Participants
Patients aged between 18 and 60 years are recruited from four Dutch tertiary clinics specialized in the treatment of anxiety and eating disorders. Patients diagnosed with AN (both restrictive and binge/purge subtype) are recruited from Altrecht Eating Disorders Rintveld, Zeist, and Rivierduinen Eating Disorders Ursula, Leiden, and patients with OCD from Altrecht Academic Anxiety Centre, Utrecht, and Overwaal Centre for Anxiety Disorders, Nijmegen. Most of the patients referred to these centres suffer from moderate to severe AN or OCD.
Eligible are patients whose AN (or eating disorder not otherwise specified, clinically referred to as AN) or OCD diagnosis is consistent with the Diagnostic and Statistical Manual of Mental Disorders, fourth edition text revision (DSM-IV-TR ).
The criteria for exclusion are: severe neurological illness (including a history of seizures, stroke, or Parkinson’s disease), severe comorbid psychiatric disorder (clinical significant major depressive disorder, current acute phase of bipolar disorder, current psychosis, substance dependence/abuse), intellectual impairment, defined as an IQ < 80, estimated with the Dutch Adult Reading Test (DART) [
61], which is a Dutch version of the National Adult Reading Test (NART) [
62], and inability to adequately speak or read Dutch. Current use of antidepressants and antipsychotics is allowed provided that dosages are kept constant during the experimental part of the study. Since benzodiazepines can dampen the effect of cognitive treatments [
63], their use is restricted to a daily maximum dose (up to 20 mg for temazepam or an equivalent dose) taken as a sleep medication. On request, patients can withdraw their participation at any time.
Inclusion and exclusion criteria are verified with the aid of the structured clinical interview for DSM-IV-TR axis I disorders (SCID-I [
64]), while for OCD patients to be enrolled, a Yale-Brown Obsessive-Compulsive Severity Scale score ≥16 is required [
65,
66]. In all patients comorbidity with either OCD or AN is allowed.
Outcome measures
The primary outcome measures are: 1) disease-specific psychopathology: Eating Disorder Examination Questionnaire (EDE-Q) [
70] for AN severity; Yale-Brown Obsessive Compulsive Scale (Y-BOCS) [
65,
66] for OCD severity.
The Eating Disorder Examination Questionnaire (EDE-Q) [
70,
71] is the self-report version of the Eating Disorder Examination (EDE) [
72], a semi-structured interview to evaluate ED psychopathology. The EDE-Q assesses attitudinal and behavioural aspects of EDs over a 28-day period. It has four subscales assessing concerns about shape, weight and eating, and restraint. The EDE-Q consists of subscale scores and a total scale score. The EDE-Q has excellent internal consistency (Cronbach α 0.78-0.93). The subscales have also excellent test-retest reliability over a 2-week period (Pearson r ranging from 0.81 to 0.94) [
73].
The 10-item Yale-Brown Obsessive-Compulsive Severity Scale (Y-BOCS) [
74] is a clinician-rated, semi-structured interview-based scale that is broadly used to assess obsessive-compulsive symptom severity. The scale has two parts, each with five questions, with each subscale assessing five aspects of OCD pathology: 1) time consumed, 2) degree of interference, 3) distress, 4) resistance, and 5) perceived control. The first subscale gives an obsession score (maximum: 20), the second a compulsion score (maximum: 20), and together yield a total score (maximum: 40). It has a strong internal consistency (Cronbach α .88-.91), inter-rater reliability (r 0.82–0.98), and test-retest reliability in clinical and nonclinical samples were excellent.
The secondary outcome measures are: 1) disease-specific psychopathology: Obsessive-Compulsive Inventory-Revised (OCI-R) [
75], 2) quality of life and daily functioning: EuroQol (EQ-5D) [
76], Eating Disorder Quality of life questionnaire (ED-QoL) [
77], Obsessive Compulsive Disorder Quality of life questionnaire (OCD-QoL) [
78], and Work and Social Adjustment Scale (WSAS) [
79], 3) cost-effectiveness (direct and indirect health-related costs as evaluated by means of patients’ self-reported QoL ratings and medical and societal burden: Trimbos/iMTA Questionnaire for Costs associated with Psychiatric illness (TiC-P) [
80], 4) general anxiety: Beck Anxiety Inventory (BAI) [
81], and 5) Depression: Beck Depression Inventory, second revision (BDI-II) [
82].
The Obsessive-Compulsive Inventory-revised (OCI-R) [
75] evaluates the frequency, and distress experienced of OCD symptoms using six subscales: washing, obsessing, hoarding, ordering, checking, and neutralizing. The OCI-R has good psychometric properties [
75,
83,
84] and is sensitive to the effects of treatment [
85]. Internal consistency was adequate (obsessing α = .82, washing α = .86, checking α = .88, neutralizing α = .86, ordering α = .90, hoarding α = .90, total scale α = .81 ) and test-retest reliability was excellent (Spearman’s correlations between .79 and .91) [
75].
The EuroQol (EQ-5D) [
76] is a standardized self-report instrument gauging five dimensions: mobility, self-care, usual activities, pain/discomfort, and anxiety/depression. Each dimension is rated on three levels (no problems, some problems, and extreme problems). Further, the EQ-5D uses a visual analog scale from 0 to 100, on which the respondent marks his general well-being. The EQ-5D is used in a variety of studies to determine health status in a wide variety of patients.
The Eating disorder Quality-of-Life questionnaire (ED-QoL) [
77] generates a disorder-specific self-rated quality-of-life score. It consists of 25 items assessing the influence of eating behaviours/body weight in four subscales: psychological, physical/cognitive, financial, and work/school. A total score is calculated as the average of the items of the four subscales. Higher scores indicate a lower quality of life. The ED-QoL has a good internal consistency (Cronbach α = 0.84–0.95), and test-retest reliability (intraclass correlations were .97 for the psychological subscale, .87 for the physical/cognitive subscale, .90 for the financial subscale, .14 for the work/school subscale, and .93 for the total score) [
77].
The Obsessive-Compulsive Disorder Quality of Life (OCD-QoL) is a self-report questionnaire [
78] gauging the influence of OCD on the respondent’s quality of life. It was adapted from the ED-QOL and thus also has four subscales: psychological, physical/cognitive, financial and work/school, whose 25 items are scored and interpreted similarly.
The Work and Social Adjustment Scale (WSAS) [
79] examines to which extent the disorder has caused the patient functional impairment. A total score of 10 or below is normal, with scores of 10–20 indicating significant functional impairment but less severe clinical symptomatology, and scores of 20 and above suggest moderate to severe impairment or more severe psychopathology. The test-retest correlation for the total WSAS score was 0.73 and the scale has an acceptable to good internal consistency (Cronbach alpha 0.79–0.88).
The Trimbos/iMTA questionnaire for costs associated with psychiatric illness (TIC-P) [
80] evaluates the utilization of medical services such as the number of contacts with the general practitioner and other care providers (e.g., medical specialists and paramedics) as well as medication used during the last 3 months. The costs associated with the experimental and control interventions delivered during this study will be calculated as recommended in the Dutch handbook on cost calculations in health care [
86]. Reference unit prices of corresponding health-care services will be used and cost-utility calculated by relating the difference in direct medical costs per patient receiving CRT + TAU and SAT + TAU to the difference in terms of quality-adjusted life years (QALY) gained (cost-utility), yielding a QALY estimate.
The Beck Depression Inventory, second revision (BDI-II) [
82] is a gold-standard self-report measure to assess the severity of depressive symptoms. The BDI-II comprises 21 categories of statements. Respondents are asked to choose the statement that describes the way they felt in the past 2 weeks. Each item is rated on a 4-point Likert-type scale ranging from 0 to 3, with higher scores indicating higher levels of depression. The maximum total score for all 21 items is 63. Score categories range from 0 to 13 (minimal depression), 14 to 19 (mild depression), 20 to 28 (moderate depression), and 29 to 63 (severe depression). The BDI-II has good psychometric properties [
87]. The BDI-II shows good validity compared to the Hamilton Depression Rating Scale (Pearson’s r of 0.71). The inventory has high internal consistency (Cronbach α = .91) [
88].
The Beck Anxiety inventory (BAI) [
81] gauges symptoms of anxiety (such as trembling hands, dizziness, sweating) by means of 21 questions. Respondents indicate if and to what extent they have experienced any symptoms during the past week on a 4-point Likert scale ranging from 0 (not at all) to 3 (severely). The BAI discriminates well between anxious and non-anxious groups and is useful as screening measure for anxiety in a variety of clinical populations. The BAI has a high validity and high internal consistency (α .92).
The following variables will be tested as potential mediators of intervention response: 1) cognitive flexibility, using a Stroop task [
89], including Delis-Kaplan Executive Function System (D-KEFS) card nr. 4, as well as Detail and Flexibility Questionnaire (D-Flex) [
90], Trail Making Test (TMT) [
91], 2) response inhibition: Stop Signal Reaction Task (SSRT) [
92], 3) Central coherence: Group embedded figures test (GEFT) [
93], 4) Intolerance of uncertainty: Intolerance of Uncertainty Scale (IUS) [
94], 5) Perfectionism: Clinical Perfectionism Questionnaire (CPQ) [
95], 6) Self-efficacy: Perceived Efficacy in Patient-Physician Interactions (PEPPI-5) [
96], 7) Behavioural Inhibition and Behavioural Activation Scales (BISBAS) [
97], and 8) therapeutic engagement Session Rating Scale (SRS) [
98].
The Stroop-task was originally developed by Stroop [
89] to easily measure selective attention and cognitive flexibility. The Stroop Color-Word Test has three components. Patients are first asked to name a series of colored words to reflect basic reading rate. They are next asked to name the color (red, blue, or green) of a bar. The final component comprises a Color-Word task in which the participant is shown the names of colors printed in conflicting ink colors (e.g., the word “green” in blue ink) and is asked to name the color of the ink rather than the word. For all tasks, patients are timed by the tester to determine the number of seconds needed to identify all the stimuli, and, for the interference condition, the number of uncorrected errors is recorded. The ‘Stroop Effect’ is calculated by subtracting the difference between the interference and color patch task times, where lower scores indicate better executive functioning. Average test-retest reliability for the Stroop task is high (
r = 0.84) [
99]. In the present study we use the Color-Word interference test from the D-KEFS test-battery [
100].
The D-Flex [
90] is a 24-item self-report scale that measures cognitive rigidity (difficulty with set-shifting/flexibility) and attention to detail (weak central coherence). The two subscales showed high internal consistency (Cronbach α 0.90–0.95), construct validity (as compared to relevant subscales of the AQ) was strong for cognitive rigidity (
r = 0.72) but moderate for attention to detail (
r = 0.26).
The Trail Making Test (TMT) [
91] is used to evaluate set-shifting abilities. Originally a pen-and-paper test, more recently a computerized version has become available. Patients numerically or alphabetically connect circles on a page in a ‘ dot-to-dot ’ fashion (trail A), and then alternatively link numbers and letters, i.e. 1–A–2–B–3–C (trail B). Time taken to complete trail B (switching task) is the outcome measure of set-shifting ability.
When performing the computerized Stop Signal Task (SSRT) [
92] patients need to rapidly select corresponding motor responses to left- or right-facing arrows appearing on a computer screen and attempt to inhibit responses when an auditory “stop signal” sounds. Using a tracking algorithm, the task estimates the time taken to internally suppress prepotent motor responses (stop-signal reaction times).
The Group Embedded Figures Test (GEFT) [
101] is a paper-and-pencil task commonly used as a measure of central coherence. The time taken to find 18 simple shapes embedded in complex designs is recorded, with the main outcome being the total time needed to locate all hidden shapes, where shorter times are assumed to result from a strong local processing capacity or a bias toward detail.
The Intolerance of Uncertainty Scale (IUS) [
94,
102,
103] was developed to assess reactions to ambiguous situations, uncertainty, and future events. The IUS has been investigated and validated in various populations to measure intolerance of uncertainty. The IUS includes 27 items relating to the idea that uncertainty is unacceptable, reflects badly on a person, and leads to frustration, stress, and the inability to take action. Respondents rate items on a 5-point Likert scale ranging from 1 = “not at all characteristic of me” to 5 = “entirely characteristic of me”. The Dutch version has high internal consistency for anxiety-disordered patients (Cronbach α = 0.94) and a satisfactory test-retest reliability (
r = 0.79).
The Clinical Perfectionism Questionnaire (CPQ) assesses degrees of perfectionism. Respondents are presented with 12 items describing different expressions of perfectionism and asked to rate these on a 4-point Likert scale in relation to the last 28 days. Total score is the sum of all scores and range from 12 to 48, with higher scores indicating higher clinical perfectionism. There is evidence showing the CPQ to be a promising and reliable measure of clinical perfectionism. It has an acceptable internal consistency (Cronbach α = 0, 71) and test-retest reliability (
r = 0.49–0.67) [
95,
104].
With the Perceived Efficacy in Patient-Physician Interactions (PEPPI-5) [
96] patients rate their sense of confidence or self-efficacy in regard to their interactions with their physician on a 5-point Likert scale ranging from ‘not at all confident’ (1) to ‘very confident’ (5). The instrument has been translated and validated for use in the Netherlands [
105]. The range of possible scores is 5 to 25, with 25 reflecting highest-level self-efficacy. The PEPPI-5 has a high internal consistency (Cronbach alpha 0.92) and a fair test-retest reliability (intra-class correlation coefficient 0.68).
The Behavioural Inhibition and Behavioural Activation Scale (BISBAS) [
97] is a 20-item instrument designed to measure behavioural inhibition (i.e., concern over and reactivity to aversive events) and behavioural activation (i.e., responsiveness to incentives, drive, and fun seeking). Patients respond to the items using a 4-point scale, ranging from 1 (quite untrue) to 4 (quite true). After reverse scoring of two items on the BIS, subscale scores are computed by summing the scores corresponding to items on the BIS scale (7 items, range 7–28, α .74). The BAS-scale (13 items, range scores 13–52) consists of the reward responsiveness scale (5 items, α .73), the drive scale (4 items, α .76), and the Fun Seeking scale (4 items, α .66). Test-retest correlations over an 8 week period are satisfactory (r from 0.59 to 0.69). The BIS correlates moderately highly with measures of trait anxiety, negative affect, and negative temperament. The BAS correlates positively with extraversion, positive affect, and positive temperament.
The Session Rating Scale (SRS) [
98] is a short, 4-item, self-report questionnaire probing therapeutic alliance. After each session patients are asked to give their impressions of the preceding session using a VAS. They rate their bond with the therapist, agreement on therapy goals and therapy tasks, and their confidence in the collaboration, where ratings can range from “there was something missing in the session today” to “Overall, today’s session was right for me”. The Dutch translation of the SRS has been shown to have a strong internal consistency (Cronbach alpha 0.85–0.95) and adequate test-retest reliability [
106], although in the Dutch study correlations with outcome and concurrent validity were lower than expected.
Moderators
The following variables will be tested as potential moderators of intervention response: 1) demographic variables (e.g. sex, age, educational level), 2) general psychopathology: Structured Clinical Interview for DSM-IV-TR Axis I disorders (SCID-I) [
64], Structured Clinical Interview for DSM-IV-TR Axis II disorders, section obsessive-compulsive personality disorders (SCID-II-OCPD) [
68], 3) Autism: Autism-spectrum Quotient (AQ) [
107], 4) Attention-deficit hyperactivity disorder: Connors’ adult ADHD Rating Scales CAARS-S:SV) [
108], 5) Cognitive flexibility: Intra dimensional-/extra dimensional shift paradigm (ID/EDS) [
109], and 6) Planning ability: Behavioural Assessment of the Dysexecutive Syndrome, Zoo MAP (BADS-Zoo map) [
110]. The SSRT, IUS, and BADS-Zoo map will be taken for OCD only.
The autism-spectrum quotient (AQ) [
107] is a self-report questionnaire that assesses traits associated with the autistic spectrum [
111]. It consists of five 10-item subscales (social skills, attention switching, communication, imagination, attention to detail). Example items of the AQ are: “I find myself drawn more strongly to people than to things” and “Trying to imagine something, I find it easy to create a picture in my mind”. Respondent rate each item on a 4-point Likert scale from “definitely agree” to “definitely disagree”. A score of ≥ 32 is indicative of an autism spectrum disorder. The index’s test-retest reliability is good (
r = .60 to .81) and internal consistency is acceptable (α = .71) [
112].
The Connors’ Adult ADHD Rating Scales self-report, short version (CAARS-S:SV) [
108] comprises 30 items on the presence of ADHD symptoms to be rated on a 4-point Likert scale ranging from 0 (not at all), to 3 (very much). The maximum score for al 30 items is 90. The CAARS-S:SV consists of 5 subscales (inattention/memory problems, hyperactivity/restlessness, impulsivity/emotional lability, problems with self-concept, and ADHD-index). The internal consistency is good, ranging from α = .80 to .89 and it has an acceptable test-retest reliability (
r = 0.58 to 0.77).
The Intra-dimensional/extra-dimensional shift task (ID/EDS) [
109] is part of the Cambridge Neuropsychological Test Automated Battery (CANTAB). The version used in the current trial is a replication of the original task. The ID/EDS is a reliable and valid computerized task, which, like the TMT, probes set-shifting ability [
113‐
115]. The paradigm involves nine-stages of a visual discrimination task with multidimensional stimuli. Two stimuli are displayed, and feedback is provided which stimulus is correct. To pass to the next stage, six consecutive correct responses are required within 50 trials; if not, the task ends. The rule for correct responding is modified at the start of each task stage.
The Zoo Map is part of the Behavioural Assessment of the Dysexecutive Syndrome test battery (BADS) [
110] and consists of two parts. In the first part, patients are instructed to plan their route through a map of a zoo, visiting a selection of locations while actively disregarding others. While planning the route, patients also instructed to obey certain rules. The second part consists of the same map with the same locations that have to be visited, but this time, instructions are provided about the precise order in which the locations must be visited. Therefore, in contrast to the first part, the second part consists of a highly structured setting that strongly reduces the involvement of planning abilities. The BADS-Zoo map subtest has been proven to be a valid indicator of planning ability in a heterogeneous patient sample [
116].
In this trial, CRT is based on the Dutch translation of the original manual by Kate Tchanturia [
119] for both patient groups. Comprising 10 twice weekly 45-min sessions, CRT is delivered by psychologists, clinical nurses, and psychology students (at the MSc level), all having been trained by experienced CRT therapists (UD, BvP, LS, DC) who themselves were trained by Dr Kate Tchanturia, expert on CRT for eating and weight disorders. Using a range of cognitive lab-based exercises, CRT aims at improving patients’ set-shifting, cognitive flexibility, global information processing skills, and at reducing perfectionism. Table
2 lists the within-session exercises used in the RCT.
Central coherence tasks | Maps Summarize a letter Hidden words Word-search task Embedded word task |
Set shifting tasks | Card Stack Visual illusions Stroop material Switching attention task Switching Time Zones Alphabet task |
Perfectionism tasks | Estimation task |
Strategy tasks | How to plant a sunflower Complex figure task |
For a more detailed description of the exercises, see Tchanturia et al. [
119]. With respect to carrying out the exercises, awareness of dysfunctional thinking styles is enhanced. Patients are encouraged to find out how these thinking styles affect their daily lives, and from session 2 onward, homework assignments are given comprising behavioural exercises that are closely linked to real-life skills and tasks to thus stimulate patients to practice more flexible behaviours in their everyday lives.
Treatment as usual
After completion of either CRT or SAT, patients receive TAU. The content of TAU comprises all the essential elements recommended in the Dutch guidelines on eating disorders and OCD, which are closely related to the international guidelines [
121,
122].
TAU for patients with AN consists of normalization of eating behaviours (weight gain), discussion of daily problems, goals, psycho-education, and for a subgroup psychomotor therapy, art therapy, social skills training, family-therapy or cognitive behavioural therapy. Psychiatric- and/or medical consultation and pharmacotherapy is given where needed.
TAU for OCD patients comprises CBT delivered in 20 to 30 weekly or twice weekly 45–90 min sessions in which exposure with response prevention forms a key element, consistent with the multidisciplinary guidelines for anxiety disorders (OCD chapter) [
121]. In line with the literature [
123,
124], the OCD protocol is flexible, i.e., patient-tailored with regard to the time spent on each component, accounting for differences in session and treatment duration. The therapy encompasses a) psychoeducation, b) cognitive therapy, c) exposure in vivo with response prevention, d) psychiatric consultation, and, in case of severe symptoms, pharmacotherapy with SSRIs, venlafaxine, clomipramine and/or atypical antipsychotics. The protocol allows for one or two sessions in which the principles of CBT are explained and one or two sessions in which an inventory is made of all anxiety/tension-provoking situations that give rise to compulsions and a construction of a list of compulsion triggers. All patients receive in- and between session exposure to anxiety-provoking thoughts and situations, coupled with prolonged response prevention tailored to their individual symptoms.
Analyses
Baseline differences between the CRT and SAT (control) groups and between the participating centres are investigated using χ2 tests for categorical variables and ANOVAs for continuous variables. Additional exploratory analyses are conducted separately for AN and OCD.
The data for the primary AN and OCD outcome indices are analysed using linear mixed models (LMM) with the baseline value as a covariate. The mixed models analysing the general efficacy of CRT include a random term for the intercept and fixed terms for condition, time contrasts (T0–T1, T0–T2, T0–T3, T1–T2, T1–T3), and the interactions between condition and time contrasts.
In order to investigate moderators of short- and long-term treatment response, time contrasts are created (T0–T1, T0–T2, T0–T3, T1–T2, T1–T3) by means of dummy coding. Potential continuous moderator variables are standardized. Interactions between condition and the potential moderators are entered into the equation. All analyses on therapy effect modificators are of an exploratory nature. The moderator and mediator analyses are calculated with 5000 bootstraps [
126].
To assess the cost-effectiveness of CRT + TAU versus SAT + TAU for ED and OCD, all relevant costs and effects are taken into account. Disease progression is described in terms of transitions between ‘states’, where a patient’s ‘state’ can shift in either direction or remain constant. Estimations of state-transition rates are based on the data from the ongoing trial and the literature, which modelling approach serves to help us assess future scenarios. Costs and effects are calculated in accordance with recommendations in the Dutch guideline for economic evaluation of health care. As a primary assessment we apply a costs-utility analysis that yields a cost per quality-adjusted life years (QALY) outcome, a common cost-effectiveness measure that allows comparison with economic evaluations of other diseases and interventions. The uncertainty estimations are computed using bootstrapping and the results expressed in an acceptability curve. If no intervention is effective, a cost-minimization calculation is performed.