Effectiveness of Acceptance and Commitment Therapy on Obsessive Beliefs Among Patients With Obsessive-Compulsive Disorder
Article information
Abstract
Objective
Both acceptance and commitment therapy (ACT) and metacognitive therapy (MCT) for obsessive-compulsive disorder (OCD) target metacognition. In this context, ACT’s therapeutic effects on OCD are presumably reflected in the Obsessive Belief Questionnaire (OBQ), which incorporates significant metacognitive factors. However, most studies have investigated ACT’s effects on OCD symptoms rather than on obsessive beliefs.
Methods
This study examined the relationship between ACT process measures—Acceptance and Action Questionnaire-II (AAQ-II) and Cognitive Fusion Questionnaire (CFQ)—and the OBQ following an 8-week group-format ACT (GACT) intervention among patients with OCD (GACT group, n=37; wait-list control group, n=42).
Results
Significant reductions were observed in AAQ-II, CFQ, and OBQ scores after treatment. Changes in AAQ-II and CFQ (particularly in the former) were significantly associated with changes in the OBQ total and subscale scores. These measures accounted for 26% and 12% of the variance in the OBQ total score and OCD symptom scores, respectively. Among OBQ subscales, changes in AAQ-II and CFQ best accounted for the variance of perfectionism and intolerance of uncertainty (OBQ-PU), followed by importance and control of thoughts (OBQ-ICT).
Conclusion
Improvements in ACT process measures through GACT were more strongly linked to obsessive beliefs than to OCD symptoms, highlighting the role of obsessive beliefs—particularly OBQ-PU and OBQ-ICT—in ACT’s effectiveness for OCD.
INTRODUCTION
Obsessive-compulsive disorder (OCD)—typically a chronic and debilitating condition—is characterized by intrusive thoughts, images, or urges that cause anxiety, followed by compulsive behaviors or mental acts performed to alleviate it [1]. Currently, exposure and response prevention (ERP), with or without cognitive therapy, is considered the most effective psychological treatment for OCD [2-4]. However, ERP poses several practical challenges, including limited availability, significant time commitment, patient reluctance, and therapist-related barriers [5]. Consequently, clinicians and researchers have been exploring alternative treatments to address these concerns and enhance ERP’s efficacy.
Wells and Matthews [6] developed the first explicitly metacognitive model of OCD. According to metacognitive therapy (MCT), thoughts can be experienced in two ways: the fused, object mode, wherein thoughts are indistinguishable from direct experiences of the self or the world; the metacognitive mode, wherein thoughts can be observed as separate events [7]. MCT emphasizes awareness of metacognitive beliefs—how one relates to and experiences thoughts and feelings. Key beliefs in OCD include fusion beliefs concerning the power and importance of thoughts and beliefs concerning the need to perform rituals to control thoughts and prevent perceived danger [8]. Detached mindfulness, the essential MCT skill, helps individuals relate more adaptively to intrusive thoughts [9].
MCT perspectives on OCD informed the development of the Obsessive Belief Questionnaire (OBQ) [10]. Among its six beliefs across three domains, importance and control of thoughts corresponds to the typical metacognitive processes, while intolerance of uncertainty is a metacognitive factor where individuals doubt their own thinking and feel the need to control or eliminate uncertainty, often leading to maladaptive coping strategies like avoidance or rumination.
Over the past decade, acceptance and commitment therapy (ACT) has emerged as a promising mindfulness-based approach for OCD [11-13]. Rooted in relational frame theory, ACT emphasizes how linguistic relational frames influence behavior, leading to cognitive fusion, whereby literal content becomes intertwined with internal experiences (i.e., thoughts and feelings). Excessive cognitive fusion can trigger experiential avoidance in OCD, typically expressed through compulsions or avoidance [14,15]. ACT focuses on non-judgmental recognition of thoughts—cognitive defusion—to change one’s relationship with them and promote value-driven actions [16].
In sum, although theoretically distinct, both MCT and ACT share a focus on metacognition and the relationship with inner experiences that drive unhelpful thinking (e.g., rumination) and coping (e.g., experiential avoidance and compulsion), with mindfulness as a core therapeutic component. From this perspective, ACT’s effects are likely reflected in the OBQ, yet few studies have examined its impact on OBQ scores, as most focus on obsessive-compulsive (OC) symptoms.
To the best of our knowledge, only one study has observed changes in the OBQ after ACT treatment. However, it included both ERP and ERP+ACT groups, rendering it difficult to attribute the effects solely to ACT; the ERP+ACT group exhibited a decrease from 195 to 138 points in the OBQ after 8 weeks of treatment [17]. Nevertheless, it presented only the total score changes and did not provide detailed findings for the OBQ, such as correlations with ACT process scores or regression analysis results. One MCT study noted a significant reduction in the OBQ score, indicating an improvement (Cohen d=0.53 from pre- to post-treatment, 0.96 from pre-treatment to follow-up) [18], but also did not present additional details. Moreover, some cross-sectional studies have assessed cognitive fusion and experiential avoidance, two key psychopathologies of ACT, along with the OBQ. Nonetheless, these studies primarily aimed to examine the extent to which cognitive fusion or experiential avoidance, alongside the OBQ, explain OC symptoms or symptom dimensions, instead of directly exploring the relationships between cognitive fusion, experiential avoidance, and the OBQ [19-22].
Therefore, this study aims to investigate the relationships between ACT process measures and the OBQ following an 8-week group-format ACT (GACT) intervention among patients with OCD—randomly assigned to a treatment group or control group. Specifically, the study aims to test the following hypotheses: 1) changes in ACT process measures are associated with changes in the OBQ; 2) changes in process measures partially account for changes in the OBQ and its subdomains; 3) if process measures effectively account for changes in the OBQ, a specific OBQ subdomain is optimally accounted for by these measures; and 4) process measures account for changes in the OBQ more effectively than those in OC symptoms.
METHODS
Participants
This study assessed patients with OCD recruited between 2018 and 2024 from psychiatric clinics at Kyungpook National University Hospital and via subway advertisements. An experienced psychiatrist conducted patients’ clinical evaluations based on the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition. Patients with current comorbid psychiatric disorders (e.g., major depressive disorder, bipolar and related disorders, schizophrenia spectrum and other psychotic disorders, substance use disorders, or intellectual disabilities), as well as those with a history of head injury, neurological disorders, or acute medical conditions, were excluded. After conducting clinical evaluations and obtaining written informed consent, 93 patients were enrolled and randomly assigned to either the ACT group (i.e., GACT) or the wait-list control group (i.e., WLC). Of these, 7, 6, and 1 patients dropped out because of academic or work-related issues, a lack of motivation, and health problems, respectively. Participants were permitted to continue their usual pharmacotherapy. The final analysis included data from 79 patients (37 females; mean age±standard deviation=27.8±8.1 years) who attended at least six sessions and completed the pre-post assessments. Table 1 presents the patients’ demographic information. Data from these participants have been partially published in our previous study [23,24]. This study was approved by the Institutional Review Board of Kyungpook National University Hospital (approval number: 2021-04-032).
Baseline demographic and clinical characteristics of patients with obsessive-compulsive disorder
GACT program for OCD
An 8-week GACT program [25]—comprising 90-minute weekly sessions—was conducted. Each GACT group comprised 3–5 participants. Participants received free treatment. Session 1 provided an overview of OC symptoms and elucidated how compulsions exacerbate them; for example, intolerance of uncertainty was highlighted as a key factor in maintaining OC symptoms. Session 2 introduced the concept of creative hopelessness and explored the relationship between cognitive fusion and experiential avoidance. Sessions 3 and 4 introduced techniques for cognitive defusion, whereby therapists guided participants through mindfulness breathing training, which continued from Sessions 3 to 7. Sessions 5 and 6 focused on how increasing willingness fosters the acceptance of distressing emotions triggered by OC symptoms. Therapists encouraged participants to embrace their internal experiences associated with obsessive thoughts without engaging in compulsive behaviors. Session 7 emphasized the significance of personal values and helped participants identify their core values and set meaningful goals aligned with them. Finally, Session 8 reviewed the content covered throughout the program and offered guidance on relapse prevention. Our previous paper describes the GACT protocol in detail [23,24]. Two psychiatrists (S.J.L. and S.W.L.) and one clinical psychologist (M.C.) participated as therapists.
Psychological measures
All participants, regardless of their assignment to the GACT or WLC group, completed a baseline psychological assessment, which was repeated after 8 weeks.
Obsessive Belief Questionnaire-44
The Obsessive Belief Questionnaire-44 (OBQ-44)—a 44-items self-report questionnaire [26]—assesses dysfunctional beliefs, which are key factors in OCD development and maintenance. It comprises the following subscales: 1) responsibility and threat estimation (OBQ-RT), 2) importance and control of thoughts (OBQ-ICT), and 3) perfectionism and intolerance of uncertainty (OBQ-PU). Responses are given on a 7-point Likert (ranging from 1=disagree very much to 7=agree very much). A higher total score indicates greater agreement with OCD-related dysfunctional beliefs. This study used the Korean translation (Cronbach’s α=0.88) [27]. In the present sample, Cronbach’s α was 0.95.
Yale-Brown Obsessive-Compulsive-Scale
The severity and degree of change in OC symptoms were assessed using the self-report version of the Yale-Brown Obsessive-Compulsive-Scale (Y-BOCS) [28]. It comprises 10 items (items 1–5 for obsessions and items 6–10 for compulsions), and each item is rated on a 5-point Likert scale ranging from 0 to 4. A higher value indicates greater distress owing to OC symptoms. The Korean self-report version of the Y-BOCS [29] was employed, and the internal consistency for the clinical and non-clinical groups in their study was 0.87 and 0.89, respectively. In the present sample, Cronbach’s α was 0.81.
Dimensional Obsessive-Compulsive Scale
The Dimensional Obsessive-Compulsive Scale (DOCS)—a 20-item self-rating scale—classifies OC-related symptoms into the following four symptom dimensions: 1) contamination; 2) responsibility for harm; 3) unacceptable thoughts; and 4) symmetry and ordering [30]. Each dimension is rated on a 5-point Likert scale ranging from 0 to 4, which has been most consistently replicated in OC symptom dimensions. This scale’s total score ranges from 0 to 80, and its subscale scores range from 0 to 20. A standardized Korean version of the DOCS was employed; its psychometric properties were found to be acceptable (Cronbach’s α=0.91–0.95) [31]. In the present sample, Cronbach’s α was 0.90.
Acceptance and Action Questionnaire-II
The Acceptance and Action Questionnaire-II (AAQ-II) is a self-report measure used to assess experiential avoidance and psychological inflexibility—central to the ACT model’s etiology and therapeutic approach [32]. During its development, 10 items loading onto a single factor were initially extracted; however, the final version included 7 items. Each item is scored on a scale ranging from 1 (not at all) to 7 (always), with higher scores indicating greater psychological inflexibility. We employed the AAQ-II’s Korean version standardized by Heo et al. [33], which comprised 8 items after excluding 2 of the original 10 items of the short-form AAQ. The scale exhibited an internal consistency coefficient of 0.85. In the present sample, Cronbach’s α was 0.88.
Cognitive Fusion Questionnaire
The Cognitive Fusion Questionnaire (CFQ)—a 7-item self-report scale—assesses the degree of cognitive fusion and defusion, as conceptualized in ACT [34]. It employs a 7-point Likert scale (1=not at all true, 7=always true), with higher scores indicating a greater cognitive fusion level. In a large-sample study involving over 1,800 participants from both clinical and non-clinical populations experiencing diverse mental health issues and stresses, the CFQ demonstrated satisfactory psychometric properties (α=0.88–0.93). We utilized the scale adapted and standardized by Kim and Cho [35]. In this sample, the internal consistency was 0.91 and 0.94 for the nonclinical and clinical samples, respectively. In the present sample, Cronbach’s α was 0.92.
Beck Depression Inventory-II
The Beck Depression Inventory-II (BDI-II)—a 21-item self-report scale—measures depressed mood and assesses the severity and type of depressive symptoms [36]. The items are categorized into cognitive, emotional, and physical factors and rated on a Likert scale ranging from 0 to 3, with higher scores indicating more severe depression. We employed the validated Korean version of the scale for adult samples [37], which demonstrated an internal consistency of 0.89. In the present sample, Cronbach’s α was 0.93.
Statistical analyses
Group differences in demographic characteristics—including age; sex; education; age at OCD onset; illness duration; OC symptoms and obsessive beliefs; and ACT process—were assessed using a t-test and chi-square test for continuous and categorical variables, respectively. Treatment effects on changes in symptom and process measures over time were examined using a 2 (Condition; GACT, WLC)×2 (Time; pre-treatment, post-treatment) mixed-design generalized linear model. Condition and Time were between- and within-subject factors, respectively. The dependent variables, consistent with this study’s primary objective, included the OBQ, as well as the OC symptom and ACT process measures. Between- and within-group effect sizes, expressed as Cohen’s d using the independent and paired t-tests for between- and within-group differences, respectively. Furthermore, to clarify process factors associated with ACT’s effectiveness on obsessive beliefs, correlation and multiple regression analyses were employed to evaluate the relationship between changes in obsessive beliefs and process measures. Data analyses were performed using IBM SPSS Statistics for Windows (version 23.0; IBM Corp.).
RESULTS
Sample characteristics
Table 1 summarizes the participants’ baseline demographic and clinical characteristics. The mean (±standard deviation) age of the GACT and WLC groups was 27.3±8.3 and 28.3±8.0 years, respectively. The group exhibited no significant differences in age, sex, and educational level. The mean age of OCD onset was approximately 20 years, while the mean illness duration was approximately 7 years, with no group differences. Participants’ Y-BOCS scores indicated that their OC symptoms generally ranged from moderate to severe. Overall, the group exhibited no baseline differences on any psychological measure.
Overall, 21 (56.8%) and 27 (64.3%) participants in the GACT and WLC groups, respectively, continued with their usual medication during the study, with no significant between-group differences in medication frequency (χ2=1.3, p=0.53). All patients receiving medication used at least one selective serotonin reuptake inhibitor or clomipramine; among them, 16 patients combined it with a small dose of antipsychotic drugs. Table 2 presents detailed information regarding the medications.
Medication status of participants
Treatment outcomes
Table 3 presents the changes in psychological assessment scores from pre- to post-treatment for the GACT and WLC groups, along with the Condition×Time interaction and effect size. Regarding ACT process measures, significant interaction effects were observed between Condition and Time (F1,77=35.8, p<0.001 for AAQ-II; F1,77=48.2, p<0.001 for CFQ). Compared to the WLC group, the GACT group exhibited significant post-treatment reductions in AAQ-II (d=0.68) and CFQ (d=1.00) scores.
Changes in obsessive-compulsive symptoms, beliefs, and ACT processes at pre- and post-treatment
Further, both the total OBQ score and three OBQ subscale scores exhibited similar patterns vis-à-vis the process measures. A significant Condition×Time interaction was observed for the OBQ total score (F1,77=28.6, p<0.001). Compared to the WLC group, the GACT group exhibited significant posttreatment reductions in the OBQ total score and subscale scores (d ranging from 0.54 to 0.89). Additionally, compared to the WLC group, the GACT group exhibited a significant post-treatment improvement in OC symptoms, as measured by the Y-BOCS and DOCS.
Relationship between changes in ACT processes and obsessive beliefs following ACT intervention
Table 4 presents zero-order Pearson correlation coefficients among all study measures, indicating considerable variation in the strength of associations between variables; the correlation coefficients ranged from 0.93 (OBQ total with OBQ-RT) to 0.25 (DOCS with OBQ-RT). We applied a Benjamini–Hochberg procedure to control for the false discovery rate [38], resulting in a corrected significance threshold of p<0.022.
Correlations between changes in ACT processes, obsessive-compulsive symptoms and beliefs within the GACT group (N=37)
Table 4 suggests that changes in the AAQ-II were significantly correlated with reductions in the OBQ total score (r=0.49, p=0.002) (Figure 1) and its subscale scores—specifically, OBQ-PU (r=0.53, p=0.001), OBQ-ICT (r=0.42, p=0.011), and OBQ-RT (r=0.38, p=0.021). Likewise, changes in the CFQ were associated with changes in the OBQ total score (r=0.39, p=0.016) and OBQ-PU (r=0.44, p=0.006).
Illustration of correlations between changes in AAQ-II and OBQ-44 (A), and between AAQ-II and YBOCS (B) scores. Full correlation coefficients are presented in Table 4. AAQ-II, Acceptance and Action Questionnaire-II; CFQ, Cognitive Fusion Questionnaire; Y-BOCS, Yale-Brown Obsessive-Compulsive Scale; OBQ-44, Obsessive Belief Questionnaire-44 items.
Regression analysis predicting obsessive beliefs and OC symptoms
Multiple regression analysis revealed that changes in the AAQ-II and CFQ collectively accounted for 26% of the variance in the OBQ total score. Notably, AAQ-II emerged as the only significant predictor in this model (Table 5). In contrast, changes in the AAQ-II and CFQ significantly accounted for 17% of the variance in the DOCS total scores; however, neither variable was a significant predictor. Meanwhile, changes in these measures did not significantly account for the variance in the Y-BOCS.
Multiple regression analysis (enter method) for explaining changes in obsessive belief and obsessive-compulsive symptoms using ACT process measures
Regression analysis predicting individual obsessive belief dimensions
Changes in the AAQ-II and CFQ significantly accounted for 19% and 31% of the variance in the OBQ-ICT and OBQ-PU obsessive belief dimensions, respectively (Table 6). By contrast, these variables accounted for only a small, non-significant portion of variance in the OBQ-RT.
Multiple regression analysis (enter method) for predicting individual obsessive belief dimensions
DISCUSSION
This study investigated the relationship between ACT process measures and the OBQ following an ACT intervention among patients with OCD. Its primary findings are as follows: First, along with significant reductions in process measures and OBQ after treatment, changes in AAQ-II and CFQ, particularly the AAQ-II, were significantly associated with changes in the OBQ total score and its three subscale scores. Second, changes in the AAQ-II and CFQ significantly accounted for 26% of the variance of changes in the OBQ total score but slightly less in the Y-BOCS and DOCS total scores—specifically, 12% and 17%, respectively. Third, among the three OBQ subscales, changes in the AAQ-II and CFQ best accounted for the variance of OBQ-PU, followed by OBQ-ICT. However, they did not significantly account for the variance in OBQ-RT.
To the best of our knowledge, this is the first report to investigate the detailed relationship between changes in process measures and obsessive beliefs using RCT data. First, following ACT intervention, the OBQ-44 total score decreased from 190.2±38.9 to 150.3±44.8 in this study. This amount of change is consistent with the only previous RCT using the same OBQ-44, although either ERP or ACT+ERP was applied [17]. However, in that study, the authors did not explicitly mention the OBQ’s results. On the contrary, several studies using mindfulness-based cognitive therapy—another mindfulness-based therapy similar to ACT—have noted that the reduction in obsessive beliefs occurs alongside a decrease in OC symptoms [39-42]. Didonna et al. [43] proposed that mindfulness-based therapy may achieve greater therapeutic success by modifying higher–order cognitive processes, such as beliefs regarding the importance and the power of thoughts instead of focusing on lower–order appraisals.
In this study, changes in the OBQ-44 exhibited moderate positive correlations with those in the AAQ-II (r=0.49) and CFQ (r=0.39). Notably, although two correlation coefficients did not differ statistically, the AAQ-II exhibited a stronger correlation with changes in the OBQ than the CFQ. Moreover, in the regression analysis, the significant predictor of the OBQ was the AAQ-II, not the CFQ. These results corroborate our previous cross-sectional study’s finding that the AAQ-II accounts for specific OC symptom dimension better than the CFQ [44]. This is possibly because it was revised from a tool designed to assess experiential avoidance into one that evaluates the broader construct of psychological inflexibility [32,45,46]. Additionally, the AAQ-II items are designed to assess the impact of memories, emotions, and experiences on daily life, the pursuit of life goals, and overall functioning. This revision may have increased its likelihood to capture a broader range of experiences among OCD patients than the CFQ, whose questions pertain to a single theme.
Moreover, this study demonstrated that changes in the AAQ-II and CFQ significantly accounted for the variance in changes in the ICT and PU, but not those in the RT obsessive belief dimensions. Interestingly, this observation aligns with a previous cross-sectional study reporting the relationships between AAQ, CFQ, and OBQ [22]. Notably, this study is significant because it demonstrated a consistent association in the changes observed following ACT treatment, thereby transcending simplistic cross-sectional correlations. First, the finding that PU exhibited greater explanatory power than ICT in relation to AAQ-II and CFQ was unanticipated. The symmetry and ordering OCD symptom dimension, which PU has been demonstrated to significantly predict [47], is frequently characterized by “not just right experiences.” This manifestation fundamentally implies an ACT-based perspective, whereby unwanted internal experiences are judged as problematic and having undesirable consequences—specifically, cognitive fusion or psychological inflexibility. In this vein, one study demonstrated that the emotion regulation and negative evaluation domains of cognitive fusion were strongly associated with symmetry and ordering OCD symptom dimension [20], Second, changes in OBQ-ICT were significantly accounted for by the AAQ-II and CFQ—consistent with a prior cross-sectional study reporting the strong correlations between OBQ-ICT and the AAQ-II and CFQ among patients with OCD [22]. Third, changes in AAQ and CFQ did not account for changes in OBQ-RT. Unlike OBQ-ICT and OBQ-PU, responsibility, identified by Salkovskis [48] as central to OCD’s etiology, is essentially a non-metacognitive belief. Furthermore, Wells argued that responsibility is a by-product of metacognitions, contributing little to nothing to explaining OCD [8,49]. Thus, we attributed this negative result to the distinct categories of metacognitive and cognitive beliefs.
This study has several limitations. First, as approximately 60% of the participants continued their usual pharmacological treatment, concluding that the treatment outcomes were solely attributable to psychotherapy is challenging. However, 40% of the patients did not take any medication during the treatment period, and a detailed comparison of medication information revealed no significant difference in drug use between the two groups. Therefore, considering that the difference in effectiveness between the groups is attributable to psychotherapy is reasonable. Second, slight differences in participant number and gender distribution between the groups, due to additional inclusion criteria for the GACT group, along with the sample being drawn from a single center and consisting primarily of young adults, may limit the generalizability of the findings. Third, the study did not include long-term follow-up assessments, limiting conclusions about the durability of the observed effects.
In conclusion, this study demonstrated that improvements in ACT process measures through GACT intervention explained obsessive beliefs—particularly OBQ-ICT and OBQ-PU, which comprise metacognitive beliefs—better than OC symptoms. These findings underscore the clinical implications of obsessive beliefs in ACT treatment for OCD.
Notes
Availability of Data and Material
The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.
Conflicts of Interest
Seung Jae Lee, a contributing editor of the Psychiatry Investigation, was not involved in the editorial evaluation or decision to publish this article. All remaining authors have declared no conflicts of interest.
Author Contributions
Conceptualization: all authors. Data curation: all authors. Formal analysis: all authors. Writing—original draft: all authors. Writing—review & editing: all authors.
Funding Statement
None
Acknowledgments
None