A randomized controlled trial of mindfulness-based cognitive therapy for bipolar disorder


Tania Perich, School of Psychiatry, University of NSW, Black Dog Institute, Hospital Road, Randwick NSW 2031, Sydney, Australia.

E-mail: t.perich@unsw.edu.au



To compare the efficacy of mindfulness-based cognitive therapy (MBCT) plus treatment as usual (TAU) to TAU alone for patients with bipolar disorder over a 12-month follow-up period.


Participants with a DSM-IV diagnosis of bipolar disorder were randomly allocated to either MBCT plus TAU or TAU alone. Primary outcome measures were time to recurrence of a DSM-IV major depressive, hypomanic or manic episode; the Montgomery-Åsberg Depression Rating Scale (MADRS); and Young Mania Rating Scale (YMRS). Secondary outcome measures were number of recurrences, the Depression Anxiety Stress Scales (DASS), and the State Trait Anxiety Inventory (STAI).


Ninety-five participants with bipolar disorder were recruited to the study (MBCT = 48; TAU = 47). Intention-to-treat (ITT) analysis found no significant differences between the groups on either time to first recurrence of a mood episode or total number of recurrences over the 12-month period. Furthermore, there were no significant between-group differences on the MADRS or YMRS scales. A significant between-group difference was found in STAI – state anxiety scores. There was a significant treatment by time interaction for the DAS – achievement subscale.


While MBCT did not lead to significant reductions in time to depressive or hypo/manic relapse, total number of episodes, or mood symptom severity at 12-month follow-up, there was some evidence for an effect on anxiety symptoms. This finding suggests a potential role of MBCT in reducing anxiety comorbid with bipolar disorder.

Significant outcomes

  • Mindfulness-based cognitive therapy (MBCT) significantly improved state anxiety for those diagnosed with bipolar disorder over a 12-month follow-up period.
  • Improvements were found in reducing dysfunctional attitudes surrounding ideas of achievement for those allocated to MBCT.


  • The small sample size and high drop-out rate over the 12-month follow-up period.
  • A shorter follow-up period than other published trials of MBCT.


A number of psychotherapies have been demonstrated to be effective adjunctive treatments in the management of bipolar disorder [1]. Cognitive-behavioral therapy (CBT) is associated with fewer hospitalizations [2], longer time to depressive relapse [3], and fewer and shorter manic and depressive episodes [4]. Interpersonal and Social Rhythm Therapy (IPSRT) is effective in improving occupational functioning [5], family therapy has been associated with longer time to relapse [6], while group-based psycho-education programs have also shown reduced rates of relapse over a 12-month follow-up period [7]. Chronic care model programs, combining a range of psychological techniques along with structured follow-up, have also noted reduced mania symptoms at 2-year follow-up [8].

Despite this, some mixed results have been noted. Scott et al. [9] reported no significant difference in either postintervention relapse rates or symptom severity in a large trial of CBT compared with treatment as usual (TAU). However, a post hoc analysis demonstrated that CBT was more effective than TAU for those with <12 prior bipolar disorder episodes [9]. Those with bipolar disorder have also been found to show less dysfunctional attitudes following CBT enhanced with emotive techniques [3], suggesting potential schematic changes from such psychological treatments.

Mindfulness-based approaches are a recent addition to the range of psychological therapies used to treat psychiatric conditions. Mindfulness has been described as ‘paying attention in a particular way: on purpose, in the present moment and non-judgmentally’ [10] with therapies incorporating mindfulness including a range of techniques designed to train participants to cultivate a ‘mindful’ approach. These techniques include mindfulness meditation or more explicit instruction on how to relate to thoughts, emotions, and bodily sensations with mindful awareness.

Mindfulness-based Cognitive Therapy (MBCT) is an 8-week group therapy program developed to prevent depressive relapse in participants with a history of Major Depressive Disorder (MDD) [11]. It is based on the Mindfulness-based Stress Reduction (MBSR) program [12] and is combined with cognitive therapy techniques including psycho-education on depression and relapse prevention strategies. MBCT has been found to be effective in preventing relapse to depression in MDD for those with three or more past episodes [13-16], in reducing current depressive symptoms [17, 18] and lowering comorbid anxiety [19].

Given the results at reducing depression and comorbid anxiety, a handful of provisional studies have examined MBCT in the treatment for bipolar disorder with mixed results. In a pilot study, Williams et al. [20] randomly allocated a mixed group of those with a history of unipolar depression (n = 14) and those with bipolar disorder in remission (n = 7) to either MBCT or a wait-list control condition. They reported that levels of depression decreased for all participants allocated to MBCT from baseline to post-treatment, and those with bipolar disorder returned lower anxiety scores compared with those allocated to the wait-list control condition [20].

Further study of bipolar disorder participants (n = 22) undertaken by Miklowitz et al. [21] reported improvements in depression from pre- to post-treatment with MBCT [21]. However, a later study of MBCT for bipolar disorder (n = 15) reported no significant differences between pre- and post-treatment mania and depression scores [22]. Another study examining the effects at 3-month follow-up noted improvements in depression symptoms, psychological wellbeing, and psychosocial functioning [23].

Despite these results, no studies have examined the effects of MBCT in a randomized controlled trial design over a longer follow-up period. The present study aimed to compare MBCT plus TAU to TAU alone in a randomized controlled trial of participants with bipolar disorder over a 12-month follow-up period. Given that MBCT has been reported to prevent depressive recurrences in those with major depressive disorder [13-15], we hypothesized that MBCT would reduce time to relapse and the number of depressive episodes in participants with bipolar disorder. It was also hypothesized that bipolar disorder participants allocated to MBCT would demonstrate lower depression scores over a 12-month follow-up period post-treatment.

Those with bipolar disorder have also been found to show less dysfunctional attitudes following CBT enhanced with emotive techniques [3], suggesting potential schematic changes from such psychological treatments.

Furthermore, as MBCT has been reported to reduce comorbid anxiety in major depressive disorder and bipolar disorder [19, 20], and given that anxiety may worsen the course of bipolar disorder [24, 25], a further aim of this trial was to investigate the effect of MBCT on comorbid anxiety in bipolar disorder, both in terms of categorically defined anxiety disorders and dimensional measures of anxiety. It was hypothesized that those allocated to MBCT plus TAU would report lower anxiety symptoms compared with TAU alone and that there would be corresponding changes in other outcome measures such as dysfunctional attitudes and response styles over a 12-month follow-up period.

Given the post hoc findings of Scott et al. [9] that CBT was more effective for those with fewer than 12 prior episodes, this study also examined for such differences in treatment outcome, hypothesizing that MBCT would be more effective than TAU alone for those with <12, compared with those with 12 or more prior episodes over a 12-month follow-up period.

In addition, given that psychotherapies such as CBT have been found in prior research to change cognitive styles such as by reducing dysfunctional attitudes [3], this study also aimed to examine any treatment effects on measures of response styles and dysfunctional attitudes. It was hypothesized that those allocated to MBCT plus TAU would show lower scores on these measures compared with those allocated to TAU alone.

Aims of the study

This study aims to compare mindfulness-based cognitive therapy to treatment as usual as an adjunctive treatment for bipolar disorder. It was hypothesized that those allocated to mindfulness-based cognitive therapy would return lower depression scores and experience fewer depressive episodes over the 12-month follow-up period than those allocated to treatment as usual. It was further hypothesized that there would be corresponding decreases in anxiety and dysfunctional attitudes as a result of mindfulness-based cognitive therapy treatment.

Material and methods


Participants were recruited via newspaper advertisements and flyers in the local community. Inclusion criteria were i) lifetime DSM-IV diagnosis of bipolar I or II disorder; ii) maintained on a mood stabilizing medication for the duration of study treatment; iii) at least 18 years of age; iv) secondary school education; v) able to provide informed consent; vi) fluent in written and spoken English; vii) currently under the care of a GP or psychiatrist; viii) at least one bipolar disorder episode (hypo/mania, depression, mixed episode) over the previous 18 months; and ix) a lifetime incidence of at least three bipolar (depression, hypo/mania) episodes. Exclusion criteria were i) current DSM-IV major depressive, hypomanic or manic episode; ii) lifetime diagnosis of schizophrenia or schizoaffective disorder, current substance abuse disorder, organic brain syndrome, antisocial or borderline personality disorder; iii) the presence of a concurrent significant medical condition impeding the ability to participate; iv) currently receiving other psychological therapy, and v) previous participation in a MBCT, MBSR or Dialectical Behavior Therapy course. The study was approved by the University of New South Wales Human Research Ethics Committee (HREC approval number 08039) and registered with the Australian and New Zealand Clinical Trials Registry (ACTRN12610000414011; UTN 1111-1115-0862). The study protocol may be accessed at http://www.anzctr.org.au/. All participants provided informed written consent.

Study design

The study utilized a two (treatment plus treatment-as-usual vs. treatment-as-usual control group) by six (pre-, post-, 3-, 6-, 9-, 12-month follow-up) mixed factorial design. Study duration was 14 months. Participant recruitment was conducted from July 2008 until May 2010, with the final follow-up being completed in June 2011.


Clinician-administered measures

  1. Structured Clinical Interview for DSM-IV-TR Disorders (SCID-I) [26]. The SCID-I is a researcher-administered interview schedule that assesses symptoms of various psychiatric conditions including bipolar disorder (mania, hypomania, depression, and psychotic symptoms), major depressive disorder (single episode and recurrent), and dysthymia. Diagnostic reliability of the SCID-I for DSM-IV criteria for bipolar disorder has been reported with a kappa value of 0.74 [27].
  2. Young Mania Rating Scale (YMRS) [28]. The YMRS is an 11-item clinician-administered scale originally designed to assess symptoms of mania in a hospital inpatient setting. It has good inter-rater reliability [28].
  3. Montgomery-Åsberg Depression Rating Scale (MADRS) [29]. The MADRS is a 10-item clinician-administered scale designed to assess the severity of depressive symptoms. It has good internal consistency (α = 0.85) [30] and inter-rater reliability (0.76) [31].
  4. Composite International Diagnostic Interview (CIDI) [32]. The anxiety disorder modules of the CIDI-AUTO [33] (simple phobia, generalized anxiety disorder, panic disorder, social phobia, agoraphobia) were administered via interviewer. The CIDI-AUTO has been reported to have fair validity for the anxiety disorder modules taken together (κ = 0.40) [33].

Self-report measures

  1. Depression Anxiety Stress Scales (DASS) [34]. The DASS is a 42-item measure comprising subscales of depression, anxiety, and stress. The DASS has reported adequate reliability for all three subscales (α = 0.93 for ‘Stress’, α = 0.90 for ‘Anxiety’ and α = 0.95 for ‘Depression’) [35].
  2. State/Trait Anxiety Inventory (STAI) [36]. The STAI is a two-dimensional measure assessing both state and trait anxiety. The STAI has good internal consistency, with alpha levels for the state anxiety subscale ranging from 0.90 to 0.91 and the trait anxiety subscale ranging from 0.86 to 0.85 [36]. Test–retest reliability is high for the trait anxiety scale (ranging from 0.73 to 0.86) [36].
  3. Dysfunctional Attitudes Scale 24 (DAS-24) [37, 38]. The DAS-24 is a 24-item measure assessing dysfunctional attitudes about achievement, dependency, and self-control. It is a shortened version of Weissman's [37] original scale. The DAS-24 has acceptable internal consistency for the subscales of ‘Achievement’ (α = 0.85), ‘Dependency’ (α = 0.74), and ‘Self-control’ (α = 0.68) and good convergent validity with the longer versions of the DAS-A (0.90) and DAS-B (0.83) [38].
  4. Response Style Questionnaire (RSQ) [39-41]. The RSQ is a 48-item measure adapted by Thomas and Bentall [39] from Nolen-Hoeksema and Morrow's [41] RSQ. It comprises three subscales: Rumination, Adaptive Coping, and Risk Taking. The adapted RSQ has internal consistency of α = 0.91 for ‘Rumination’, α = 0.82 for ‘Adaptive Coping’, and α = 0.68 for ‘Risk Taking’[40]. The subscale scoring for this study was derived from Knowles et al.'s [40] factor analysis of the items in the measure.
  5. Mindful Attention Awareness Scale (MAAS) [42] is a 15-item uni-dimensional measure of trait mindfulness. It has good internal consistency (α = 0.87) [42]. Test–retest reliability at 4 weeks is 0.81 [42].


Potential participants were screened via telephone. Eligible participants were assessed with the SCID-I to confirm a lifetime diagnosis of bipolar disorder. Participants were also administered the CIDI (anxiety disorder section), YMRS, and MADRS and required to complete a battery of self-report questionnaires. Interviews were conducted within 2 weeks prior to the first MBCT therapy session.

After all assessments were completed, participants were randomly allocated to either MBCT or TAU. Central randomization was conducted by an independent researcher who was not involved with the trial conducting the randomization process using a computer-generated randomization list. A block design was used for randomization sequencing with 16 participants in each block (eight per condition). The randomization sequence was created by a statistician not associated with the research study.

Assignment of treatment condition was concealed until after the baseline assessment interviews were completed; participants were then informed of their allocation prior to the first therapy session. Interview assessments of the YMRS and MADRS were conducted by an independent rater blind to the participant's treatment allocation. Any material that may have identified participant allocation was concealed. Participants were also required to complete a self-report questionnaire booklet in addition to their interview assessments at mid- and post-treatment as well as a daily log of their mood throughout the 8-week study period.

At 3-, 9-, 6- and 12-month follow-ups, participants were asked to complete a questionnaire booklet including the baseline questionnaires and additional questions on the number of depressive, hypomanic, or manic episodes they had experienced over the preceding 3 months. At each time point, interview assessments of the YMRS and MADRS were also conducted by an independent rater blind to the participant's treatment allocation.

At 12-month follow-up, in addition to the YMRS and MADRS, the 12-month CIDI anxiety disorder section was administered. Participants were also questioned in a clinical interview using a DSM-IV check-list to determine whether any bipolar disorder episodes had occurred over the previous 12 months.

All interviews were conducted by trained researchers with a minimum of 4 years of psychology undergraduate study and who had been trained in administering these measures by psychologists and psychiatrists at the Bipolar Disorders Clinic, Black Dog Institute, Sydney Australia.

Treatment protocols


The MBCT treatment was an adaptation of the 8-week MBCT course developed by Segal et al. [11]. The MBCT program consists of weekly mindfulness meditation practice and cognitive therapy regarding depression including psycho-education [11]. In this program, psycho-education and relapse prevention information were adapted to include education about bipolar disorder, depression, hypo/mania, and anxiety. For example, in the session examining relapse to depression (Session 7), information was included about addressing mania symptoms in addition to depression. The program was also modified slightly such that the purchase of the book ‘Full Catastrophe Living’ [12] was optional rather than compulsory, and the DVD ‘MBSR’ was omitted as it was unavailable in Australia at the time of study commencement. Homework was assigned according to the MBCT protocol developed by Segal et al. [11]. The MBSR program CD set, ‘Guided Mindfulness Meditation’, was provided to each participant. However, the Yoga CD was omitted, and, instead, participants were required to alternate between a 40-min ‘body scan’ and the 40-min sitting meditation CD during the weeks where this was the required homework. This CD was removed as the study did not exclude participants with mobility issues at baseline who may have been unable to complete the mindful yoga CD if allocated to MBCT.

All group sessions were conducted at the Black Dog Institute, Prince of Wales Hospital, Sydney, Australia, with each group comprising four to eight participants. Each group session lasted for approximately 2–2.5 h. All sessions were conducted by the same registered psychologist (TP) with previous experience in conducting MBCT group therapy sessions and over 10 years experience in mindfulness meditation. Training in mindfulness meditation was obtained from Buddhist schools including those from Mahāyāna and Theravada traditions. Training in Vipassana (S. N. Goenka; Sayagyi U Ba Khin tradition) had also been obtained. Participants in the MBCT plus TAU condition also received identical weekly psycho-education material to those allocated to TAU.


Participants in the TAU condition were sent weekly handouts comprising information about bipolar disorder via email or mail. This information included topics such as the causes of bipolar disorder, available treatments, and common symptoms associated with the disorder.

Treatment fidelity

All therapy sessions were audio recorded. Treatment fidelity was assessed using the Mindfulness-based Cognitive Therapy Adherence Scale (MBCT-AS), which assesses the key constructs of MBCT during group sessions and has good reliability with α = 0.8 [43]. The item referring to the use of video material in the MBCT program was scored as 0 as this was not used in this study.

A research assistant who was not associated with the treatments rated the audiotapes. Audiotapes were randomly selected from the group sessions. The mean score on the MBCT-AS was 15.13 (SD = 3.31). Mean score reported in the validation study of the MBCT-AS was M = 15.7 (SD = 2.9) [43].

Outcome measures

Primary outcome measures were time to recurrence of a DSM-IV major depressive, hypomanic or manic episode; the MADRS; and YMRS. Secondary outcome measures were number of depressive, hypomanic or manic recurrences; DASS; STAI; DAS-24; RSQ and the MAAS.

Statistical analyses

An intention-to-treat (ITT) analysis was conducted using the sample of 95 participants who had enrolled in the study at baseline. Survival curves and relapse rates were assessed using the Cox proportion hazard regressions model examining the time (weeks) to first recurrence during the 12-month follow-up period using number of prior bipolar disorder episodes as a covariate in the model.

For the dimensional analyses, a mixed linear model (MLM) using the restricted maximum likelihood estimator (REML) was fitted in PAWS [44]. The MLM allowed all available data to be used without excluding missing data. Furthermore, the MLM did not require the imputation of any data. The ITT analysis included an assessment of the primary and secondary outcome measures. Measurement points included pre-, mid-, and post-treatment and then 3-, 6-, 9-, and 12-month follow-up post-treatment.

Power analyses for ancova indicated that a sample size of 120 with 60 participants being allocated to either MBCT or TAU (with education material) would provide at least 80% power to detect a difference equivalent to a moderate effect size of 0.35 (α = 0.05) or greater.


Of the 266 participants who were screened for the trial, 95 were eligible for inclusion. Forty-eight were randomized to MBCT and 47 to TAU. Figure 1 illustrates the flow of participants into the study. Reasons were noted when they were given by participants, such as dissatisfaction with treatment; however, some participants were not able to be contacted and were classified as declining to participate.

Figure 1.

Flow diagram of trial participants.

Fourteen participants (29%) (including 10 drop outs; see Fig. 1) did not complete the MBCT condition requirements, and 22 (47%) (including 18 drop outs; see Fig. 1) did not complete TAU requirements. Completers were defined as those who had completed both the post-treatment interview, and questionnaire assessments, and for those allocated to MBCT, had participated in a minimum of four group therapy sessions. There was a trend for those allocated to TAU to show higher drop-out and non-completion rates than those allocated to MBCT (χ2 = 3.14, P = 0.076). Participants allocated to the MBCT condition completed a mean number of seventreatment sessions (SD = 0.94; range: 4–8).

Nine participants in the MBCT condition and 11 in TAU had a change in medication initiated by their own physician during the 8-week trial. Changes were defined as any changes in dose, class, or type of medication that the participant indicated. One person in the MBCT condition did not provide information about medication use. There was no significant difference between the conditions on changes in medication during the trial period (χ2 = 1.575, P = 0.209).

Further demographic and clinical characteristics of the sample by treatment condition are described in Table 1. Using chi-square analyses, there were no significant differences between the conditions on any demographic variables. An anova found no significant difference between the conditions at baseline on the YMRS, MADRS, DASS subscales (depression, anxiety and stress), STAI (state/trait anxiety) subscales, RSQ subscales, DAS subscales, or MAAS (see Table 2).

Table 1. Demographic and clinical characteristics of the intention-to-treat sample at baseline
N (%)N (%)
  1. CIDI, Composite International Diagnostic Interview; MBCT, mindfulness-based cognitive therapy; TAU, treatment as usual.

Male17 (35)16 (34)
Female31 (65)31 (66)
Marital status
Married13 (29)12 (26)
Never married21 (47)27 (57)
Separated/Divorced/Widowed11 (24)7 (15)
Country of birth
Australia33 (70)35 (74)
Other14 (30)12 (26)
Some secondary3 (6)4 (9)
Completed secondary9 (19)5 (11)
Vocational qualification18 (38)16 (34)
Bachelor degree or higher16 (33)19 (40)
Other2 (4)3 (6)
Full time21 (45)22 (47)
Part time12 (26)3 (6)
Casual3 (6)4 (9)
Unpaid1 (2)1 (2)
Unemployed/Retired10 (21)17 (36)
Illicit drug use
Past use17 (35)19 (40)
Current7 (15)4 (9)
No history of drug use24 (50)24 (53)
Lithium13 (27)18 (39)
Sodium Valproate18 (38)14 (31)
Lamotrigine5 (10)2 (4)
Other9 (19)11 (24)
Not stated1 (2)1 (2)
Not on medication2 (4)0 (0)
Number of previous episodes
1–11 episodes18 (37)20 (42)
12 or more20 (42)21 (45)
Unknown10 (21)6 (13)
Number of previous hospitalizations
None14 (29)16 (34)
1–321 (44)20 (43)
4 or more10 (21)9 (19)
Unknown/not reported3 (6)2 (4)
Lifetime bipolar disorder diagnosis
Bipolar I27 (56)32 (68)
Bipolar II21 (44)14 (30)
Bipolar NOS0 (0)1 (2)
Anxiety disorders
CIDI-AUTO diagnosis
Any anxiety disorder33 (69)36 (77)
Panic disorder18 (38)15 (32)
Agoraphobia without panic3 (6)5 (10)
Social phobia15 (31)18 (38)
Generalized anxiety disorder13 (27)14 (30)
Specific phobia18 (38)23 (49)
Table 2. Mean scores of outcome measures – intention-to-treat sample
 BaselineMid-treatmentPost-treatment3-month follow-up6-month follow-up9-month follow-up12-month follow-up
M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)M (SD)
  1. MADRS, Montgomery-Åsberg Depression Scale; YMRS, Young Mania Rating Scale; DASS, Depression Anxiety Stress Scales; STAI, State Trait Anxiety Inventory; RSQ, Response Style Questionnaire; DAS-24, Dysfunctional Attitudes Scale 24; MAAS, Mindful Attention Awareness Scale; MBCT, Mindfulness-based Cognitive Therapy; TAU, treatment as usual.

MADRS11.71 (8.24)14.57 (10.91)9.81 (9.21)12.74 (10.86)7.14 (7.31)11.07 (9.26)4.58 (7.36)11.61 (11.64)6.67 (7.05)6.33 (5.15)8.60 (10.40)10.55 (14.29)9.60 (10.55)8.00 (5.94)
YMRS4.98 (4.49)5.47 (4.36)3.92 (4.16)5.83 (5.57)3.97 (4.57)4.44 (4.38)4.31 (5.47)2.78 (2.58)3.71 (3.65)5.00 (5.66)4.15 (6.55)2.45 (2.46)3.60 (4.13)3.31 (5.39)
DASS – Depression14.79 (12.11)19.50 (14.16)8.89 (8.12)17.97 (13.73)10.03 (9.49)13.48 (11.26)9.23 (8.81)15.57 (13.70)9.81 (7.86)13.20 (12.20)13.08 (12.91)15.19 (14.30)13.68 (11.94)15.73 (14.65)
DASS – Anxiety11.92 (10.39)12.91 (11.18)7.78 (6.08)10.33 (9.83)7.11 (6.307.74 (9.03)6.68 (7.16)9.48 (10.17)5.81 (6.82)6.80 (9.23)8.62 (9.22)9.10 (9.34)9.73 (9.50)5.40 (9.12)
DASS – Stress21.13 (14.74)22.35 (15.67)15.05 (10.67)19.07 (9.14)15.14 (11.69)16.30 (10.23)12.00 (9.74)17.48 (10.31)11.38 (8.13)13.27 (10.16)13.54 (10.35)14.38 (11.64)15.36 (12.28)13.80 (12.11)
STAI – State Anxiety47.06 (16.42)48.57 (13.87)43.00 (13.22)50.10 (11.55)40.65 (13.21)44.19 (12.50)38.10 (12.82)46.26 (15.17)39.69 (11.64)43.27 (16.04)44.50 (16.51)42.76 (11.70)39.45 (14.03)42.80 (7.64)
STAI – Trait Anxiety51.72 (14.23)54.59 (13.03)48.24 (13.95)52.43 (12.72)45.57 (13.57)51.07 (12.72)43.97 (14.11)49.96 (16.04)44.12 (11.67)47.47 (14.77)49.19 (15.22)46.90 (11.63)47.82 (13.02)49.73 (13.68)
RSQ – Rumination32.94 (12.86)39.38 (12.02)  33.93 (14.70)34.95 (13.96)33.93 (14.70)34.95 (13.96)29.56 (14.00)35.86 (13.74)32.31 (13.58)31.57 (10.44)30.64 (14.61)34.80 (10.73)
RSQ – Adaptive14.48 (7.82)14.49 (7.10)  17.94 (7.98)15.70 (5.62)17.86 (7.31)15.59 (6.91)18.52 (7.74)17.43 (7.65)16.73 (6.00)17.62 (7.02)17.09 (6.13)14.73 (6.85)
RSQ – Dangerous3.09 (3.47)5.13 (4.58)  3.03 (3.97)4.48 (5.05)2.89 (2.83)2.86 (4.24)3.17 (3.93)2.71 (4.86)2.88 (3.90)2.76 (3.59)2.59 (3.69)1.40 (1.88)
DAS-24 – Goal Attainment27.17 (7.18)26.81 (7.60)  22.44 (7.53)23.41 (8.12)23.13 (6.80)24.95 (8.40)22.04 (8.54)21.90 (9.15)21.27 (8.15)18.40 (7.54)22.55 (9.83)19.67 (10.45)
DAS-24 – Dependency17.88 (5.20)17.74 (5.51)  14.44 (5.18)17.67 (5.84)15.06 (4.85)17.55 (6.05)14.73 (4.84)15.87 (5.76)15.54 (5.04)16.62 (6.58)15.14 (5.92)16.73 (7.16)
DAS-24 – Achievement22.63 (5.78)22.30 (6.99)  19.36 (7.18)20.74 (5.30)18.06 (6.48)21.68 (6.46)17.58 (7.58)17.60 (6.16)19.04 (6.72)19.76 (7.22)18.86 (7.22)18.33 (6.44)
MAAS3.32 (0.93)3.30 (0.98)  3.87 (0.90)3.49 (1.29)3.95 (0.86)3.45 (1.05)3.88 (0.95)3.65 (1.05)3.78 (1.08)3.52 (1.12)3.82 (0.86)3.61 (1.35)

Thirty-two participants completed the 12-month follow-up interview assessments of the CIDI (MBCT N = 22; TAU N = 12) with 10 (46%) participants allocated to the MBCT condition and 4 (33%) participants in the TAU condition met diagnostic criteria for an anxiety disorder. Chi-square analysis found no significant difference between the conditions on whether or not the participant met criteria for a 12-month anxiety disorder diagnosis (χ2 = 0.083, P = 0.773).

Primary outcome measures: recurrence rates

Survival curves and recurrence rates were analyzed using the Cox proportion hazard regressions model to examine the time (weeks) to 12-month recurrence for hypo/manic (Fig. 2) and depressive episodes (Fig. 3). Data were utilized from 12-month interview assessments (MBCT N = 22; TAU N = 12) and self-reports at post-treatment, 3, 6, 9, and 12-month assessments (MBCT N = 12; TAU N = 17). Those who reported discontinuation from the trial due to symptoms (MBCT N = 3; TAU = 3) or significant clinical symptoms on the YMRS and MADRS at baseline (MBCT N = 4; TAU = 5) were included in the analysis. Last observation data at baseline was used for the remaining participants (MBCT = 7; TAU = 10).

Figure 2.

Proportion of participants who survived without relapse to hypo/mania over 12 months by treatment condition.

Figure 3.

Proportion of participants who survived without relapse to depression over 12 months by treatment condition.

Nineteen (59%) participants in the MBCT condition and 10 (48%) in the TAU reported a hypo/manic episode, while 20 (59%) in the MBCT condition and 17 (68%) in TAU reported a depressive episode over the 12-month follow-up period. Median time to hypo/mania relapse was 130 days for MBCT and 143 days for TAU. Median time to relapse for depression was 53 days for MBCT and 46 days for TAU.

There was no significant difference between the conditions on risk for recurrence or time to hypo/manic (hazard ratio = 1.813; 95% confidence interval 0.782–3.622; χ2 = 1.813, P = 0.183) or depressive (hazard ratio = 0.974; 95% confidence interval 0.509–1.864; χ2 = 0.006, P = 0.936) relapse.

Primary outcome measures: symptom severity scores

A MLM analysis was conducted on the MADRS and YMRS. Fixed effects added to the model were treatment condition, whether or not criteria were met for an anxiety disorder, and number of prior episodes (1–11 or 12 or more). There was no significant treatment by time interaction, nor treatment by time by number of prior episodes interaction for either MADRS or YMRS scores. Mean scores per condition are reported in Table 2.

Secondary outcome measures

A MLM analysis was conducted on the secondary outcome measures of the DASS subscales (depression, anxiety, and stress) and the STAI subscales (state and trait anxiety). Fixed effects added into the model were treatment condition, whether or not participants met DSM-IV criteria for an anxiety disorder and number of prior bipolar episodes.

For the DASS stress subscale, a trend was noted for a significant treatment by time interaction (F = 1.864, P = 0.088). The interaction of treatment by time by number of prior episodes interaction was not significant. No significant interactions were found for the DASS depression or anxiety subscales.

There was a significant treatment by time interaction for the STAI-state anxiety subscale (F = 2.158, P = 0.048) and a trend for the STAI-trait anxiety subscale (F = 1.944, P = 0.075). No treatment by time by number of prior episodes interaction was found. Mean scores for each condition are reported in Table 2.

A MLM analysis was conducted on the RSQ – Rumination, Dangerous Activities and Adaptive Functioning subscales, the DAS-24 – Dependency, Goal Attainment, and Achievement subscales and the MAAS. Fixed effects added into the model were treatment condition, whether or not participants met DSM-IV criteria for an anxiety disorder and number of prior bipolar episodes. A trend was found for the treatment by time interaction for the DAS-24 – dependency subscale (F = 2.186, P = 0.058). The DAS-24 – achievement subscale – demonstrated a significant treatment by time interaction (F = 2.534, P = 0.03) with a significant quadratic interaction being noted (t = 2.077, P = 0.039). The RSQ – rumination subscale – demonstrated a small trend for the treatment by time interaction (F = 1.894, P = 0.098). No other significant results were found (see Table 2).


In contrast to the findings of Teasdale et al. [13] on major depressive disorder, MBCT does not appear to offer a protective benefit in terms of delaying time to recurrence of an episode of depression or hypo/mania for those with bipolar disorder. There was also no significant treatment by time interaction for depressive or hypo/manic scores over the 12-month period, consistent with the findings of Weber et al. [22]. This may not be surprising, given that some trials examining psychological interventions such as CBT have also found that the intervention did not reduce relapse to depression or hypo/mania in comparison to a TAU condition [9].

Those in the MBCT condition returned significantly lower scores on state anxiety than TAU, with similar trends also noted for trait anxiety – a finding that was consistent with that of Williams et al. [20]. There was also a trend for self-reported stress scores to be lower in the MBCT condition than in TAU. These findings suggest that MBCT may offer some assistance in managing anxiety for those with bipolar disorder and in particular for managing state anxiety.

Although Scott et al. [9] reported that those with <12 prior bipolar episodes experienced reduced episode recurrence compared with those with 12 or more episodes, we did not replicate those results in this study. The number of prior episodes was also not associated with reductions in symptom severity measures.

Participants in the MBCT condition returned significantly lower scores on the Achievement subscale of the DAS over the 12-month follow-up period. The Achievement subscale assesses high personal standards of achievement, and it is possible that MBCT may address this through the non-judgmental nature of mindfulness practice and its emphasis on not striving to achieve a particular outcome. There was also a trend for lower scores on the Dependency subscale of the DAS in the MBCT condition. Lower scores on this measure indicate less reliance on others to achieve personal happiness. The emphasis on personal autonomy in mindfulness practice and the solitary nature of mindfulness meditation practice may reduce beliefs about personal happiness being dependent on others.

Rumination scores also showed a trend for an interaction of treatment by time, with MBCT participants returning lower scores than the TAU participants. However, as decreases in depression were not observed, and results were only at trend level, further research is required.


The small sample size and the drop-out rate at 12-month follow-up were high, with 22 participants allocated to MBCT and 12 allocated to TAU completing interview assessments. However, this study also utilized data that were obtained at other time points (3-, 6-, 9-month follow-up) for the survival analyses, and the additional use of a RCT design provides some confidence that these findings are robust. Other studies examining the protective benefits of MBCT in preventing relapse to depression in participants with major depressive disorder have assessed participants over an 18-month follow-up period [45], while this study utilized a 12-month follow-up period. It could be argued that a longer follow-up period is also required to best assess the outcome of MBCT in bipolar disorder. The group sizes in this study were also smaller than those in previously published studies of MBCT, which may also have influenced the result. Changes in medications that occurred throughout the trial may also have impacted on results, while self-reported episodes may also be not as reliable as other forms of measurement.

To conclude this is the first reported definitive RCT of MBCT in bipolar disorder. MBCT did not reduce time to recurrence of depressive or hypo/manic episodes over a 12-month follow-up period, nor was it associated with a reduction in mood symptom severity scores. However, MBCT was associated with a reduction in state and trait anxiety and levels of stress, indicating benefits to bipolar disorder patients with comorbid anxiety.

Funding sources

The project has been funded by National Health and Medical Research Council of Australia (NHMRC Program Grant 222708 and Program Grant 510135) and Rotary Australia.

Declaration of interest

Over the past 5 years, PM has received remuneration from AstraZeneca, Eli Lilly, Janssen-Cilag, and Lundbeck for lecture honoraria, consultancies, medico-legal reports, or advisory board membership. He has not been a member of an industry advisory board since late 2007 and has not accepted remuneration from industry since early 2009. The authors have no other conflicts of interest to declare.