Effectiveness and cost effectiveness of cognitive behavioral therapy (CBT) in clinically depressed adolescents: individual CBT versus treatment as usual (TAU)

The aim of this study is to investigate the effectiveness and the cost-effectiveness of the individual CBT program the “D(o)epression course” in a sample of referred adolescents with a Depressive Disorder according to DSM-IV-TR [25] in a randomized controlled trial. We expect that CBT will be more effective than TAU (without CBT). Furthermore, cost-effectiveness of CBT and the cost of illness of clinical depression in adolescents will be established. Potential moderators (comorbidity, severity of depression, age, ethnicity, gender, suicidal thoughts and psychopathology in parents) and mediators (negative automatic thoughts, cognitive emotion regulation and attribution style) for the effectiveness of CBT will be studied. The role of non-specific treatment variables (therapeutic alliance, client expectancy, client satisfaction, treatment adherence) will be taken into account as well (see Figure 1).

This study is designed as a multi-site, randomized controlled clinical trial in which individual CBT will be compared to TAU. Four multiple informant (adolescent, parent, therapist) assessments are done: prior to treatment (pre-test assessment), immediately after treatment (post-test assessment or after 15 sessions), 6 months after treatment (6 month follow-up) and 1 year after treatment (1 year follow-up). To investigate potential mediators two temporary assessments are conducted, each after 5 therapy sessions.

All sessions will take place within routine outpatient care within second order public mental health clinics to enhance external validity. All treatments in both conditions are delivered by psychologists with at least a master's degree-level and two years of experience within professional mental health care. These professionals are trained before delivering the treatments to enhance treatment integrity.

Adolescents who are diagnosed with a depression according to the K-SADS-PL [26, 27], who meet inclusion criteria and do not meet the exclusion criteria are randomly assigned to either individual CBT or TAU. Random assignment per adolescent is executed using computer generated block randomisation and stratified per mental health centre. See Figure 2 for patient’s flow chart.

The design of this study is according to the guidelines specified by the Task force on promotion and dissemination (1995) and approved by the independent Medical Ethics Committee (METC) of the Utrecht Medical Centre at Utrecht University.

Based on previous research [16] an effect size (Cohen’s d) of 0.53 is expected. Power calculations indicated that 70 adolescents per condition (assuming an alpha of 0.05, a statistical power, 1-beta, of 0.80 and a drop-out of 20%; power calculations in STATA) are required to detect a difference in depression diagnosis. In total 140 adolescents will be included.

Participants in both groups will be compared afterwards for their match on the following characteristics: age, ethnicity, gender, educational level parents and severity of the depression. If significant differences between two groups appear, these variables will be controlled for in analyses.

The participants are recruited from fourteen different specialized mental health care institutions spread all over the Netherlands. An experienced psychologist within the mental health centre informs the adolescents and their parents about the study. After written informed consent by the adolescent and his/her parents, a trained independent researcher carries out an interview to check for the inclusion and exclusion criteria.

The experimental treatment is a protocolled individual CBT program, the D(o)epressie course. It consists of 15 weekly sessions of 45 minutes, two parent sessions after 3 and 9 weeks and a meeting with the parent at the end of treatment. The D(o)epressie course is, as the CWD-A, based on the social learning theory about the aetiology of depressions by Lewinsohn [28]. According to this theory, there is a connection between the number of positive interactions between a person and his environment on one hand, and depression on the other. A triggering event such as a stressful life event causes a negative spiral of less positive interactions leading to more negative thoughts and a deteriorating depressed mood. The aim of the intervention is to reduce depressive complaints in adolescents with a depressive disorder. Since depressive episodes are multi-factorial determined, the focus of the intervention is broad. The intervention contains representative components of CBT [29] namely: psycho-education (information about depression and the rationale for the aetiology of the complaints and the treatment of them), setting attainable goals (translate large goals into realistic short term goals), self monitoring (registration of the mood, activities and thoughts), activation (planning frequent, joyful activities), improving social skills and communication skills (improvement and stimulation of social behaviour), relaxation techniques, cognitive restructuring (identifying and changing unrealistic negative thoughts about the self, others and events), role play and problem solution skills (teaching the creation of solutions for problems via brainstorm, choosing, trying and evaluating) and relapse prevention. Exercises are executed within the sessions and are generalised into daily life by means of homework assignments. In the parent sessions, parents will receive psycho education and information on CBT. Therapists are not allowed to conduct treatments in both conditions.

The control treatment consists of TAU for clinical depression. After a short telephone survey we concluded that TAU in the Netherlands consists of: Interpersonal Therapy (IPT), family therapy, parent counselling, medication, mindfulness training, acceptance commitment therapy (ACT), psychodynamic therapy (short duration), (non-directive) counselling, creative therapy and running therapy, and CBT. However, in this study CBT is not allowed within TAU. The content of TAU and the treatment techniques used, are monitored.

In Table 1, concepts, source and time of assessment of all used instruments are presented. The adolescent and parent complete the self-report questionnaires online at home. Both, the adolescent and the parent have a separate login code to secure privacy. The parents fill in a paper version of the cost diary at home. The therapist completes the questionnaires at the office.

The primary outcome measure is the presence of the depression diagnosis, as measured by the Kiddie-Schedule for Affective Disorders and Schizophrenia, present and lifetime version (K-SADS-PL) [26, 27], a widely used semi-structured diagnostic interview. The K-SADS-PL assesses a wide range of diagnoses (present and life time) including their severity. The view of the adolescent, the parent and the independent clinician are taken into account. Concurrent validity of the K-SADS-PL is supported [26]. Also the interrater agreement is high (range 93% to 100%) and test –retest reliability is excellent for present and lifetime diagnoses of major depression (.77 to 100) [26]. Convergent validity of the depression screen criteria and the diagnoses generated with the K-SADS-PL was confirmed but divergent validity is only partly supported within an inpatient sample [30].

A broad range of secondary outcome measures will be assessed namely total symptoms of major depressive disorder or dysthymic disorder, severity and improvement of depression, global functioning, quality of life, suicide risk, comorbidity, alcohol and drug use, parental depression and psychopathology, parenting and conflicts. Costs, moderators, mediators and treatment characteristics will be assessed as well.

The degree of depressive symptoms are measured with a self report measure, the Child Depression Inventory-2 (CDI-2) [31, 32]. The CDI-2 is a revision of the CDI [33, 34] and was translated in Dutch. It was expanded with a version for the parents (CDI-P) [31] as well. The severity of the depression is rated by the independent clinician on the K-SADS-PL (see above) and by the therapist on the Clinical Global Impression-severity scale (CGI-S) [35]. Improvement of depression, in reference to the severity of the depression at the start of the treatment, is also rated by the therapist on the Clinical Global Improvement scale (CGI-I) [35]. Global functioning of the adolescent is measured by an independent clinician and therapist on the Children Global Assessment Scale (CGAS) [36, 37]. The Dutch version of the EuroQol Questionaire (EQ-5D adolescent and parent version) [38] is used to establish quality of life as expressed in quality adjusted life years (QALYs). Apart from the K-SADS-PL, suicide risk is also assessed with a newly developed self-report questionnaire, which focuses on frequency of suicidal thoughts, wishes, plans and actions over the past two weeks.

Comorbidity and psychopathology is assessed with the K-SADS-PL, but also with the Youth Self Report scale (YSR) for adolescents and the Child Behavior Check List (CBCL) for parents [39, 40]. Comorbidity between depression and anxiety is very high [12] therefore anxiety symptoms are also assessed separately with the Scared-5 [41].

Personality of the adolescent is assessed with the Quick Big Five Personality Inventory (QBF) [42]. For this study, we also constructed the Life Event Scale (LES) [43], which is a self-report measure about life events (including drug abuse, bereavement, maltreatment and suicide attempts), their date of occurrence and their impact on the adolescents well being.

Psychopathology of both parents is measured with the Adult Self-Report (ASR) [44]. The degree of depressive symptoms in parents is assessed with the Dutch version of the Beck Depression Inventory, second edition (BDI-II-NL) [45, 46].

Parenting, in particular consistency, responsiveness, positive parenting, harsh discipline, psychological control and behavioral control is assessed with subscales of different instruments filled in by both the adolescent and the parent. The Parenting Dimensions Inventory (PDI) [47, 48] was used to measure consistency, the degree to which the parent shows predictable discipline behaviour. The Nijmeegse Rearing Questionnaire (NOV) [49, 50] measures responsiveness, the degree to which the parent is responsive for the needs, signals and condition of the child, and attachment, the degree to which the parent feels emotionally connected to the child. Positive parenting is measured with 6 items from the Alabama Parenting Questionnaire (APQ) [51]. The Ghent Parental Behavior Questionnaire (SOG) [52] assesses physical harsh discipline or physical punishment. Psychological control is assessed using the Psychological Control Scale (PCS) [53], which measures the degree in which the parent tries to control the child in an intrusive way. The degree in which parents monitor their children, that is behavioral control is measured with 6 items of the (Monitoring questionnaire, VTH) [54]. Competence, the parents’ perspective on their ability to address rearing practices, is established with the subscale competence of the Nijmeegse Parental Stress Index (NOSI) [55, 56].

Attachment from child to parent is measured using 13 items of the Psychological Availability and Reliance on Adult (PARA) [57]. The degree of conflicts (quarrels, irritations and antagonism in the child–parent relationship) was measured with 6-item Network of relationship inventory (NRI) [58].

The economic evaluation is done by registration of costs in a cost diary based on the Trimbos Institute and Institute of Medical Technology Assessment Questionnaire on Costs Associated with Psychiatric Illness (TiC-P) [59] and PRODISQ [60]. The registered costs are directly related to health care or indirect health care (out-of-pocket costs, costs of informal care) and direct costs outside health care (monetary value of production losses caused by absence and reduced productivity). The costs will be considered separately from the perspective of mental health and from society. The mental health costs are the costs, which are credited to the mental health care budget, the decision-maker’s perspective. The societal costs are the costs of direct (mental) health care as well as indirect costs such as lost productivity, school absent and out-of-pocket costs.

Potential moderators that are analysed are severity of the depression, comorbidity, parental depression and psychopathology and demographic variables. Demographic information is gathered by adding questions about gender, age, ethnicity, education level and family income to the self-report questionnaires.

Three mediators are investigated namely negative automatic thoughts (CNCEQ) [61], cognitive emotion regulation (CERQ) [62] and attribution style (CASQ) [63]. The Cognitive Negative Cognitive Error Questionnaire (CNCEQ) measures cognitive errors namely the underestimation of the ability to cope, personalizing without mind reading, selective abstraction, over generalizing and mind reading. The Cognitive Emotion Regulation Questionnaire (CERQ) measures a broad set of cognitive emotion regulation strategies which are used in response to the experience of threatening or stressful life events; Self-blame, Other-blame, Rumination, Catastrophizing, Positive refocusing, Planning, Positive reappraisal, Putting into perspective and Acceptance. The Children’s Attributional Style Questionnaire (CASQ) is a self-report measure with three dimensions of attribution; internal- external, stable-unstable and global- specific.

Several non-specific treatment variables will be investigated. The client’s credibility, expectancy and involvement regarding treatment are assessed with the Parent Expectancies for Therapy Scale (PETS) [64] which was revised for adolescents. Previous treatments for depression, including complementary and self-help treatments, are administered with the inventory of History of Treatments (VEHI) [65]. Satisfaction with treatment is measured with the Service Satisfaction Scale (SSS) [66]. The Cooperation With Treatment scale (CWT) [67] is used to assess the degree of cooperation with treatment as observed by the therapist. The quality of the therapeutic alliance is assessed with the Therapy Alliance Scale for Adolescents (TASC) [68]. The content of treatment is assessed in both conditions with the Therapy Procedure Checklist (TPC) [69]. Treatment integrity will be established by recording two randomly chosen sessions that are observed and rated.

Missing values will be imputed. Intent-to-treat as well as completer analyses will be conducted. The effect of the intervention is analyzed with a chi-square test on the dichotomous primary outcome measure, presence of the depression diagnosis (K-SADS-PL). The cost-effectiveness analyses will be based on the comparison of costs and effects in both conditions and will be done according the international guidelines [70]. The cost- effectiveness analyses will be done separately from the perspective of mental health and society over a period of 6 months and 1 year. The effects of societal costs will be expressed in years to live, corrected for quality of life (QALYs). The costs of CBT versus CAU will be expressed in 1) incremental costs per QALY and 2) incremental costs per adolescent with a depression in full remission.

The secondary continuous outcome measures will be analyzed with repeated measures MANCOVA using the pre-test as a covariate as is recommended for RCT’s with pre-, post- and follow-up measurements [71]. For each questionnaire, the effect size (Cohen’s d) from pre- to post-treatment is defined as (M_pre-M_post)/SD_pooled, where SD_pooled = [(SD_pre ² + SD_post ²)/2]. CBT versus TAU pre-post effect sizes are calculated as follows [(M_{CBT post}-M_{TAU post})/SD_{pooled post}] – [ (M_{CBT pre}-M_{TAU pre})/SD_{pooled pre}]. Furthermore, analyzes will be conducted to establish clinical significance and the reliable change index [72].

Moderators will be analysed by multi-group analyses for dichotomous variables. The continuous variables will be analysed using hierarchic regression analyses. Mediator effects will be analysed using hierarchical regression analyses and structural equation modelling.