Session-specific effects of the Metacognitive Group Training for Obsessive–Compulsive Disorder: significant results for thought control

The investigation of the session-specific effects is central for the understanding of psychological interventions. For the present study, we investigated the session-specific effects of the Metacognitive Group Training for Obsessive–Compulsive Disorder (MCT-OCD), which was revised based on data of a pilot study. Thirty-four outpatients with OCD participated in the MCT-OCD once a week over 8 weeks. Different metacognitive beliefs (e.g., thought control) and cognitive beliefs (e.g., intolerance of uncertainty), OC symptoms, as well as associated comorbid symptoms were assessed before and after each session. Linear mixed effects models showed that patients’ obsessions and compulsions, thought control, the belief of being well informed about the disorder, and action fusion improved over the course of the training. The only session-specific effect emerged for thought control, which improved immediately after the respective module. We were able to replicate the findings of the pilot study and thus corroborate the session-specific effect of the module targeting thought control. Moreover, we generated information on the mode of action of the individual modules of the MCT-OCD that allows a more in-depth evaluation of the intervention. Notably, we were able to eliminate the adverse effects of the pilot version of the MCT-OCD. Trial Registration: German Clinical Trials Register (Deutsches Register Klinischer Studien [DRKS]; DRKS-ID: DRKS00013539; registration date: 22/02/2018).


Results
See Table 1 for the demographic and psychopathological data of the sample. Patients participated in six modules on average (M = 5.82, SD = 1.66), with a range from two to eight. The main current comorbid disorders as verified by the M.I.N.I. were depression (17.8%) and anxiety disorders (agoraphobia: 17.6%; generalized anxiety disorder: 14.7%; social anxiety disorder: 11.8%; panic disorder: 8.8%).
For the treatment effect over time the random intercept model was superior for the majority of the variables, with the exception of both variables on the belief of being well informed about the disorder, reduced intolerance of uncertainty 2, perfectionism 2 (see Supplementary material A for a detailed description of all items), both variables on action fusion, overestimation of threat 1, inflated sense of responsibility 2, and biased attention. For those models, the random intercept, random slope model was a better fit. The random intercept model showed an improvement in thought control (b = − 0. 10 For most within-session analyses, the random intercept model was used (see Table 2). In accordance with our hypotheses, the calculations demonstrated that patients' thought control reduced the most after module #5 (control of thoughts; with a small to medium effect of d = 0.42) compared to the average score on thought control after all other modules. The change in all other variables remained insignificant (see Table 2).
For most between-session analyses, the random intercept model was used (see Table 3). Patients' intolerance of uncertainty 2 worsened more one week after module #1 (false assumptions about OCD; with a medium effect size of d = 0.52) compared to the average score on intolerance of uncertainty 2 after all other modules. All other variables remained insignificant (see Table 3).

Discussion
Based on the results of the pilot study 14 , we aimed to investigate the session-specific effects of the revised metacognitive training for patients with OCD (MCT-OCD) using expanded pre-and post-questionnaires. In accordance with the results of the uncontrolled pilot study on the session-specific effects of the MCT-OCD 14 and our hypotheses, patients' thought control, obsessions, and compulsions improved over the duration of the treatment.
In line with our hypotheses, patients' belief of being well informed about the disorder and patients' degree of action fusion also improved over treatment. The other (meta)cognitive beliefs and associated comorbid symptoms did not improve over the duration of the treatment.
Regarding session-specific effects, we were able to replicate and specify the results of the pilot study 14 showing that patients' control of thoughts improved most after the relevant module (module #5). As the pilot version of the MCT-OCD comprised two (meta)cognitive beliefs per module (e.g., thought-action fusion/control of thoughts), this was not self-evident. Hence, the present study increased the clarity and strengthened the validity of the finding. However, none of the other modules improved the specific (meta)cognitive belief that was targeted in the respective module. Several potential causes may have contributed to this result: (1) the items did not assess the (meta)cognitive belief adequately and thus need to be revised and extended, (2) the modules did not address the (meta)cognitive beliefs as specifically as necessary and need to be revised, (3) it takes longer for certain (meta)cognitive beliefs to change, or (4) the (meta)cognitive beliefs that are addressed by the modules did not change the respective (meta)cognitive belief. The reasons for the insignificant within-session findings of the present study cannot fully be determined on the basis of our data. Future studies need to assess variables apart from the (meta)cognitive beliefs that are targeted in the MCT-OCD, in order to identify the mode of action of all modules in greater depth. In contrast to the pilot study showing that patients' mood worsened after the module on overestimation of threat/exaggerated sense of responsibility, patients' mood did not worsen after any of the revised modules in the present study. Thus, the revision of the MCT-OCD can be regarded as successful in respect to the elimination of adverse effects.
Adverse effects of between-session effects of the pilot study (i.e., intensification of compulsions) were not evident in the revised MCT-OCD. This can be regarded as a further improvement. However, between-session changes contradicted our prior assumptions because no hypothesis-conformant result was found. Surprisingly, patients' intolerance of uncertainty worsened (i.e., increased) the most one week after module #1 (false assumptions about OCD) compared to all other modules. Given that the module has a psychoeducational character, it might be assumed that patients developed a certain clarity about their symptoms and concluded that they were now more certain about particular concerns (e.g., the identification of their OC symptoms), and they therefore scored higher on the item that asked for confidence in choices. Thus, the module might either be slightly revised by conveying that some uncertainty will always remain and encourage patient acceptance of this, or the item that assessed this metacognition should be revised in order to assess it more precisely. However, the finding needs to be replicated in future studies in order to draw more valid conclusions.
The absence of other between-session effects is in line with our pilot study 14 , which did not show hypothesesconformant results for between-session effects. Between-session effects could have been expected if the patients applied and practiced what they had learned from the session between sessions. However, the clinical impression was that patients often did not apply and practice what they had learned, which may have significantly reduced the between-session effects.
In addition to the several strengths of the present study (e.g., the specific analyses of session-specific effects in contrast to the overall efficacy of the MCT-OCD, secured diagnoses by trained raters in in-person assessments, revised intervention, and pre-and post-questionnaires), some limitations need to be acknowledged. First, the Table 1. Demographic and psychopathological data: mean (M) and standard deviation (SD) and frequency (n) and percent (%). HDRS Hamilton Depression Rating Scale, Y-BOCS Yale-Brown Obsessive Compulsive Scale. www.nature.com/scientificreports/ study and the therapy were conducted by the developers of the MCT-OCD (LJ, BH, FM), leading to a potential allegiance effect. Second, the between-session analyses may have been influenced by other concurrent outpatient therapies because patients were allowed to continue their treatment as usual. Third, the main analyses relied on self-report, which has several advantages (e.g., anonymity and thus possibly more honest feedback) but also comes with the disadvantage that items may have been misunderstood as well as the general disadvantages of self-report 15 . Forth, we did not explore whether changes in beliefs during the MCT-OCD are predictive of treatment outcome. This was not an aim of the current study, but it should be investigated in future studies. Fifth, no a priori power analysis was conducted for the present study. Sixth, single items were analysed in the present study and no quality criteria of the pre-and post-session questionnaires are available, which limits conclusions.

Conclusions
The results of the present study could be used to inform another revision of the MCT-OCD. Future revisions could include revising module #1 (false assumptions about OCD) so that the increased intolerance of uncertainty one week after module #1 would be eliminated (see discussion on between-session effects). Notably, we now know that module #5 (control of thoughts) specifically targets patients' need to control their thoughts (replicated result) 14 . Thus, the module can be considered successful in improving the need to control thoughts. Moreover, this module might be used not only within the scope of the MCT-OCD but also in individual therapy for patients who excessively control their thoughts. As mentioned above, homework should become mandatory   www.nature.com/scientificreports/ in the MCT-OCD in order to strengthen the between-session effects as well as the transfer of the learning into patients' everyday life. For future studies, the pre-and post-questionnaires need to be revised to include more items on other potential mechanisms of change in the different modules instead of only those items that are related to the topics of the modules. Based on the results of the present study, most modules (except for module #5) do not specifically improve the specific (meta)cognitive belief (e.g., perfectionism) that is targeted in the module but might act through something different (e.g., self-esteem). This needs to be explored in future studies. Based on our results, it is apparent that research on session-specific effects can be very fruitful for the evaluation of a therapy and the design of its revision as well as for the determination of the specific effects of the intervention or certain components of it. Using session-specific analyses instead or on top of the evaluation of the overall efficacy of an intervention can help to optimize therapies.

Materials and methods
Design. We conducted a randomized wait-list controlled trial. The study design (study protocol) of main trial have already been reported 15 . After baseline assessment (blinded rater), patients were randomized to participate either in MCT-OCD or in the wait-list control group and were assessed after eight weeks (post assessment). Both groups had access to care as usual. Informed consent was obtained from all patients, and the study was conducted in accordance with the Declaration of Helsinki. Sample. Patients were screened for eligibility through a telephone interview. If they fulfilled the inclusion criteria, they were invited for a baseline in-person assessment and asked to complete a range of questionnaires. Patients were included if they were between 18 and 70 years of age, were willing to participate in the MCT-OCD, were suitable for outpatient group therapy (i.e., no acute suicidality), and had sufficient German language comprehension. Patients with current or lifetime psychotic symptoms, mania, a severe neurological disease, and/or current substance dependence were excluded. Due to recruitment difficulties, we broadened the inclusion and    www.nature.com/scientificreports/ exclusion criteria to also include patients with a lifetime mood disorder with psychotic features, a lifetime high substance use disorder, or a neurological disease not associated with OC symptoms). For the present study, only patients who were randomized to the MCT-OCD, who participated in at least one MCT-OCD module, and who filled out at least one pre-and one post-session questionnaire were included in the analyses, resulting in a total sample of N = 34 for the present study.
Metacognitive training. A psychotherapist in training (master's degree in psychology) and a psychologist with a bachelor's degree carried out the MCT-OCD. The sessions lasted approximately 90 min and were administered once a week over a period of 8 weeks. The group size was allowed to range between three and ten patients. The variation in the number of participating patients arose due to the open group format (i.e., patients were able to enter and drop of the group at any time). In the patient's first session, they received a booklet with exercises for each session (designed to facilitate the transfer of the intervention into everyday life) and summaries of the most important content of the modules (exercises). Strategy of data analysis. LMMs were calculated to analyze the main research questions of the present study due to the hierarchical structure of the data. Three different sets of calculations were run in order to identify (1) the amount of within-session changes for each module and each dependent variable, (2) the treatment effect over time for each dependent variable, and (3) the magnitude of between-session changes for each module. All analyses were run with all 19 dependent variables and thus all items of the pre-and post-session questionnaires. The LMMs consisted of two levels: (1) repeated measures (pre-and post-session questionnaires) and (2) between-subject factor (i.e., the patients). For all analyses, a random intercept model and a random intercept, random slope model were calculated, and ANOVAs compared the two models in order to choose the model that provided the best fit. The decision on model fit was based on the Akaike information criterion (AIC): A significantly lower AIC indicated a better model fit 19 . If the random intercept, random slope model was used, it indicated a significant variability in the patients' level of the specific variable at the start of the session and in the change within or between sessions. We decided to also test random slopes for the analyzed models by allowing the respective effect to vary between patients. For the approximation of parameters, the maximum-likelihood method was used 20 , and the beta values (B) represent the estimated effect sizes. In particular, B displays the magnitude of standard deviations of the changes due to the module under consideration compared to all other modules 21 . See Supplementary material B for the equations for all calculations and detailed information of the analyses and Supplementary material C for the AICs of each model. This strategy of data analysis was also used in two previous studies 14,22 . As exploratory analyses, we computed whether any other module had a stronger effect on the specific items compared to all the other modules. As we conducted exploratory analyses for the seven remaining modules, a Bonferroni-corrected alpha level of 0.05/7 = 0.007 was applied. Effect sizes of significant findings are presented in the results section and can be interpreted similar to the classification by Cohen 23  Power analyses. We used R package simr to estimate the power to find a certain effect size. According to the power analysis, the smallest effects that can be found for the used models (only insignificant results) and the sample of the present study with a power of greater than or equal to 0.80 range between small and moderate effects for the within-session analyses (0.39 ≤ d ≥ 0.70).

Measures
Ethical standards. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional guides on the care and use of laboratory animals.

Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.