CBT vs. Schema Therapy: Is There a Difference?
The simple answer to this question is yes. Schema therapy tends to be initiated once CBT has not been successful for patients and includes some elements of CBT. Let’s get into what each therapy entails and exactly what the difference is.
CBT (Cognitive Behavioral Therapy)
CBT is a popular form of psychotherapy that examines the relationships between thoughts, feelings, and behaviors. Its purpose is to help people overcome emotional problems, make patients aware of automatic thoughts, and how those automatic thoughts affect how patients feel and behave.
The psychiatrist and patient work together to expose patterns of thinking and determine the best ways to positively change those patterns. Because CBT is an active intervention, the patient should expect homework and practice outside of therapy. CBT emphasizes changing the ways people think in order to improve their moods.
Schema Therapy (ST)
Schema therapy combines CBT, gestalt, imagery and other techniques to help weaken the maladaptive schemas and coping styles and re-build the patient’s healthy side. In a sense, schema therapy works with the patient’s inner child to help correct the emotional difficulties experienced during childhood in a strong effort to change them.
Schema therapy doesn’t simply target feelings of depression or anxiety; it works to rectify past emotional disturbances and change long-term patterns.
The Difference
While schema therapy combines elements of CBT, these elements are often practiced later. The purpose of schema therapy is to bring to light schemas suffered by a patient during childhood that have entrenched themselves in their adult life.
In CBT, recognizing automatic thoughts and how they make patients feel and behave is sufficient. However, in schema therapy, the focus is to do all of the above while changing the schemas so that they are no longer a hindrance to the patient’s adult life.
To learn more about CBT and schema therapy, contact Dr. Scott Shapiro at 212-631-8010 or [email protected] for a fifteen minute phone consultation at no charge.
Dr. Shapiro is a NYC psychiatrist who is also certified in CBT and schema therapy.
Primary and secondary measures for outcome comparison
We choose BDI-II [87] as primary outcome to capture the change of symptom severity over the course of 7 weeks. BDI-II represents a widely used and well established self-assessment of MDD [104, 105] and thus assures comparability of results with former research. In order to overcome some of BDI-II related restrictions such as sensitivity to maladaptive personality traits [106], we added MADRS [88], as clinical assessment of change in MDD as secondary outcome. Further secondary outcomes are recovery rate (change in diagnosis) measured by the Munich-Composite International Diagnostic Interview (M-CIDI) [107], change in general psychopathology using the Brief Symptom Inventory (BSI) [108], change in global functioning and quality of life according to the World Health Organization Disability Assessment Schedule (WHODAS) [109] and the World Health Organization Quality Of Life (WHO-QOL) assessment [110]. Further secondary outcome measures are drop-out rates, and remission rates. Based on previous research, our main hypotheses are that ST including its intervention techniques is more effective (superior) in the treatment of MDD compared to IST as a nonspecific-common factor psychotherapy [111] (H1), and 2) ST is non-inferior compared with CBT regarding treatment response and recovery rates operationalized by primary and secondary outcomes after the intervention and after 6 months [24] (H2).
Further measures of potential MOCs for process comparison
In order to delineate the MOCs of ST and compare them with underlying mechanisms of CBT and IST, we include multiple potential mediators, most of them measured in weekly intervals (cf. [62]). Depending on treatment condition, we hypothesize different MOCs to be relevant.
In accordance with the theoretical concept of ST, we consider schemas (Young Schema Questionnaire – Short Version 3) [112], Young Positive Schema Questionnaire) [113], modes (Schema Mode Inventory) [114], therapeutic alliance (Working Alliance Inventory) [65, 112], attachment style (Relationship Scales Questionnaire) [115], affect (Need for Affect Questionnaire [116], the Positive and Negative Affect Schedule) [117, 118], and emotion regulation (Emotion Regulation Questionnaire) [119, 120] to function as MOCs of ST regarding the change in MDD (H3).
Based on previous research [62] and in contrast to ST, for CBT we assume cognition related aspects of depression (Automatic Thought Questionnaire [121], Dysfunctional Attitude Scale [122], Cognitive Style Questionnaire – Short Form in German [123], Internal and External Control Beliefs Scale [124], and self-efficacy measured by Allgemeine Selbstwirksamkeitsskala [125]) to mediate change in MDD (H4).
Additionally, we expect all three conditions, but especially the non-specific treatment arm (IST) to work through common factors of psychotherapy, such as goal setting (Goal Attainment Scale) [126], therapy expectations (Patient Questionnaire on Therapy Expectation and Evaluation) [100], further general mechanisms (Scale for the Multiperspective Assessment of General Change Mechanisms in Psychotherapy) [127], and non-specific session characteristics (Session Evaluation Questionnaire) [128] to be associated with change (H5).
In order to examine temporal and causal relations between MOCs and outcomes measures, measurements should happen at the same time point [55, 56]. We therefore include outcome measures (BDI-II, BSI) and MDD related constructs such as thought-action fusion in the context of suicidality [129], resilience (Brief Resilience Scale) [130], perceived stress (Perceived Stress Scale) [131, 132], and coping with depressive symptoms (Response Style Questionnaire) [133] on a weekly base. See Additional file 1: Appendix A for details regarding time points of measurement.
Beyond that, the exploration of potential MOCs of ST should not be restricted to weekly self-reports, but also include information from other sources in order to get a more comprehensive understanding of the course and processes during psychiatric care [62]. Therefore, we complement the set of weekly self-report variables by adding actimetry measures to assess locomotor activity, psychophysiological measures during the actual psychotherapy session, and continuous measures of depression-related parameters such as mood and repetitive negative thinking:
Actimetry
Changes in locomotor activity and unbalanced rest-activity cycles are widely known as key features of depression [134, 135]. Actimetry provides an objective and unobtrusive mean of assessing sleep and activity with high temporal resolution, so that activity- and sleep-related symptoms (e.g., insomnia) and potential treatment effects (e.g. change in activity) can be dynamically captured [136].
In the current study, locomotor activity and sleep behavior are assessed using actimetry wrist-watch devices. The devices are worn by participants throughout the treatment phase of the study over the course of 7 weeks continuously except during activities that may damage the device or be a risk to participants (e.g., swimming/contact sports).
Psychophysiology
Psychophysiology in general and interpersonal physiology in particular, is related to psychotherapy processes [137, 138] which are specifically relevant in ST [15], such as therapeutic alliance [139] and emotion regulation techniques applied by the therapist [140]. We therefore assess physiological parameters such as heart-rate (HR), electrodermal activity, and body temperature in a sub-sample of patients during psychotherapy sessions through hand-wrist devices that are worn by patients and therapists. Thus, we aim to gain insight into the synchronicity of physiological processes underlying ST specific intervention techniques, e.g. imagery rescripting, and to investigate how these processes differ from CBT techniques such as cognitive modification [141].
EMA of depressive core symptoms and repetitive negative thinking (RNT)
During the seven-week treatment phase, patients are asked to participate in an app-based EMA [142] that acquires momentary states of different core symptoms of MDD (e.g., mood, withdrawal) and RNT [143]. EMA offers several advantages compared to traditional questionnaire assessments, which are particularly relevant for the investigation of clinical processes, such as being closer in time to the experienced phenomenon, reducing recall bias (specifically relevant for MDD samples) [144], collecting data in naturalistic settings [145], and examine within-person processes which are especially important in psychotherapy. The measurements take place three times a day and comprise a total of eleven items. The app is installed on the patient’s personal smartphone or a provided device.
Such high-frequency measures of depressive symptoms and RNT can provide new insights into the dynamic changes over the course of the treatment phase and are particularly useful, given the variability of symptoms within and between days and individuals.
Further measures for exploring treatment prediction
Beside outcome and process comparison, the current study includes a variety of potentially MDD-related predictors from the domains of neuropsychology, ECG, biology, and cognitive and social neurosciences in order examine patterns of patient characteristics using an exploratory approach. For an overview of all assessment domains and time points of measurements, see Table 1. Some of the listed measures are obligatory, while others are optional sub-studies that are not applied to all participants (see Table 1). In order to ensure the practicability and implementation of the measures in clinical routine, a feasibility study was conducted prior to the start of the actual trial and subsequently, processes were adapted if necessary.
Table 1 Overview on time points of measurement
Full size table
Note: T0 = baseline measures before treatment start; T1 = first study week; etc.; T8 = 6 months follow-up; T9 = 24 months follow-up; x = one time point of measurement per week; xx = two time points of measurement per week; xxx = quasi continuous data collection ranging from three times per day (EMA) to every 30 s (actimetry). *Assessment domains are obligatory; Secondary outcome self-reports include BSI, WHODAS, and WHO-QOL; For further details on the MOC measures (particularly common factor measures) see Additional file 1: Appendix A.
Neuropsychology
Cognitive impairment plays a key role as a transdiagnostic factor in psychiatric disorders in general and MDD in particular [146,147,148]. These deficits affect different cognitive domains such as memory, executive functions, attention, and learning [149, 150] and in many cases outlast the remission of depressive symptoms [151, 152]. They function as a mediator of functional impairment in MDD in general [153] and thus, as potential predictor and working mechanism of treatment. Therefore, we assess cognitive functions in order to identify its influence on the outcome effects of treatment before treatment (T0), after treatment (T7), and 6 months after completion of the study at follow up assessment (T8). We assess three basic domains of cognitive processing: attention, executive functions, and memory. The cognitive test battery includes tests from the Test of Attentional Performance [154], which is used to assess cognitive inhibition (Go/No-go-Task), working memory (Dual n-back), and cognitive flexibility. The test battery “Materialien und Normwerte für die neuropsychologische Diagnsotik” [155] is administered to test episodic memory, word fluency and sensitivity to interference (Stroop task). Additionally, we collect information on attention and cognitive flexibility with the Trail Making Test (TMT) and the d2-R [156], and assess intelligence with the “Mehrfachwahl-Wortschatztest” [157].
Electrocardiography
Heart-rate variability (HRV) parameters carry important information on the status of the autonomous nervous system that is unstable in stress-related disorders [158, 159] and has been found to correlate with the severity of depressive symptoms [160]. Therefore, we included Two-channel mobile ECG is routinely obtained during standardized conditions (5 min of resting state, 1 minute of deep breathing, overnight measurement of twelfe hours) at T0 and T7 to extract of HR and HRV parameters.
Biological parameters
Biological parameters would be most welcome and important tools in predicting response to specific psychotherapeutic or psychopharmacological interventions [161]. Recently, genome-wide association studies for unipolar depression have revealed a number of significantly associated loci [162, 163] and epigenetic modifications are considered to play an important role in the pathogenesis and therapy response in patients suffering from this disorder [164]. Gene-environment-interactions such as the role of trauma exposure have been discussed to shed a light to the genetics of MDD [165]. Even though, there is an emerging body of research extending these gene-environment-interactions to learning-contexts of psychotherapy [166], so far there are no satisfying validated biological parameters to assist in the decision-making process regarding the best treatment option for patients suffering from MDD. Studies aiming to integrate underlying pathomechanisms in this process have been designed [167]. In this study, we are aiming to identify biological parameters to contribute to the understanding of the response to psychotherapy.
We test for biological parameters of therapy response according to current and future evidence from clinical and preclinical data. Serum and plasma samples are stored for the analyses of parameters possibly associated with therapy response. The possible levels of investigation include genetics, epigenetic measures such as deoxyribonucleic acid (DNA) methylation, non-coding ribonucleic acid (RNA) but also other epigenetic markers such as histone modifications as well as proteomics, gene expression and metabolomics. Blood is drawn before treatment start (T0), in treatment week 4 (T4) and after treatment (T7) as well as during follow-up assessments at 6 months and 24 months after end of treatment. The sample includes ethylenediamine tetra acetic acid (EDTA) blood for DNA extraction (genome-wide genotyping and DNA methylation), RNA tubes for microRNA expression as well as small non-coding RNAs and serum and plasma for proteomics and metabolomics. Plasma can also be used to assess miRNAs circulating in exosomes. Finally, only at the baseline visit, we collect peripheral blood mononuclear cells (PBMCs) using Ficoll separation. At least 30 Mio. cells are stored for each individual and these can be used as a source tissue for induced pluripotent stem cell programming as well as functional assays in live mononuclear cells. Cells are stored with dimethyl sulfoxide (DMSO) as the stabilizer in liquid nitrogen. Induced pluripotent stem cells are established from blood cells and tested. In addition, the patients receive the standard safety routine blood draws of the clinical routine. The total blood volume drawn before treatment (T0) is 55 ml, the total blood volume drawn after 4 weeks (time point T4), 7 weeks (T7) after 6 and 24 months (T8) is 28 ml. All samples are entered into the biobank at the Max Planck Institute of Psychiatry which has been approved by the ethics committee of the Ludwig-Maximilians-University, Munich, under the project-ID 338–15.
Neuroimaging
In OPTIMA we offer a basic neuroimaging protocol acquired on 3-Tesla clinical MRI system (General Electric, Milwaukee, USA) in order to extract information on macroscopic and microscopic brain features as well as brain function.
Neuroimaging in the context psychotherapy research regarding MDD is built on evidence that the clinically heterogeneous condition of MDD is reflected in structural and functional abnormalities of brain circuits [168, 169]. These abnormalities, on one hand, represent target systems that are modified by the learning processes stimulated by psychotherapy. On the other hand, heterogeneity of these abnormalities across subjects is expected and hypothesized to hold a predictive value with regard to the most effective type of psychotherapy for an individual [170]. This latter hypothesis will mainly be tested using a response-status-by-treatment-type interaction framework applied to extracted MRI features or to voxelwise/vertexwise measures. Analyses are designed anatomically explorative and will thus be controlled for multiple testing. Post-hoc analyses will comprise pair-wise group comparisons (e. g. responders of ST against non-responders of ST), comparisons of responders of one treatment against pooled non-responders, and a general responder/non-responder comparison. Examples for established feature extraction techniques are listed per MRI subdomain in the following:
-
(1)
A high resolution T1-weighted imaging with high contrast between grey matter, white matter and cerebrospinal fluid serves as basis for voxel-based and surface-based morphometry analyses using established imaging post-processing approaches. Discrete cortical thickness and surface area features, voxelwise volume maps and surface meshes will be calculated, and the above defined group comparisons performed for the entire anatomical space (either covered by extracted anatomical features or by voxels/vertex points).
-
(2)
Diffusion tensor imaging (DTI) is acquired in order to allow for the reconstruction of fiber tracks as basis for structural connectivity and to calculate voxel wise maps of measure of fiber integrity such as fractional anisotropy. DTI is suited to probe specific hypotheses on ‘hard-wired’ connectivity patterns of specific networks as anatomical basis of functional re-organization. Probabilistic region-by-region-connectivity values using the FreeSurfer cortical/subcortical parcellation for ROI-definition will represent the main target features of this domain.
-
(3)
Resting state functional MRI (rs-fMRI) over 6.5 min is acquired in an eyes-open-crosshair-fixation design with parallel eye-tracking. Respiration and pulse measurements are taken for later denoising steps. Resting state fMRI allows for different types of functional connectivity analyses at the whole brain level (e. g. functional connectivity density maps forwarded to second level analyses) or at the level of specific circuitries. For the latter, group independent component analysis will be used to extract a set of within-network and between-network connectivity using validated analysis pipelines.
-
(4)
Task fMRI: In order to acquire information on social interaction information processing, a shortened version of an established social interaction task is performed, which involves gaze contact with an interaction partner who reacts in real-time in the context of an object selection task (same image geometry and parameters as rs-fMRI for optimal coupling) [171, 172]. This task validly recruits specific neural systems that are involved in social processes and that are highly relevant to participation in psychotherapy. The above-mentioned group comparisons represent (voxelwise) second level analyses based on first level activation maps that hold information on the individual’s strength of the social network recruitment.
MRI measurements (1)–(3) will be repeated at post treatment, yet we expect dropouts here and thus refrained from building the main hypotheses on longitudinal MRI data. For details of the tested domains and applied procedures see Additional file 1: Appendix B.
Bayesian social learning task
Different psychotherapy approaches rely on basic processes of learning. This is particularly relevant for ST, which aims to overcome EMS, that is enable individuals to make new interpersonal experiences, and focuses on the therapeutic alliance and patient-therapist interaction [15, 27]. Therefore, we included a social learning task, that enables additional insights into the underlying learning and decision-making mechanisms, in other words, why and how participants learn and behave a certain way [173].
We use a reward-based learning task that requires the integration of non-social and social cues in conjunction with computational modeling [174]. In this task, participants have to learn about the winning probabilities of two cards in order to win points, which will be turned into a financial gain at the end of the study. In addition, a face in the center of the screen looks at one of the two cards, before the participant can make his/her choice. The probability of this gaze shift being helpful or not is also systematically manipulated. Both card and gaze probabilities fluctuate according to a fixed schedule, which is unbeknown to the participant, and do so independently from one another. Behavioral responses to this task are collected to assess participants’ performance in terms of total points achieved. Here, the impact of the non-social and the social domain can also be studied. Furthermore, computational modelling allows to assess learning and decision-making parameters estimated for each participant from their behavior. These parameters could help to shed new light onto psychotherapeutic processes, which also rely on social learning [175], and potentially help to predict treatment outcomes [176].
Sociodemographic, clinical and personality parameters
In the past, sociodemographic and clinical parameters have been used to predict outcome of the treatment of depression [177,178,179], but little is known about their associations and interplay with potential predictors as described above. Therefore, we include sociodemographic (such as age, gender, socioeconomic status, etc.), clinical parameters (such as comorbidities, symptom severity, etc.) and personality measures (Assessment of the DSM-IV Personality Disorders (ADP-IV) [180], the DSM-V Level of Personality Functioning Scale – Self Report (LPFS-SR) [181], the Personality Inventory for DSM-5 (PID-5) Short Version [182,183,184]), stressful and traumatic life events (Childhood Trauma Questionnaire (CTQ) [185], Social Readjustment Scale (SRRS) [186]), and motivation related constructs (Behavioral Inhibition System/Behavioral Activation System Scales (BIS/BAS) [187]). Data analysis and data management.
Power and statistical analysis
Power estimation for the current investigation is based on the main research question on the effectiveness of ST and hypothesis H1 on the superiority of ST over IST regarding the primary outcome (BDI-II). We presume that the minimal clinically important difference (MCID) regarding BDI-II scores should be related to initial depression severity and a patient perspective of perceived improvement [188]. Button and colleagues estimated a minimum reduction of 17.5% of BDI-II scores as MCID. Based on our baseline pilot data, an effect size of d = 0.4 would allow us to detect all outcome differences that can be considered as MCID in our target sample which consists of moderate to severe depressed patients in an inpatient and day clinic setting. If setting power to 0.80 and α to 0.05 while using two-sided t-tests and following a 1:1:1 randomization to ST, CBT, or IST, it is necessary to recruit n = 99.1 per group resulting in an overall sample size of N = 300 (rounded) participants to identify differences in BDI of d = 0.4.
Regarding H2 on the non-inferiority of ST compared with CBT, a sample size of n = 100 per group, a one-sided significance level of α = 0.05, and setting power to 0.80 will allow us to evaluate a non-inferiority margin of d = .36. This is even lower than what can be considered as MCID [188] and takes into account the potential role of concurrent medication during the psychotherapy treatment.
We will apply different methods such as Holm procedure [189] in a scientifically appropriate manner regarding all future data analyses derived from the current study in order to prevent family wise error rate. The Holm method can be applied in same cases like Bonferroni correction to control for multiple testing, but is a more advanced and powerful tool [190].
For the analysis of primary and secondary outcome variables, we will apply linear mixed-effect models (e.g. to explore predictors) which outperform other approaches like analysis of covariance when data is missing not completely at random [191] . Additionally, linear mixed-effect models have increased power compared to simple linear regression approaches as the intra-class coefficient increases [192], thus, a targeted sample size of N = 300 is likely a conservative estimate to assess treatment effects with a power of 80% and an alpha level of 0.05. When investigating treatment outcome in clinical trials, the role of missing values needs to be considered. In order to deal with them adequately, we will follow an intention-to-treat approach using multiple imputation techniques [193] and use more specialized statistical analyses such as survival analyses e.g., to investigate dropout as a clinically relevant secondary outcome. For investigating specific and non-specific MOCs in ST, CBT, and IST treatment arm (H3 – H5), we will fit growth models within multilevel and structural equation model frameworks taking into account the nested structure of the data and potential meaningful growth over time [194, 195]. Multilevel models are able to differentiate within- from between-person variation considering the hierarchies within the data (such as time point of measurements within individuals) and by including random slopes and random intercepts [194, 196].
Data management and monitoring
Data collection and management is conducted according to German law. Here, patient data is stored on encrypted institute servers in pseudonymized manner to restrict access to full details (i.e., personal identifying and study data) to dedicated study personnel only. In order to ensure data quality, double data entry is applied. If requested by a participant, all individual data is removed from all servers immediately. Data presented in publications will be fully anonymous and will not allow identification of study participants. Study documents will be kept at the Max Planck Institute of Psychiatry for the duration of the study and consecutive data analysis. All data that is not kept in the biobank, will be deleted 25 years after end of the study.
The occurrence of adverse events, defined as the development of acute suicidality, and serious adverse events, defined as suicide attempt, will lead to the immediate exclusion of the participant from any study procedures. In such cases, necessary psychiatric care will be provided. Serious adverse events are reported to the Institutional Ethic Committee of the Faculty of Medicine of Ludwig-Maximilians-University Munich. If they are related to study procedures, the study is terminated immediately.
Since it is a psychotherapy trial, blinding of participants and personnel (except from raters) in the OPTIMA trial is impossible and potential adverse events or deteriorations directly assignable to treatment conditions. Therefore, the establishement of a data monitoring committee, which is normally installed in masked trials to supervise adverse events and potential relations to the experimental treatment condition, is not necessary.
