The Effect of Mindfulness-Based Cognitive Therapy On Rumination and Intrusive Thoughts in Patients With Bipolar Disorder: Secondary Analyses From a Randomized Controlled Trial

Background: Preliminary evidence suggests that Mindfulness-Based Cognitive Therapy (MBCT) is a promising treatment for bipolar disorder (BD). One of the proposed working mechanisms of MBCT in attenuating depressive symptoms is through the reduction of depressive rumination. In BD the effect of MBCT on rumination is less well studied. The primary aim of this study was to investigate the effect of MBCT on self-reported trait depressive rumination and an experimental state measure of negative intrusive thoughts. On an exploratory note, we investigated the effect of MBCT on positive rumination and positive intrusive thoughts. Methods: The study population consisted of a subsample of bipolar type I or II patients participating in a multicenter randomized controlled trial comparing MBCT + treatment as usual (TAU) (N = 25) to TAU alone (N = 24). Trait depressive rumination (RRS brooding subscale) and negative intrusive thoughts (breathing focus task (BFT)) were assessed at baseline and post-treatment. During the BFT, participants were asked to report negative, positive and neutral intrusive thoughts while focusing on their breathing. Results: Compared to TAU alone, MBCT + TAU resulted in a signicant pre- to post-treatment reduction of trait depressive rumination (R 2 = 0.16, F(1, 28) = 5.30, p = .029; medium effect size (f 2 = 0.19)) and negative intrusive thoughts on the BFT (R 2 = .15, F(1, 28) = 4.88, p = .036; medium effect size (f 2 = 0.17)). MBCT did not signicantly change positive rumination or positive intrusive thoughts on the BFT. Conclusions: MBCT might be a helpful additional intervention to reduce depressive rumination in BD which might reduce risk of depressive relapse or recurrence. Future research is required to replicate our ndings and to explore whether this reduction in rumination following MBCT indeed mediates a reduction in depressive symptoms and leads to relapse prevention in BD.


Background
Bipolar disorder (BD) is an affective mental disorder characterized by a chronic course of recurrent depressive, (hypo)manic and/or mixed episodes. It is one of the leading causes of disability worldwide (1), causing high economic costs (2,3). BD patients suffer from mood symptoms half of their life, and depressive symptoms seem to predominate and contribute most to their disability (4). Despite the effectiveness of existing pharmacological and psychological interventions, in more than 40% of BD patients residual mood symptoms remain (5). Therefore, it is of great importance to explore novel psychological interventions for BD to reduce these residual mood symptoms and prevent relapse at the longer term.
One of the key features of BD patients is their tendency to engage in maladaptive emotion regulation strategies such as depressive rumination, which they have in common with patients suffering from major depressive disorder (MDD) (6)(7)(8). Depressive rumination can be described as the process of thinking perseveratively about one's negative feelings and problems and their possible causes and consequences and is associated with depressive symptoms (9), also in BD (10,11). Moreover, there is considerable evidence that depressive rumination plays an important role in the onset of new depressive episodes and maintenance of MDD (9,12,13) and BD (14,15).
Mindfulness based cognitive therapy (MBCT) is an effective treatment for major depressive disorder (MDD) (16) that may sort its clinical effects by reducing depressive rumination (17). One of the proposed working mechanisms of MBCT is the increased ability to disengage from automatic maladaptive cognitive processes, such as depressive rumination. One of the core skills to be learned during MBCT is the ability to become aware of self-perpetuating ruminative thinking patterns and to let go of them. By becoming increasingly aware of automatic maladaptive cognitive processes and learning to decenter and disengage from them, patients prevent themselves to enter a vicious cycle of ruminative thinking that could otherwise aggravate symptoms of depression (18). As MDD and BD patients have many characteristics in common, such as the tendency to engage in ruminative thinking (6,8), it has been hypothesized that MBCT may also be an effective treatment for BD (19). Indeed, various systematic reviews concluded that MBCT seems promising as a treatment for BD, and that rumination may be an important therapeutic target (20)(21)(22).
Two systematic reviews showed that MBCT reduced depressive rumination in MDD and that reduction of depressive rumination may mediate the reduction of depressive symptoms (17,23). The effect of MBCT on depressive rumination in BD is less well studied. An RCT that included 95 BD patients found a statistical trend towards reduced depressive rumination following MBCT (24). In addition, an open-label trial with 12 BD patients found depressive rumination to be signi cantly reduced after MBCT (19).
In the context of an RCT on the effectiveness of MBCT in BD (25), we aimed to investigate the effect of MBCT on depressive rumination in BD patients. As traditional measures of rumination rely on self-report and therefore are prone to response and recall bias, the current study included both a self-report measure of trait rumination and an experimental measure, as suggested by van der Velden et al. (2015). We use the breathing focus task (BFT) (26) as an experimental state measure of negative, neutral, and positive intrusive thoughts patients report during a 5-minute breathing exercise. The number of negative intrusive thoughts has previously been conceptualized as a state measure of depressive rumination (Cladder-Micus et al. 2019). However, as negative intrusive thoughts measured by the BFT are not necessarily ruminative in nature, we will refer to the BFT as a state measure of negative, positive, and neutral intrusive thoughts for the remaining of this paper.
The main objective of the current study is to investigate the effect of MBCT in addition to treatment-asusual (TAU) on depressive rumination and negative intrusive thoughts in BD. In addition, we aim to establish whether the experimental state measure of negative intrusive thoughts is related to a self-report measure of depressive symptoms and depressive rumination. Apart from depressive rumination, bipolar patients also engage in positive rumination, which has been de ned as repetitively thinking about positive self-qualities and one's current positive state (27). However little is known about the role of positive rumination in the course of BD (but, see (15). Therefore, on an exploratory note, we assessed the relation between positive intrusive thoughts, manic symptoms, and positive rumination and assessed whether MBCT changed these variables.

Trial design
This study was part of a larger randomized, multicenter, evaluator-blinded, prospective clinical trial assessing the clinical effectiveness of MBCT as an additional treatment for BD (25). In total, 144 patients were randomly assigned to an 8-week MBCT training in combination with treatment as usual (TAU) or TAU only. Assessments were conducted up to 15 months follow-up. For the current study, baseline and post-treatment assessments were used.

Participants and procedure
The following inclusion criteria were applied: (a) age ≥ 18; (b) SCID-I con rmed diagnosis of bipolar type I or II; (c) suffered from at least (i) two lifetime depressive episodes (current or in (partial) remission), and (ii) one affective episode within the year prior to baseline; (d) No current (hypo)manic episode (young mania rating scale score <12) (28). Exclusion criteria were: (a) manic episode within the last 3 months before start of the trial; (b) lifetime diagnosis of schizophrenia or schizoaffective disorder, current substance abuse disorder, organic brain syndrome, antisocial or borderline personality disorder; (c) risk of suicide or aggression; (d) presence of a concurrent medical conditions impeding the ability to participate.
Participants were recruited from seven specialized outpatient clinics for adults with BD. Participants received a letter from their attending clinicians that informed them about the study. After verbal consent was obtained, participants were screened to assess eligibility and detailed information about the study was provided. When interested and eligible, a research interview was conducted by a research assistant during which written informed consent and a baseline assessment were obtained. Due to practical reasons (laptop not available, lack of research assistance, time schedule too tight), only a subset of participants (74 out of 144) were invited to perform a set of cognitive tasks at baseline and at posttreatment, including the breathing focus task (BFT; see paragraph 'Measures'). Twenty-two of those invited did not participate because they refused (n = 8), could not be planned anymore (n = 9) or because of other practical reasons (n = 5). Sociodemographic characteristics of the 52 participants who performed the experimental tasks were neither signi cantly different from the 22 participants who were invited but not participated (appendix table 3), nor from the total group of 92 participants who did not perform the tasks (appendix table 4). Research assistants who conducted the assessments were blind for treatment allocation.

Measures
Inventory of Depressive Symptomatology -Clinician administered (IDS-C) The IDS-C is an observer-rated 30-item questionnaire that assesses the severity of depressive symptoms (range 0 -84) over the past week (29). The IDS-C has good psychometric qualities (30,31) and was administered by trained research assistants (25).

Young Mania Rating Scale (YMRS)
The YMRS is a reliable, valid and sensitive 11-item questionnaire that assesses the severity of (hypo)manic symptoms (range 0 -60) (28) and was administered by trained research assistants.

Brooding subscale of Ruminative Response Scale -Extended version (RRS-EXT)
The brooding subscale of the RRS consists of 5 items that assesses a self-report measure of brooding (range 5 -20): a form of rumination strongly related to levels of depression (32). The previously reported adequate internal consistency (α = .77) is comparable to consistency in the current sample (α = .81).

Breathing focus task (BFT)
The BFT, originally developed by Borkovec and colleagues (34), is considered to be an experimental measure of intrusive thoughts (35)(36)(37). Generally, the BFT consists of a rst assessment phase followed by a worry or negative mood induction phase and a second assessment phase. Due to ethical concerns regarding inducing negative mood in a clinical sample, the BFT has also been conducted with the rst assessment phase only (37). As our BD sample also included clinically depressed patients, this last version of Cladder-Micus et al. (2019) was used.
The BFT consisted of a practice phase and the actual task. During the practice phase, participants were asked to practice focusing on their breathing for 20 seconds. After that, participants were asked to concentrate on their breathing for 45 seconds, while noticing distracting intrusive thoughts. During this period, a computer-generated tone sounded 3 times at random intervals of 10-20 seconds. After each tone, participants verbally reported whether they were focused on their breathing or distracted by an intrusive thought. When distracted by a thought, participants reported a short word label (e.g. "cannot concentrate") and classi ed the thought as negative, positive or neutral. When participants were focused on their breathing, they responded by saying 'breath' (Dutch: 'adem'). During the actual task, participants were asked to focus on their breathing for 5 minutes and responded to 12 tones at random intervals of 20-30 seconds analogously to the practice phase.
After the 5-minute breathing period, participants were asked to ll in a self-report measure of (i) 'percentage of time distracted by negative thoughts' (VAS, 0-100%), (ii) 'percentage of time distracted by positive thoughts (VAS, 0-100%), (iii) 'percentage of time focused on breathing (VAS, 1-100%), (iv) 'how di cult it was to focus on breathing' (very di cult -not at all di cult).

Mindfulness-Based Cognitive Therapy (MBCT)
Patients were randomly assigned to either (i) MBCT + TAU, in which patients received MBCT in addition to usual care typically consisting of pharmacotherapy, psycho-education and self-management interventions, or (ii) TAU alone. MBCT offered in the current study is based on the manual developed for relapse prevention in unipolar depression (38), and was slightly adapted to address the needs of BD patients (25). The MBCT training consisted of 8 weekly sessions of 2.5 hours, one 6-hour silent day, and daily home practice (±45 min). MBCT was taught by a therapist with knowledge of BD together with a MBCT teacher meeting the advanced criteria of the Association of Mindfulness Based Teachers in the Netherlands and Flanders (Belgium) which are in concordance with the Good Practice guidelines of the UK Network of Mindfulness-Based Teacher Trainers (39).

Statistical analyses
Depressive symptoms, depressive rumination and negative intrusive thoughts Data were analyzed by using the SPSS 25.0 software package and visualized by R, and Graphpad Prism version 8.0. The main goal of the current study was to get more mechanistic insight into the relation between depressive symptoms, depressive rumination and negative intrusive thoughts, and whether MBCT could change those variables in BD. Therefore, we performed per protocol analyses for the pre-post data: participants from the MBCT + TAU group who received a minimum effective dose of 4 or more sessions were included, as proposed by Teasdale et al. (2000) (40). At rst, demographic variables and baseline scores were compared between the conditions using independent sample t-tests, Mann-Whitney tests, Fisher's Exact test, and χ2 tests respectively. Secondly, spearman's rho correlations were used to evaluate the association of negative intrusive thoughts on the BFT at baseline with depressive symptoms, depressive rumination, and the self-reported time patients were distracted by negative thoughts. Thirdly, the effect of MBCT on depressive symptoms was assessed in the current sample by repeated measures analysis of variance (rmANOVA), with time (baseline to post-treatment) as within subject factor and group (MBCT + TAU vs TAU) as between subject factor. Fourth, the effect of MBCT on depressive rumination and intrusive thoughts was assessed by bootstrap linear regression analysis. This type of analysis was chosen because count data from the BFT was positively skewed, and although the (negative) binomial distribution is commonly used for positively skewed count data, these distributions did not adequately t our data. Change in (i) RRS brooding score, and (ii) number of negative, positive, neutral and total intrusive thoughts (post-treatment -baseline) was entered as dependent variable while group (MBCT + TAU vs TAU) was entered as predictor. Bias corrected and accelerated (BCa) 95% con dence intervals (CI's) and signi cance values were calculated based on 5000 bootstrap samples. Because bootstrap analysis does not rely on assumptions of normality or homoscedasticity, they provide us with an accurate estimate of the unstandardized regression coe cient B for group as a predictor variable. Cohens f 2 , the standard effect size measure for linear regression, was calculated by the following equation: , in which f 2 ≥ 0.02, f 2 ≥ 0.15 and f 2 ≥ 0.35 represent small, medium, and large effect sizes (41). The bootstrapped linear regression models were run 5 times to con rm robustness of the output.
Manic symptoms, positive rumination, and positive intrusive thoughts On an exploratory note we rst explored with spearman's rho correlations whether there is an association between baseline positive intrusive thoughts, manic symptoms, positive rumination, and self-reported time distracted by positive thoughts. Secondly, we assessed the effect of MBCT on manic symptoms using rmANOVA. Third, we explored the effect of MBCT on emotion focused and self_focused positive rumination by bootstrap linear regressions, with change in those variables as dependent variable, while group was entered as predictor variable.

Sample characteristics
From the 52 participants who performed the BFT at baseline, three were excluded from analyses due to missing data (n = 2) or because the valence was not reported for >33% of the generated tones (n = 1).
Thus, BFT data was available for 49 participants at baseline (MBCT + TAU, N = 25; TAU, N = 24). Of these participants, 34 (69%) completed the BFT post-treatment (not signi cantly different in terms of sociodemographics and clinical characteristics from non-completers; appendix table 5), of which 32 participated in at least 4 MBCT sessions. Two participants were excluded from pre-post analyses because the valence was not reported for >33% of the generated tones post-treatment, resulting in a nal sample of 30 participants equally divided over the MBCT + TAU (N = 15) and TAU group (N = 15).
Sociodemographic and clinical characteristics, and baseline scores on outcome measures are presented in Table 1. At baseline, participants had mild depressive symptoms and were in remission of manic symptoms. Participants in the MBCT + TAU group showed higher levels of self-focused and emotionfocused positive rumination than the TAU group. There were no other signi cant differences between both groups. Table 1 Sociodemographic and clinical characteristics, and baseline outcome measures and the breathing focus task at baseline.

Effect of MBCT on trait rumination
Then, we investigated whether group (MBCT + TAU compared to TAU) predicted a change in the questionnaire-based measure of depressive rumination and the number of negative intrusive thoughts on the BFT. Table 2 shows bootstrapped BCa 95% con dence intervals for all conducted linear regression models. Receiving MBCT + TAU compared to TAU resulted in a decrease of 2.05 points on the RRS brooding subscale from baseline to post-treatment (R 2 = . 16

Sensitivity analysis
When the linear regression models were run another four times, bootstrapped BCa 95% con dence intervals and corresponding p-values for B were comparable, indicating robustness of these result (appendix Table 6 -12).

Discussion
We found that MBCT resulted in a signi cant pre to post-treatment reduction in self-reported trait depressive rumination and the experimental measure of negative intrusive thoughts in BD patients. To our best knowledge, this is the rst study to use both an experimental (state) measure of intrusive thoughts and a self-report (trait) measure of depressive rumination in BD. Both trait rumination (the general tendency to ruminate) and the number of negative intrusive thoughts on the BFT were affected by MBCT, strengthening the assumption that MBCT changes dysfunctional cognitive patterns such as depressive rumination.
Our current nding that MBCT reduced depressive rumination in BD patients is in line with previous controlled studies in MDD showing reductions in trait depressive rumination (17,23,42). The effect of MBCT on trait rumination in BD is less well studied, yet, available evidence points in the same direction (19,24). A statistical trend towards reduced depressive rumination after MBCT was observed in an RCT that included 95 BD patients (24), and depressive rumination was signi cantly reduced following MBCT in an open label trial with 12 BD patients (19).
To put the reduction in rumination through MBCT in perspective, we nominally compare our data to the study of , where never depressed controls on average scored 9.3 (SD = 2.9) and remitted patients on average 11.0 (SD = 3.0) on the RRS brooding scale. Comparing these scores to those reported in the current study (from pre-11.7 (SD = 3.7) to post-MBCT 9.1 (SD = 2.3)) suggests that MBCT decreased rumination in our sample to levels comparable to never depressed controls (43).
We also found a reduction of negative intrusive thoughts on the BFT, which is in line with a controlled study in MDD showing a reduction of negative intrusive thoughts after MBCT (37). To our knowledge, the effect of MBCT on negative intrusive thoughts, or other state measures of negative thinking have not been studied in BD to date. Such an experimental state measure provides added value to self-report trait measures as it provides information regarding MBCT-induced changes in 'on-line' experience of negative intrusive thoughts during task performance, and is therefore less susceptible for recall and response bias (44). Thus, our ndings indicate that MBCT reduced both the general tendency to ruminate and also reduced negative intrusive thoughts during an experimental task.
Of interest, MBCT reduced trait rumination and negative intrusive thoughts in a relatively euthymic sample, even without a signi cant effect on depressive symptoms. This reduction in depressive rumination and negative intrusive thoughts may be bene cial for the course of bipolar disorder, as has been reported in an uncontrolled study of MBCT in MDD showing that post-treatment levels of depressive rumination predicted the risk of relapse in a 12-month follow-up period, even when controlled for previous numbers of depressive episodes and residual depressive symptoms (13). In addition, also in bipolar disorder depressive rumination seems to be involved in the onset of new depressive episodes (14) and was associated with greater lifetime depression frequency (15). Thus, we show that MBCT reduces depressive rumination in BD even in relatively euthymic patients, which potentially reduces the risk on depressive relapse or recurrence.
We were also interested in the extent to which the state measure of negative intrusive thoughts relates to the trait measure of depressive rumination. We did not nd the number of negative intrusive thoughts on the BFT to be signi cantly correlated with trait depressive rumination. This null-nding may be explained by our small sample size. Moreover, it could also be that negative thoughts on the BFT are in fact not all ruminative and intrusive in nature and therefore do not always correspond to ruminative depressogenic thoughts. Previous work on the BFT in chronically depressed patients indeed showed similar results: no relationship was found between the RRS-brooding subscale and the number of negative intrusive thoughts (37). In addition, state measures, such as negative intrusive thoughts on the BFT are much more affected by situational cues. Thus, our state measure of negative intrusive thoughts and our trait measure of depressive rumination may be different constructs that may both change over time, but this change is not necessarily related and does not per se happen simultaneously.
As secondary objective we exploratively investigated the effect of MBCT on positive rumination. MBCT did not change self-reported trait positive rumination nor positive intrusive thoughts on the BFT. These ndings may be caused by the fact that most patients in our sample were in remission of manic symptoms (showing very low scores on the YMRS), resulting in oor effects. In addition, positive intrusive thoughts on the BFT may not entirely re ect (hypo)manic intrusive thoughts. BD patients might not appraise (hypo)manic thoughts as positive and may therefore not report them as positive on the BFT.
Thus, this null-nding warrants further investigation.

Strengths, Limitations And Future Research
One major strength is the innovative character of this study. This is the rst study to our knowledge that includes, apart from conventional self-report measures, an experimental measure to assess the effect of MBCT on rumination and intrusive thoughts in BD. The use of an additional experimental measure provides complementary knowledge triangulating (45) information derived by questionnaires (17). Here, we showed that both measures were affected by MBCT, strengthening the assumption that MBCT reduces depressive rumination and negative thinking in BD.
The most important limitation of this study is the relatively small sample size, which in uences the reliability of the effects we found. Another limitation is the absence of an active control group, which prevents drawing conclusions on what speci c aspects of the MBCT may have contributed to the reduction of depressive rumination and negative intrusive thoughts. Another point of attention is the BFT itself, that may yet need re nement. In this study all reported BFT scores fall within the lower regions (0 - 3) of the measurement tool, with many participants reporting 0 negative intrusive thoughts. The zeroin ated outcomes observed on the BFT in this and other studies (35)(36)(37)(46)(47)(48) might prevent nding a relation with trait measures and may restrict the interpretation of the BFT outcomes. Future studies might bene t from reintroducing the worry/rumination induction which has been previously used in context of the BFT (35,36,47) or extending the duration of the task (e.g. 24 beeps in 10 minutes). However, despite the relatively low scores on the BFT, in our study the BFT was sensitive to pick up MBCT-induced reductions of negative intrusive thoughts. Lastly, one of the key components of MBCT is that participants are speci cally trained to pay attention to and become aware of their breathing (mindful breathing). One could speculate that patients receiving MBCT are therefore less easily distracted by and would report less intrusive thoughts in general, irrespective of valence. However, in line with previous research (37) we only found a reduction in negative intrusive thoughts after MBCT. In addition, an increase in neutral intrusive thoughts and no effect on total intrusive thoughts was observed. This might be explained by MBCT leading to a reinterpretation of thoughts with a negative content as neutral thoughts. For future research it seems valuable to include a self-report measure of state rumination to relate to the BFT, as for example the Brief State Rumination Inventory (BSRI) (49).
The current study is based on a pre-post design, which prevents conclusions on whether change in depressive rumination or negative intrusive thoughts precedes a change in depressive symptoms or is related to depressive relapse. To investigate whether depressive rumination is a mediator of the effect of MBCT on depressive symptoms and depressive relapse in BD, future well-powered longitudinal studies including multiple time points are required (50). A better understanding of the underlying mechanisms in the bene cial effects of MBCT may eventually provide insight into the individual differences in effectiveness, and may help to improve effectiveness of MBCT.

Conclusion
MBCT might be of added value to regular treatment of BD to reduce depressive rumination, a known riskfactor for relapse in BD (14,15). In addition, our study provides evidence that MBCT changes dysfunctional cognitive patterns such as depressive rumination (18)  The study protocol (NL63319.091.17) has been approved by the local medical ethics committee CMO Arnhem-Nijmegen. Verbal as well as written consent to participate were obtained.

Consent for publication
Not applicable.

Availability of data and material
The dataset used and/or analyzed for the current study are available from the corresponding authors on reasonable request.

Competing interests
IH and MH (mindfulness teachers), AS (professor/psychiatrist), DG (psychiatrist/researcher) and JL (researcher) all work at the Radboudumc Centre for Mindfulness.

Funding
The overarching RCT is externally funded by a grant from ZonMw, the Netherlands Organization for Health Research and Development (Grant Number: 843002903). The grant was awarded to prof. dr. A. E.
M. Speckens and dr. M. J. Huijbers. Data collection, management, analysis, interpretation of data, writing of the report and the decision to submit the report has not been in uenced by the funding agency.