is a recommended treatment for posttraumatic stress disorder (PTSD; ISTSS Guidelines Committee, 2018; World Health Organization, 2013). It has been found to be capable of processing memories of distressing events (Shapiro, 2017). Individuals with MDD frequently report an adolescent onset, often following exposure to distressing experiences (Mandelli, Petrelli, & Serretti, 2015; Monroe, Slavich, & Georgiades, 2014). Strongest evidence for a relationship between childhood adverse events and the development of MDD has been found for interpersonal experiences, like humiliation and entrapment (Kendler, Hettema, Butera, Gardner, & Prescott, 2003), and different forms of abuse, primarily emotional abuse and neglect (Hovens et al., 2010; Mandelli et al., 2015).

In the past 5 years, several studies have been conducted demonstrating preliminary evidence for the efficacy of EMDR therapy in the treatment of MDD in adults. Promising results were obtained from studies investigating EMDR therapy as an adjacent therapy to CBT (Hofmann et al., 2014), to pharmacological treatment (Minelli et al., 2019; Ostacoli et al., 2018) and to inpatient treatment (Hase et al., 2015, 2018). Three studies, investigating the efficacy of EMDR as a stand-alone treatment, demonstrated significant reductions of depressive symptoms (Gauhar, 2016), even for patients with long-term depression (Wood, Ricketts, & Parry, 2018) and treatment-resistant depression (Minelli et al., 2019). Treatment of MDD also resulted in significant decreases of trauma symptoms (Gauhar, 2016) and anxiety symptoms (Minelli et al., 2019), improved social functioning (Minelli et al., 2019) and quality of life (Gauhar, 2016).

While research involving EMDR treatment for adults with depression is emerging rapidly, research on the effectiveness for adolescents has not followed at the same pace. To our knowledge only one case series (Bae, Kim & Park, 2008) has been published, which included two adolescents. Although these adolescents did not report traumatic events in their history, they had experienced loss and rejection in family and peer relationships. EMDR was targeted on these memories involving loss and rejection. Both adolescents displayed a significant decline of depressive symptomatology after three and seven EMDR sessions, respectively. This result was maintained at 2- and 3-month follow-up. Given that distressing or traumatic events have been found to be associated with the onset and maintenance of depressive disorders it is conceivable that adolescent MDD is responsive to EMDR therapy when the memories of these events are targeted and resolved. Therefore, the purpose of the present study was to investigate the effectiveness of EMDR in adolescents (12–18 years) with a primary diagnosis of MDD (DSM-IV; American Psychiatric Association, 1994). It was hypothesized that the application of EMDR therapy would be associated with a significant decrease in the severity of depressive symptoms and in the percentage of patients meeting DSM-IV criteria for MDD. Furthermore, we hypothesized that treatment would be associated with a significant decrease in the severity of co-morbid symptoms (i.e., posttraumatic stress symptoms, anxiety, somatic and emotional-behavioural problems). In addition, we examined whether duration of MDD, baseline posttraumatic stress disorder severity, number of comorbid disorders, or having a history of emotional abuse, emotional neglect or physical neglect would significantly predict post-treatment outcome. Moreover, to determine the safety of the intervention for this target group the number of adverse events was recorded.

Method

Participants

Patients were recruited from the regular referrals to the Mental Health Institution (MHI) Rivierduinen Leiden Children and Youth department, an outpatient mental health care unit, between December 2015 and March 2018. Inclusion criteria were: (a) age 12–18 years (b) mild to moderate depressive disorder according to the criteria of the Dutch guidelines (Dutch Multidisciplinary Guideline for Depression in Youth, 2009), i.e. five to eight symptoms according to DSM IV (American Psychiatric Association (APA), 1994), interference of the condition with a maximum of three out of four life domains (school, social situations, leisure, and home/family) and a Global Assessment of Functioning (GAF) >45 (c) identified memories of at least one distressing or traumatic event related to the depressive symptomatology. Exclusion criteria were: (a) severe suicidal or psychotic symptoms, (b) a suicide attempt or serious non-suicidal self-injury requiring hospitalization in the 3 months prior to intake (c) substance dependence (d) IQ estimated to be ≤80 based on information from the referral letter or diagnostic phase (e) insufficient Dutch language skills.

Procedure

Patients referred for treatment of depressive symptoms at the participating department were screened for eligibility by the first and third author. After the institutions’ regular intake assessment, adolescents who had depressive symptoms were informed about the study by their clinician. Next, pre-treatment assessment (T0) was administered and a session with the EMDR therapist was planned. This session was aimed at checking the inclusion criterion ‘identified memories of at least one distressing or traumatic event related to the depressive symptomatology’; since no standardized instruments are available to make an inventory of depression-related memories beforehand. Subsequently, remaining in- and exclusion criteria were again checked, and in case of eligibility and willingness to participate, informed consent of both adolescents and their caregivers was obtained. Following EMDR treatment, remaining symptoms and need for further treatment of each adolescent were discussed with participants, parents and the multidisciplinary staff.

Outcomes were measured post-treatment (T1) and 3 months after treatment (follow-up, T2) by a team of eight independent assessors (i.e., trained clinicians and master level students), who were not involved in the treatment. Adverse events, such as suicidal attempts, serious self-injurious behaviour and crisis contacts, were recorded using a checklist by the therapists at the start of each session.

For administration of the ADIS-C, all assessors were trained according to a protocol consisting of observing live and videotaped interviews and completed an exam to prove adequate administration of the interview. Supervision was provided for each assessment and the reports were reviewed and discussed to ensure that administration, scoring and reporting would not drift. Therapists who conducted the EMDR sessions were blind to assessment data.

Intervention

The Dutch version of the standard EMDR procedure with age-specific adaptations for children and adolescents (De Roos, Beer, de Jongh, & Ten Broeke, 2015) was used for the present study. This procedure includes eight phases: history taking, preparation, assessment, desensitization, installation, body scan, closure and re-evaluation (Shapiro, 2017). Treatment consisted of six weekly 60-min individual treatment sessions. Memories were placed in a hierarchy based on the Subjective Units of Disturbance (SUD) and were treated subsequently from high to low SUD. Each session was followed by a 10- to 15-min meeting with the adolescent and one or both parents. The content of this meeting was discussed beforehand with the adolescent and could comprise any one of the following elements: (1) an outline of the content of the session (2) parents’ view on the course of symptoms in the week before the session and (3) the need and possibilities for emotional support of the adolescent after the session.

In the present study, EMDR therapy was carried out by a team of seven clinical psychologists. Six of them were registered EMDR Europe Practitioners. All sessions were videotaped and all therapists participated in monthly two-hour supervisions by a certified EMDR Europe Child and Adolescent Consultant (second author). Additional supervision by email or telephone was provided on request. Early completion of treatment (<6 sessions) was assigned in cases where all target memories from the case conceptualization could be retrieved without emotional disturbance (i.e. SUD related to the memory was reduced to zero).

Assessment instruments

The participants’ demographic characteristics (e.g. living condition, level of education, history of mental health service use) were assessed at baseline. All measurements were administered at all assessments (baseline (T0), post-treatment (T1) and 3-month follow-up (T2), except the Childhood Trauma Questionnaire (CTQ), which was only administered at T0).

The primary outcome measure of this study was the presence of an MDD diagnosis on the Anxiety Disorders Interview Schedule for DSM-IV Child version (ADIS-C). The ADIS-C assesses a wide range of diagnoses according to DSM-IV criteria (American Psychiatric Association, 1994; Siebelink & Treffers, 2001; Silverman & Albano, 1996). The ADIS-C has strong evidence for providing reliable and valid diagnoses and proved to possess adequate sensitivity to clinical change in treatment outcome research (Silverman & Ollendick, 2005).

The Dutch version of the Children’s Depression Inventory (CDI; Kovacs, 1985; Timbremont, Braet, & Roelofs, 2008) was used to assess affective, behavioural and cognitive aspects of depressive symptoms in the past 2 weeks. The CDI includes 27 items dealing with sadness, self-blame, loss of appetite, insomnia, interpersonal relationships, and school adjustment which are scores on a 3-point Likert scale (0–2, total range 0–54). Acceptable levels of internal consistency, validity and test–retest reliability have been established (Kovacs, 1985; Roelofs et al., 2010). Reliability of the total scale in the current study was acceptable (α = .78).

The University of California at Los Angeles Post-traumatic Stress Disorder Reaction Index Adolescent version (UCLA PTSD RI; Steinberg, Brymer, Decker & Pynoos, 2004) was used to screen for exposure to traumatic events and to assess PTSD symptoms. The symptom scale consists of 22 items which are scored on a 5-point Likert scale (0–4; total range 0–88) and assesses the frequency of occurrence of PTSD symptoms during the past month. The original list of traumatic events covering medical trauma, natural disasters, community violence and abuse was adapted for the present study by adding four items concerning experiences of loss and separation (death and separation from loved ones) and humiliation (bullying and being isolated/ignored). These experiences, considered as ‘attachment trauma’ (Hofmann et al., 2014) have been identified as being connected to the onset of depressive episodes (see, e.g., Bae et al., 2008; Kendler et al., 2003). Reliability of the UCLA total scale in the current study was excellent (α = .91).

The Dutch version of the Screen for Child Anxiety Related Emotional Disorders (SCARED; Muris, Bodden, Hale, Birmaher, & Mayer, 2007; Muris, Merckelbach, Schmidt, & Mayer, 1998) was used to assess signs of anxiety disorders in the past 3 months. The SCARED is a 41-item inventory rated on a 3 point Likert-type scale (0 = ’not true’ or ‘hardly ever true’; 1 = ‘somewhat true’ or ‘sometimes true’; 2 = ’very true’ or ‘often true’; total range 0–82). Reliability of the SCARED total scale in the current study was excellent (α = .92).

The Children’s Somatization Inventory (CSI; Garber, Walker, & Zeman, 1991; Dutch version: Treffers, Goedhart, & Siebelink, 1998) was used to assess the extent and frequency of 35 somatic complaints (e.g. headaches, constipation, dizziness) in children and adolescents in the past 2 weeks. Items are scored on a 5-point Likert scale (0: ‘not at all’, 4 ‘a whole lot’) (total range 0–140). Reliability of CSI total scale in the current study was excellent (α = .93).

The Dutch adolescent version of the Strengths and Difficulties Questionnaire (SDQ; Goedhart, Treffers, & Van Widenfelt, 2003; Goodman, 1997) was used as a global assessment of psychological problems. The SDQ consists of 25 items which are scored on a 3-point Likert scale ranging from ‘not true’, ‘somewhat true’ or ‘certainly true’ (total range 20–80). In this study, the ‘total difficulties scale’ was used in the analyses. Reliability of the SDQ total scale in the current study was good (α = .80).

The Childhood Trauma Questionnaire (CTQ; Bernstein et al., 2003) was used to assess experiences of childhood maltreatment. The CTQ is a self-report list consisting of 28 items which are scored on a 5-point Likert scale. The CTQ has a good criterion validity in both a clinical and a healthy sample (Bernstein et al., 2003). The subscales Emotional neglect, Emotional Abuse and Physical Neglect were used in the analyses. Reliability of the CTQ total scale in the current study was excellent (α = .90).

Data analyses

All statistical analyses were performed using IBM SPSS Statistics for Windows (version 24). Descriptive statistics were produced to describe the demographic characteristics and baseline variables of the sample. To investigate the effect of EMDR therapy time contrasts were created (T0-T1, T0-T2) by means of dummy coding. Linear mixed model analyses were used for the main analyses. The mixed model for investigating the general efficacy of the EMDR intervention included a random term for the intercept and fixed terms for time contrasts (T0-T1, T0-T2). The covariance matrix was set to scaled identity. Effect sizes were calculated using Cohen’s d (Cohen, 1992), and determined by calculating the mean difference between scores from baseline (T0) to post-treatment (T1) and from baseline (T0) to follow-up (T2), dividing the result by the pooled standard deviation (Cohen, 1988). Cohen’s d was calculated for both depressive symptoms and comorbid symptoms.

To identify possible predictors of treatment outcome (depressive symptoms as measured by the CDI), baseline posttraumatic stress symptom severity (UCLA), number of comorbid disorders (ADIS-C) and depression-specific baseline factors (i.e., history of emotional abuse or neglect (CTQ), and duration of the disorder) were entered separately in the linear mixed model analyses. The same time contrasts as described above were used (i.e. T0-T1 and T0-T2). The level of significance was set at a =.05.

Results

Patient flow and sample characteristics

Before referral to our institution, and entering the study, the majority of the patients (n = 23; 72%) received some form of treatment. Based upon the UCLA, 73% of the adolescents had experienced a non-criterion A-event (i.e., bullying/humiliation, being ignored/isolated and bereavement of a loved one) prior to therapy. The characteristics of the study population are presented in Table 1. The treatment sample was characterized by a long duration of MDD (M = 72.4 weeks, SD = 74.42, range 18–364 weeks; depressive disorder was present in the family in 59%) and a high number of comorbid disorders (M = 2.39, SD = 1.38; for all but one patient comorbid disorders were classified at T0). These comorbid disorders comprised primarily social phobia, generalized anxiety disorder and dysthymic disorder. From the different forms of childhood trauma, emotional neglect (32% above cut-off), emotional abuse (23% above cut-off) and physical neglect (19% above cut-off) were reported most frequently (Table 2).

Table 1.

Sample characteristics.

Characteristics of the study population	N	%
Gender, male	5	16
Nationality, Dutch	28	88
Living condition
Living with both parents	19	59
Living with one parent	8	25
Parents divorced, living with both, alternating	2	6
Other (adoptive parents, grandparents, shared student household)	3	9
Level of education (n = 31)
Low to middle level secondary education or vocational education	12	39
High level: secondary education/high school/professional	18	58
education	1	3
History of mental health service use
Outpatient psychiatric treatment	13	41
No treatment	9	28
Social work/school counselling	7	22
Multiple treatments	2	6
Other treatment	1	3
Receiving psychotropic medication	1	3
Index trauma from UCLA PTSD-RI at T0 (n = 30)
Bullying/humiliationa	10	33
Being ignored/isolateda	7	22
Bereavement of a loved one	5	17
Serious accident	2	6
Sexual assault	2	6
Illness/medical trauma	1	3
Natural disaster	1	3
Other experience with violence/serious danger	2	2
Total number of comorbid DSM-IV classifications on ADIS-C at T0 (n = 31)
0	3	10
1	5	16
2	9	29
3	7	23
4	5	16
5	2	7
Comorbid DSM-IV classification on ADIS-C at T0 (n = 31)
Social phobia	18	56
Generalized anxiety disorder	12	38
Dysthymic disorder	9	28
Attention-deficit/hyperactivity disorder	5	16
Specific phobia	5	16
Posttraumatic stress disorder	4	13
Obsessive compulsive disorder	2	6
Panic disorder	2	6
Separation anxiety disorder	2	6
Agoraphobia	1	3

UCLA PTSD-RI: University of California at Los Angeles Post-traumatic Stress Disorder Reaction Index Adolescent version; ADIS-C: Anxiety Disorders Interview Schedule for DSM-IV – Child version.

a Category was added to the original list of possible traumatic experiences for this study.

Table 2.

Results of linear mixed model analyses.

				T0 vs T1				T0 vs T2
	M (SD)N = 32T0	M (SD)n = 24T1	M (SD)n = 25T2	Unstandardized β(95% CI)	t-test (df)	p value	Cohens’ d	Unstandardized β (95% CI)	t-test (df)	p value	Cohens’ d
CDI	26.59(7.64)	20.25(9.74)	16.92(9.73)	−6.183(−8.833 to −3.533)	t(50.07) = −4.69*	<.001	0.72	−9.868(−12.519 to −7.218)	t(50.07) = −7.48*	<.001	1.11
UCLA	36.00(15.28)	19.22(13.21)	18.77(14.54)	−16.488(−22.651 to −10.324)	t(49.94) = −5.37*	<.001	1.17	−17.881(−23.044 to −11.717)	t(49.94) = −5.83*	<.001	1.16
SCARED	56.80(19.28)	37.43(22.46)	36.70(22.06)	−18.235(−23.945 to −12.526)	t(45.54) = −6.43*	<.001	0.93	−19.235(−25.018 to −13.451)	t(46.14) = −6.69*	<.001	0.97
CSI	36.83(23.16)	26.35(19.34)	22.91(21.14)	−11.373(−19.429 to −3.317)	t(48.87) = −2.84*	.007	0.49	−15.213(−23.342 to −7.083)	t(49.84) = −3.76*	<.001	0.63
SDQ total	19.50(5.05)	15.96(7.06)	15.17(6.71)	−37.311(−43.953 to −30.668)	t(50.33) = −11.28*	<.001	0.58	−38.2997(−44.971 to −31.629)	t(50.07) = −7.48*	<.001	0.73

CDI: Children’s Depression Inventory; UCLA: University of California at Los Angeles Post-traumatic Stress Disorder Reaction Index Adolescent version; SCARED: Screen for Child Anxiety Related Emotional Disorders; CSI: Children’s Somatization Inventory, SDQ: Strengths and Difficulties Questionnaire.

Figure 1 shows the patient flow through the study. In total, 32 patients were included with a mean age of 15.8 years (SD = 1.50). Five (15.6%) were early completers and needed only four (n = 1) or five (n = 4) EMDR sessions. Seven (21.9%) dropped out before the end of treatment; three withdrew from treatment and study because of a lack of interest, one because of spontaneous remission, one because the parent demanded more intensive treatment, one because of insufficient ability to attend treatment sessions and one because the adolescent was not able to experience the emotional load related to the identified depression-related memories (SUD). Independent sample t-tests were performed to compare dropouts with completers on age, gender, baseline severity of MDD symptoms and posttraumatic stress reactions, duration of MDD and number of comorbid disorders at baseline. Therapist factors were excluded because of the low number of patients in each cell of the crosstabs (32 clients were treated by 8 therapists). From all tested variables, only duration of MDD differed significantly between the groups, with dropouts having a shorter duration of depressive symptoms (m = 36.86, SD = 11.60) compared to completers (m = 86.22, SD = 79.84) (t (df = 27.14) = 2.98, p = 0.006).

Figure 1.

Patient flow chart.

Depression

As can be seen from Figure 2, 14 out of 23 adolescents who completed treatment and T1 assessment (60.9%) no longer fulfilled the criteria of a MDD diagnosis as determined by the ADIS-C after treatment (T1). For the intent to treat group this rate was 43.8% (14 out of 32). The percentage no longer obtaining a MDD diagnosis further increased to 69.8% for completers (16 out of 23) and to 50.0% for the intent to treat group (16 out of 32) at follow-up (T2). Table 3 shows a significant decrease of depressive symptoms (CDI), with a sharp reduction of symptoms during treatment (T0-T1: Cohen’s d = 0.72) and a further decrease afterwards (T0-T2; Cohen’s d = 1.11).

Figure 2.

Proportion of patients completing treatment meeting DSM-IV criteria for MDD (ADIS-C) at different points in time.

Table 3.

Results of predictor analyses on treatment outcome as measured by level of depressive symptoms (CDI).

	Time × predictor effects
	Β	t (p)	95% CI
Severity of PTSD symptoms	0.27	2.85 (.01)	0.08–0.46
Number of comorbid disorders	1.37	1.23 (.23)	−0.87–3.62
Duration of MDD	1.65	0.52 (.61)	−4.75–8.05
Emotional abuse	0.29	1.02 (.31)	−0.29–0.87
Emotional neglect	0.28	0.92 (.36)	−0.34–0.91
Physical neglect	−0.27	−0.24 (.81)	−2.54–1.99

PTSD:Posttraumatic Stress Disorder, MDD: Major Depressive Disorder.

Cohen’s κ was calculated for the inter-rater agreement on the ADIS-C at T1. Half of the interviews (n = 12) were double-scored and κ was calculated at .082.

Comorbid symptoms

Posttraumatic stress symptoms (UCLA), anxiety symptoms (SCARED), somatic complaints (CSI) and general social-emotional problems (SDQ) showed a significant decrease following EMDR therapy which was maintained at follow-up. For all measures, medium to large effect sizes were found (see Table 2).

Prediction of post-treatment outcome

Severity of PTSD symptoms significantly predicted treatment outcome as measured by the CDI (p < .01; see Table 3); with higher levels of PTSD symptoms at baseline predicting greater reductions of MDD symptoms during treatment. No other significant predictors of treatment outcome could be identified.

Adverse events

No adverse events were reported during the study.

Discussion

To our knowledge, besides one study on two cases (Bae et al., 2008) this is the first outcome study that examined the effectiveness of EMDR therapy in adolescents with a primary diagnosis of MDD. The results demonstrated a significant decrease in depressive symptoms and comorbid posttraumatic stress, anxiety, somatic complaints and overall social-emotional functioning. More than 60% of the adolescents completing treatment no longer fulfilled the criteria of an MDD diagnosis after treatment. The medium to large effect sizes suggested clinically relevant effects that were maintained at 3 months of follow-up.

It is noteworthy that remission of depressive symptoms was achieved after only six one hour sessions. For the completers, most of the MDD related memories that were identified were processed using EMDR therapy. This suggests that the number of six sessions seemed sufficient for most of the adolescents. To this end, there are no similar studies with adolescents to compare our results to, but the present findings are consistent with the case study by Bae et al. (2008), and studies reporting the effect of EMDR therapy as a stand-alone intervention in adult MDD (Gauhar, 2016; Minelli et al., 2019; Wood et al., 2018).

Treatment result in our study seemed to be significantly influenced by baseline severity of posttraumatic stress reactions, meaning that adolescents with high levels of posttraumatic stress reactions demonstrated a larger decrease of depressive symptoms during treatment. None of the remaining predictors (i.e., number of comorbid disorders, duration of MDD and having a history of emotional abuse, emotional or physical neglect) seemed to have influenced treatment outcome. To this end, it could be particularly hopeful that the results suggest that a long duration of MDD and having many comorbid problems (74.2% had two or more comorbid disorders) did not significantly interfere with the effects of EMDR therapy in this population, which is in contrast to what is often observed in studies that used cognitive behavioural therapy (Weersing et al., 2017).Yet, these results should, of course, be interpreted with caution, since this is a first small feasibility study with limited power.

The fact that adverse events, such as suicidal attempts, serious self-injurious behaviour and crisis contacts, did not occur suggests that treatment of MDD using a trauma-focused approach is safe. Related to this, the drop-out rate (21.9%) was comparable to drop-out rates obtained in other studies of EMDR therapy as a stand-alone treatment of MDD (e.g. Gauhar, 2016: 23%; Minelli et al., 2019: 15.4%; Wood et al., 2018: 30%). On the other hand, compared to the dropout-rate (57%) of a Dutch CBT study with a comparable population (Stikkelbroek et al., 2013) the drop-out rate of the present study can even be considered as low.

This study is a pilot study and has, therefore, several limitations. The most important limitation is that the absence of a waiting list and/or an active control condition so we cannot rule out that the observed improvements were either an artefact of time or due to placebo effects. Secondly, the sample size was small and the follow-up period of 3 months was relatively short. Thirdly, given the wide array of studies showing that this population often suffers from suicidal intentions we wanted to be cautious and excluded individuals with severe depression. Although the results of the present feasibility study do not support the notion that the use of EMDR is unsafe in terms of adverse events, it could be argued that the exclusion of a severe subgroup might make the results less generalizable. Interestingly, however, there are few studies with which we can compare our results on this point. For example, while in the study by Stikkelbroek et al. (2013) with comparable mean CDI total scores, severity of depression was not an exclusion criterion, acute suicide risk was. Besides the obvious limitations of the present study, some strengths should also be noted. An important strength of this study is that it included a representative group of adolescents seen in routine mental health care, in terms of a relatively long duration of depressive symptoms, many comorbid problems, and having received unsuccessful prior treatment or counselling, which makes the results highly generalizable. Another strength is the use of a semi-structured DSM based clinical interview, conducted by trained interviewers. Finally, the therapists used a manualized treatment protocol, session checklists and video-recorded sessions, which were evaluated and discussed during supervision to enhance treatment integrity.

Yet, despite the promising results, most patients still suffered from symptoms of depression after completion of EMDR treatment. More specifically, 39.1% of adolescents who completed treatment still fulfilled the diagnostic criteria of MDD according to ADIS-C; 62.5% scored above CDI cut-off of 16, and 68% of the patients received additional treatment interventions (i.e., CBT, emotion regulation training, parent counselling, family interventions, medication or a combination of these) for the remaining complaints after the study. Further research is needed to determine whether the addition of evidence-based interventions aimed at cognitive restructuring or family interventions may lead to stronger symptom reduction, even lower drop-out, and less after care. More generally, given the heterogeneous nature of MDD, it would be naïve to expect that one single treatment approach, e.g. trauma-focused treatment, would be sufficient to cure all different appearances of MDD. That is, for a certain subgroup of adolescents with MDD treatment using EMDR therapy might be of value as our results suggest, but the treatment of other subgroups, with regard to which (combinations of) interventions are the most successful, requires further investigation.

In conclusion, the results of this study showed that the application of EMDR therapy was safe and associated with a significant reduction of depressive symptoms and comorbid psychiatric problems. Clearly, randomized controlled trials with sufficient statistical power are needed to establish the efficacy of EMDR therapy in adolescents with mild to moderate or severe MDD.