Introduction
Bipolar disorder (BD) is a disabling and lifelong condition characterised by recurrent manic, hypomanic and depressive episodes, with an estimated 12-month prevalence of 0.8–1.8% (De Graaf et al., Reference De Graaf, Ten Have, Van Gool and Van Dorsselaer2012; Kessler et al., Reference Kessler, Petukhova, Sampson, Zaslavsky and Wittchen2012; Merikangas et al., Reference Merikangas, Jin, He, Kessler, Lee, Sampson and São Paulo2011) and a lifetime prevalence of 1.3–2.5% (De Graaf et al., Reference De Graaf, Ten Have, Van Gool and Van Dorsselaer2012; Kessler et al., Reference Kessler, Petukhova, Sampson, Zaslavsky and Wittchen2012; Merikangas et al., Reference Merikangas, Jin, He, Kessler, Lee, Sampson and São Paulo2011). There is a high risk of recurrence and long-term studies suggest that the risk of recurrence increases with the number of previous episodes (Kessing et al., Reference Kessing, Hansen and Andersen2004a,b). Between mood episodes, many patients suffer from subsyndromal symptoms (Marangell et al., Reference Marangell, Dennehy, Miyahara, Wisniewski, Bauer, Rapaport and Allen2009; Solomon et al., Reference Solomon, Leon, Coryell, Endicott, Li, Fiedorowicz and Keller2010), and impairment in work, social and family functioning (Merikangas et al., Reference Merikangas, Jin, He, Kessler, Lee, Sampson and São Paulo2011; Sanchez-Moreno et al., Reference Sanchez-Moreno, Martinez-Aran, Tabares-Seisdedos, Torrent, Vieta and Ayuso-Mateos2009). The presence of even mild subsyndromal depressive symptoms may interfere with functional recovery in patients with BD after symptomatic recovery from a manic or hypomanic episode (Gitlin et al., Reference Gitlin, Mintz, Sokolski, Hammen and Altshuler2011). Also, patients with BD appear to experience more affective lability and intensity of emotions, suggesting that BD is characterised not only by mood episodes but also affects emotional reactivity between episodes (Henry et al., Reference Henry, Van den Bulke, Bellivier, Roy, Swendsen, M’Bailara and Leboyer2008; Saunders et al., Reference Saunders, Cipriani, Rendell, Attenburrow, Nelissen, Bilderbeck and Geddes2016). Developing treatments effectively targeting (subsyndromal) symptomatology, lability and intensity of emotions, and level of functioning in bipolar disorder, are therefore of great importance.
Cognitive behavioural therapy (CBT) – a psychological treatment, of which the effectiveness has been widely studied and demonstrated (Amick et al., Reference Amick, Gartlehner, Gaynes, Forneris, Asher, Morgan and Lohr2015; Bisson et al., Reference Bisson, Roberts, Andrew, Cooper and Lewis2013; Kaczkurkin and Foa, Reference Kaczkurkin and Foa2015; Zhu et al., Reference Zhu, Zhang, Jiang, Li, Cao, Zhou and Li2014) – aims to improve these aspects of BD. Guidelines emphasise the use of CBT as one of the psychological interventions in the treatment of BD in addition to psycho-education (Hirschfeld, Reference Hirschfeld2007; Hirschfeld et al., Reference Hirschfeld, Bowden, Gitlin, Keck, Perlis, Suppes and Yager2002; Kupka et al., Reference Kupka, Goossens and van Bendegem2015; NICE, 2006). It is a relatively short-term, focused treatment for many types of psychiatric disorders, which helps individuals to identify dysfunctional thoughts and attitudes and behaviours and learn healthier skills and habits (Beck, Reference Beck2011). CBT, in addition to treating specific symptoms, aims to improve patients’ ability to cope with their illness and possibly their sense of mastery. Sense of mastery is described as the extent to which a person feels that he has control over his life and environment (Pearlin and Schooler, Reference Pearlin and Schooler1978). Higher level of mastery is associated with lower level of depression and buffers the adverse impact of disability on depression (Jang et al., Reference Jang, Haley, Small and Mortimer2002). In BD, CBT is offered predominantly in the non-acute phase of the disease.
In the last decade, research on CBT for BD suggests that CBT effectively reduces depressive symptoms (McMahon et al., Reference McMahon, Herr, Zerubavel, Hoertel and Neacsiu2016; Salcedo et al., Reference Salcedo, Gold, Sheikh, Marcus and Nierenberg2016) and increases time to mood episode relapse or recurrence (Reinares et al., Reference Reinares, Sanchez-Moreno and Fountoulakis2014; Salcedo et al., Reference Salcedo, Gold, Sheikh, Marcus and Nierenberg2016), especially in recovered patients and patients with less previous episodes (McMahon et al., Reference McMahon, Herr, Zerubavel, Hoertel and Neacsiu2016; Reinares et al., Reference Reinares, Sanchez-Moreno and Fountoulakis2014; Scott et al., Reference Scott, Paykel, Morriss, Bentall, Kinderman, Johnson and Hayhurt2006). A meta-analysis of 55 trials with n = 6010 participants suggests that both individual and group psychological interventions were associated with reduced relapse rates at post-treatment and follow-up (Oud et al., Reference Oud, Mayo-Wilson, Braidwood, Schulte, Jones, Morriss and Kendall2016). However, further high-quality research is needed. Future research needs to incorporate both more cost-effective treatment methods for patients suffering from BD such as group therapy, and tailor research designs to clinical practice. Recent guidelines (Goodwin et al., Reference Goodwin, Haddad, Ferrier, Aronson, Barnes, Cipriani and Young2016) stress the importance of studies targeting both treatment of episodes, as well as inter-episode distress and describe the need for research on improving interventions for patients suffering from BD. An example of this is the integrative cognitive model and the TEAMS approach based on this, which describes that multiple and extreme appraisals of changes in internal state and their reciprocal impact on behaviour, physiology and the environment provide the core mechanism in maintaining and escalating bipolar symptoms (Kelly et al., Reference Kelly, Dodd and Mansell2017; Searson et al., Reference Searson, Mansell, Lowens and Tai2012). One other promising line of research are the studies in which CBT is enhanced using imagery techniques for patients suffering from BD (Hales et al., Reference Hales, Di Simplicio, Iyadurai, Blackwell, Young, Fairburn and Holmes2018).
In order to evaluate a group intervention that could possibly be more cost-effective than an individual one, to address the problem of small research samples, as well as gaining more insight into inter-episode symptoms, we investigated the effects of group CBT on mood symptomatology and several other relevant outcomes on a sample of patients who were euthymic or suffering from subsyndromal symptoms at the start of the intervention. The best known CBT interventions for BD are developed by Basco and Rush (Basco and Rush, Reference Basco and Rush2007), Lam and colleagues (Lam et al., Reference Lam, Jones and Hayward2010) and Scott (Scott, Reference Scott2010), which have all been used in research. In the Netherlands, several intervention protocols have been developed, based on these and other sources. We used a group CBT protocol developed by S. Demacker and N. Jabben (unpublished: Cognitieve gedragstherapiegroep voor mensen met een bipolaire stoornis. Draaiboek voor de therapeut and Cognitieve gedragstherapiegroep voor mensen met een bipolaire stoornis. Werkboek voor de patiënt), which is one of the most elaborate protocols on this subject in the Netherlands. As a limited number of cross-sectional measurements can neglect patients’ difficulties with ongoing mood instability (Holmes et al., Reference Holmes, Bonsall, Hales, Mitchell, Renner, Blackwell and Di Simplicio2016) we used daily mood monitoring as an outcome, which is a recent development in the research on psychological treatment for BD. Case series experimental designs are increasingly recognised as useful methods in clinical research practice to investigate individual patients’ progress and to determine whether an intervention works (Barlow et al., Reference Barlow, Nock and Hersen2009). They offer analysis methods for complex data on individual changes, while fewer participants are required who can act as their own control (Arntz et al., Reference Arntz, Sofi and van Breukelen2013; Borckardt et al., Reference Borckardt, Nash, Murphy, Moore, Shaw and O’Neil2008). Sequence analysis was performed on case series data of daily mood states (Gabadinho et al., Reference Gabadinho, Ritschard, Müller and Studer2011), in addition to a repeated measures design.
The current study aims to estimate the efficacy of a relatively brief group CBT as an intervention for BD, adjunctive to treatment as usual according to Dutch guidelines (Kupka et al., Reference Kupka, Goossens and van Bendegem2015), by: (1) daily monitoring of mood symptoms in individual participants, and (2) assessing changes in occurrence of psychological and somatic symptoms, dysfunctional attitudes, sense of mastery, psychosocial functioning and quality of life.
We expected that the intervention would be associated with: (1) a decrease in lability and intensity of (subsyndromal) depressive and (hypo)manic symptoms and (2) a decrease of comorbid physical and psychological symptoms, and dysfunctional attitudes, and an improvement in sense of mastery, psychosocial functioning, and quality of life.
Method
Participants
Patients attending an outpatient clinic at a large general hospital, and patients from two specialised outpatient clinics for BD in the Netherlands, participated in this study. Inclusion criteria were age 18 years or older, a primary DSM-IV diagnosis of bipolar I or II disorder, euthymic or only mildly depressed or hypomanic at the start of the intervention as defined by clinical impression, at least one mood episode in the past 18 months, and receiving treatment as usual according to the Dutch guideline of management of BD (Kupka et al., Reference Kupka, Goossens and van Bendegem2015). Participants were excluded if comorbid conditions were obviously impeding their ability to participate (e.g. substance abuse, mental retardation, organic brain disorder, the presence of a significant medical condition, severe suicidality, psychosis) or if they were currently receiving other psychological treatments. Pragmatic exclusion criteria were formulated to facilitate the translation of study results to clinical practice. Written informed consent was obtained from all participants.
Design
As two kinds of data were collected: (1) pre- and post-test measurements with validated questionnaires and (2) daily life chart scores, two different analytical approaches were appropriate. For the pre- and post-test effects, repeated measures analysis was applied for two time points. The patterns of the series of daily life scores could be identified using sequence analysis (Gabadinho et al., Reference Gabadinho, Ritschard, Müller and Studer2011).
Sequence analysis of case series data
A case series design was used to gain insight into self-reported daily mood fluctuations in individual participants, potentially ranging from severe depressive symptoms to euthymic to severe manic symptoms, during a baseline, treatment and follow-up phase. After inclusion, participants started charting their mood daily, until the booster session 8 weeks after the intervention ended. Sequence analysis was performed on the case series data (Gabadinho et al., Reference Gabadinho, Ritschard, Müller and Studer2011).
Repeated measures analysis
A repeated measures design was used to assess changes in several outcomes (see ‘Measures and assessments’ section), measured pre- and post-treatment and at follow-up at 2 and 12 months. All interventions (one CBT group at each centre) took place in 2014.
Intervention
A protocol for group CBT for bipolar disorder was used (protocol available on request from the first author; S. Demacker and N. Jabben, unpublished: Cognitieve gedragstherapiegroep voor mensen met een bipolaire stoornis. Draaiboek voor de therapeut and Cognitieve gedragstherapiegroep voor mensen met een bipolaire stoornis. Werkboek voor de patiënt), based on the book ‘Overcoming Mood Swings’ by Scott (Scott, Reference Scott2010) and a protocol for CBT for recurrent depression (Bockting, Reference Bockting2003, Reference Bockting2009a,b). The used treatment protocol is one of the most elaborate protocols for group CBT for BD in the Netherlands. The intervention intends to teach participants to:
Monitor and become more aware of their mood swings and precipitating factors;
Become familiar with their symptoms of depression and mania;
Recognise their changes in thinking and behaviour preceding a manic or depressive episode and how mood influences their thinking and behaviour;
Distinguish between ‘normal’ and ‘bipolar’ mood swings;
Become aware of, and change their dysfunctional beliefs and attitudes;
Gain increased sense of mastery over their disorder.
The intervention starts off by explaining both patterns of depression and mania from a CBT perspective as well as simultaneously practising techniques to improve self-management by teaching techniques to cope with cognitive, behavioural and emotional consequences of both manic and depressive episodes of the disorder (by learning to recognise own patterns of depressive and manic thinking, reacting on it by deploying contra-behaviour and formulating a plan on what to do in case of increase of symptoms). In the second half of the intervention, dysfunctional beliefs and assumptions are targeted using cognitive techniques and behavioural experiments (in which participants investigate their own dysfunctional beliefs and assumptions and are challenged to test new helping behaviour that contradicts these).
The group intervention consists of 10 weekly sessions of two hours, with a booster session 2 months after the intervention ends. In each centre, the intervention was conducted by a qualified psychologist and a co-therapist (either another psychologist or a nurse specialised in bipolar disorder). Peer supervision was offered for all therapists taking part in this intervention.
Measures and assessments
Inclusion
Modules A and D of the Structural Clinical Interview for DSM-IV Axis I Disorders (SCID-I) (First et al., Reference First, Spitzer, Gibbon and Williams1996; Van Groenestijn et al., Reference Van Groenestijn, Akkerhuis, Kupka, Schneider and Nolen1998) on mood disorders were administered by therapists trained in this interview, to confirm the diagnosis of DSM-IV bipolar I or II disorder which had previously been assessed by the participant’s psychiatrist. The Questionnaire for Bipolar Illness (QBP) (Akkerhuis et al., Reference Akkerhuis, Van Groenestijn and Nolen2005; Leverich et al., Reference Leverich, Nolen, Rush, McElroy, Keck, Denicoff and Post2001) was used to collect data on sociodemographic variables, age of onset, lifetime number of manic and depressive episodes, and hospitalisations due to mania or depression. The QBP has been used previously in a large naturalistic cohort study (Leverich et al., Reference Leverich, Nolen, Rush, McElroy, Keck, Denicoff and Post2001; Suppes et al., Reference Suppes, Leverich, Keck, Nolen, Denicoff, Altshuler and Post2001).
Mood monitoring
After inclusion (before start of the intervention), participants started charting their mood daily with the National Institute of Mental Health Life Chart Methodology (NIMH LCM) – Prospective Self-Rating (Denicoff et al., Reference Denicoff, Leverich, Nolen, Rush, McElroy, Keck and Post2000; Leverich and Post, Reference Leverich and Post1998) until the booster session 8 weeks after the intervention ended. Participants rated the severity of their mood symptoms (mania as well as depression) and related level of dysfunction on a 5-point scale (0 = no dysfunction or euthymia, 1 = mild, 2 = low moderate, 3 = high moderate, 4 = severe dysfunction). This instrument is validated by Denicoff and colleagues (Denicoff et al., Reference Denicoff, Leverich, Nolen, Rush, McElroy, Keck and Post2000), who found high correlations between LCM ratings and ratings on IDS-C (r = –0.875, p < 0.001), YMRS (r = 0.656, p < 0.001) and GAF scores (r = 0.732, p < 0.001).
Outcomes
Prior to starting the intervention, participants completed a battery of questionnaires in an online secured software environment, including:
The Altman Self-Rating Mania Scale-NL (ASRM-NL) (Altman et al., Reference Altman, Hedeker, Peterson and Davis1997): assessing the presence and severity of symptoms of mania, higher scores indicating more severe symptoms. Altman and colleagues (Altman et al., Reference Altman, Hedeker, Peterson and Davis1997) found a specificity of 85.5 and a sensitivity of 87.3.
The Inventory of Depressive Symptoms-Self Report (IDS-SR) (Rush et al., Reference Rush, Gullion, Basco, Jarrett and Trivedi1996): assessing the presence and severity of depressive symptoms, higher scores indicating more severe symptoms. The IDS-SR has good internal consistency (Cronbach’s alpha 0.79–0.85) and correlates with the Hamilton Rating Scale for Depression (HRSD, 0.67) and the Beck Depression Inventory (BDI, 0.78–0.93) (Corruble et al., Reference Corruble, Legrand, Duret, Charles and Guelfi1999a,b; Rush et al., Reference Rush, Giles, Schlesser, Fulton, Weissenburger and Burns1986). Rush and colleagues (Rush et al., Reference Rush, Gullion, Basco, Jarrett and Trivedi1996) found a correlation between the IDS-C (clinician) and the IDS-SR (self-report) of 0.88–0.91.
The Brief Symptom Inventory (BSI) (Derogatis, Reference Derogatis1975; Derogatis and Melisaratos, Reference Derogatis and Melisaratos1983): a screening instrument for psychopathology, presenting an overall score indicating the general level of physical and psychological symptoms, and consisting of nine dimensions of which outcomes are presented in Table 2. Again, higher scores indicate more symptomatology. The reliability (α = 0.71–0.96) and validity of the scales of this instrument are good (De Beurs and Zitman, Reference De Beurs and Zitman2006).
Table 1. Characteristics of the study sample (n = 23)
1 Education is coded comparable to ISCED-2011 by UNESCO (2015): 1 = primary education; 2 = lower secondary education; 3 = upper secondary education; 4 = Bachelor’s; 5 = Master’s/doctoral; 6 = postgraduate.
Table 2. Treatment effect analyses (n = 20)
1 In bold: p < 0.05; in italic: p < 0.10; —, not significant; >, significantly greater; <, significantly smaller; =, no significant difference; SD, standard deviation.
2 ASRM-NL, Altman Self-Rating Mania Scale–Dutch translation; IDS-SR, Inventory of Depressive Symptoms–Self-Report; BSI, Brief Symptom Inventory; DAS-A-NL, Dysfunctional Attitude Scale–Dutch version; PMS, Pearlin Mastery Scale–Dutch translation; FAST-NL-P, Functioning Assessment Short Test–Dutch translation; WHOQOL-BREF, World Health Organisation Quality of Life Questionnaire, short version–Dutch translation.
3 Adjusted for multiple comparisons (Bonferroni). Adjusted α = 0.025.
4 r = z/√n (Rosenthal, 1991; as mentioned in Field, Reference Field2013).
The Dysfunctional Attitude Scale-A-NL (DAS-A-NL) (Weismann, Reference Weismann1979; Weismann and Beck, Reference Weismann and Beck1978): measuring dysfunctional cognitions or assumptions, a higher score indicating more dysfunctional cognitions or assumptions. Although it is mood-state dependent, it has also been shown to predict vulnerability to depression relapse in unipolar depression as well as in some studies of BD (Zaretsky et al., Reference Zaretsky, Velyvis and Parikh2004). The Dutch version is reported to have good psychometric properties (Raes et al., Reference Raes, Hermans, Van den Broeck and Eelen2005).
The Pearlin Mastery Scale (PM) (Pearlin and Schooler, Reference Pearlin and Schooler1978): measuring sense of mastery, a higher score indicating a lower level of sense of mastery. The basic psychometric properties are good (Pearlin and Schooler, Reference Pearlin and Schooler1978), and it has been used in several different populations (Marshall and Lang, Reference Marshall and Lang1990). Cronbach’s alpha values of 0.64 (Pearlin et al., Reference Pearlin, Lieberman, Menaghan and Mullan1981) and 0.75 (Scheier et al., Reference Scheier, Carver and Bridges1994) have been reported.
The Functioning Assessment Short Test-NL-P (FAST-NL-P) (Rosa et al., Reference Rosa, Sanchez-Moreno, Martinez-Aran, Salamero, Torrent, Reinares and Vieta2007): assessing impairment or disability in six specific areas of functioning of which outcomes are presented in Table 2; the higher the scores, the more serious difficulties in functioning. This questionnaire has been shown to have good psychometric proportions (Rosa et al., Reference Rosa, Sanchez-Moreno, Martinez-Aran, Salamero, Torrent, Reinares and Vieta2007). The modified self-report version that we used has not yet been validated.
The World Health Organization Quality of Life Questionnaire-BREF (WHOQOL-BREF) (The WHOQOL Group, 1996): assessing quality of life on four domains (see Table 2), and two separate items questioning an individual’s overall perception of quality of life and an individual’s overall perception of their health. Higher scores denote higher quality of life. The Dutch version was studied in 533 psychiatric outpatients, and content validity, construct validity and reliability appeared to be good (Trompenaars et al., Reference Trompenaars, Masthoff, Van Heck, Hodiamont and De Vries2005).
Participants completed the online battery again at the end of the intervention and at follow-up at 2 and 12 months.
Data analysis
Sequence analysis of case series data
To study trajectories over time of the daily mood recordings, sequence analysis was used with R-software (package TraMineR version 1.8-10) (Gabadinho et al., Reference Gabadinho, Ritschard, Müller and Studer2011) on the daily scores. Sequences consist of a series of subsequent states – daily scores – and similarities between sequences were calculated with the optimal matching technique (OM). Sequences were defined as day scores from 7 days prior to starting the intervention until 100 days after the start. Three out of 24 participants did not register (enough) daily mood data. The average number of filled out daily severity scores from 7 days before treatment until day 100 after treatment onset was 101 (range 71–108, SD 10.4). The average number of missing daily scores in this period was 6.8 (SD 10.4).
Rather homogenous groups of trajectories – clusters – were derived from the distance matrix that results from the OM technique. To this end, cluster analysis was applied.
Shannon entropy of transversal state of the scores on each day was calculated and plotted (Gabadinho et al., Reference Gabadinho, Ritschard, Müller and Studer2011). Entropy is a measure of diversity of states within sequences and varies between zero (all sequences in the same state at time T) and 1 (maximum diversity at time T). To explore the relationships between diversity in states and background characteristics, this longitudinal entropy was regressed on co-variates. Multiple (simultaneous) regression was applied with (1) gender, (2) comorbidity, (3) number of lifetime depressive episodes, (4) age and (5) education, as covariates (method enter). Regression of clusters on these covariates was also applied.
Finally, sub-sequences of scores at the start of the treatment were compared with sub-sequences 2 months after start of treatment, to explore change over time. The first 14 days after the start of intervention were compared with a subsequence of 14 days after 2 months (days 60 to 73 after start treatment). As an indicator of functioning, the modal scores were calculated and compared.
Repeated measures analysis
For evaluation of the possible effect of the intervention on symptoms of depression and mania, other psychological and physical symptoms, dysfunctional attitudes, sense of mastery, psychosocial functioning and quality of life, a repeated measures analysis of variance was performed with time as the within-subjects factor. As assumptions for using parametric tests were violated, we performed Friedman’s ANOVAs comparing baseline data with data 1 year after the intervention ended. When significant, these were followed by pairwise comparisons using a Wilcoxon signed rank test, in which baseline data were compared with the data of the two follow-up measurements. Effect sizes were calculated for significant outcomes to interpret the magnitude of change. Statistical analyses were performed using SPSS version 23.
Results
Clinical and sociodemographic characteristics
Clinical and sociodemographic characteristics of the participants are summarised in Table 1. Twenty-four participants were included into the study, of which 18 were female. A repeated measures analysis of variance (RM ANOVA) with the treatment location as a between-subject factor with three levels showed no significant effects of treatment location on each of the outcome variables, therefore we can assume that the observations are independent of treatment location.
Assessment of drop-out and missing values
Despite being included, one participant decided not to start in the intervention because she felt her symptoms were not severe enough to justify this treatment. Twenty-three participants attended at least seven sessions and completed questionnaires post-treatment. Seventeen participants attended the booster session, and 22 participants completed questionnaires at that point. Twenty-one participants registered their mood daily for a substantial period of time (see ‘Sequence analysis of case series data’ section). Twenty participants completed questionnaires at 12 months after the last session.
Sequence analysis of case series data
Three clusters were distinguishable in the data, as shown in Fig. 1. A first cluster consisted of nine patients who scored mostly euthymic during the full period of monitoring, yet also had some days with light depression and light mania scores. The second cluster of five cases contained more severe depression and mania scores, but showed a greater probability of more euthymic days at the end of the sequence. The third cluster can be characterised as the most diverse, cycling through different states, yet with a greater probability of less severe depression scores at the end of the sequence.
Figure 1. Results sequence analysis (n = 21) with distinguishable clusters of participants by mood monitoring.
The level of entropy of scores on each day is presented in a plot in Fig. 2. This plot suggests that variation diminished over the observation period, which indicates that variation in mood states diminished over time during the intervention. The only covariate that produced a significant effect was the reported number of lifetime depressions in the QBP; the more lifetime depressive episodes participants experienced, the more entropy or variation in daily mood scores occurred (T = 3.36, p < .01) but also these participants showed a decrease in variation of mood states during treatment. Regression of the three clusters on covariates did not produce any significant effect.
Figure 2. Results diversity of daily mood states (n = 21).
Comparing mood states of participants the first 2 weeks after start of treatment to their mood states after 2 months of treatment (also 2 weeks), suggests that the (bi)modal scores (the most occurring scores) for the start subsequence were 1 (mild depression, or little or no functional impairment, 20%) or 2 (moderate light depression, or functioning with some effort, 20%). Approximately 40% of the 21 participants had these modal values in the start sub-sequence. A modal score of 0 (euthymic, stable) was found at the end sub-sequence (50% of the participants). Clearly, there is a change from depressive to more euthymic states during treatment, as recorded with daily life chart scores: modal depressive values at the onset of the observation period into modal euthymic scores (healthy functioning) after 60 days of CBT treatment.
Repeated measures analysis
Scores on the self-report questionnaires at baseline (start intervention), post-treatment and the two follow-ups and the treatment effects, are shown in Table 2. Scores of a general level of physical and psychological symptoms (measured by the BSI total) varied significantly between start and 12-month follow-up. Post hoc analyses revealed significant decreases of symptoms between start and 2-month post-treatment and between start and 12-month follow-up, both with a medium effect size (Cohen, Reference Cohen1988). Scores of the BSI subscale psychoticism (consisting of symptoms common in a withdrawn lifestyle, as seen in several psychiatric disorders), the subscale depression, and the subscale somatisation also differed significantly, with post hoc analyses revealing significant decreases of symptoms with a medium effect size (Cohen, Reference Cohen1988). A non-significant improvement was visible in depressive symptoms measured with the IDS-SR. Scores of psychosocial functioning (measured with the FAST-NL-P) varied significantly between start and 12-month follow-up, with a significant decrease in difficulties in psychosocial functioning between start and 12-month follow-up, but not between start and at follow-up 2-month post-treatment. However, the specific domains of psychosocial functioning did not show significant changes on the different time points. Scores of the WHOQOL-Bref domain psychological health varied significantly between start and 12-month follow-up, with post hoc analyses showing significant increases of self-reported psychological health between start and 2-month follow-up and start and 12-month follow-up, with a medium effect size (Cohen, Reference Cohen1988). Other outcomes on the WHOQOL-BREF, the DAS-A-NL and PMS did not show significant changes between baseline measurement and at 12-month follow-up.
Discussion
This study aimed to estimate the efficacy of a group CBT as an adjunctive intervention for BD in euthymic or mildly symptomatic patients. This is relevant, as many bipolar patients suffer from subsyndromal symptoms and functional impairment between episodes, which may negatively influence the course of their illness. A novel aspect of our study is the use of sequence analysis on case series data, which provides more detailed insight in the ongoing mood instability experienced by bipolar patients and the possible effects of treatment on this instability. In sequence analysis, the position of each successive state receives a meaningful interpretation in terms of elapsed time. Groups with similar patterns can be identified, and the relationship with covariates can be studied (Gabadinho et al., Reference Gabadinho, Ritschard, Müller and Studer2011).
Our study showed several clinically relevant findings. First, the acceptability of the intervention and study by participants appeared to be high. The drop-out rate during the intervention was very low and the rate of completed questionnaires and daily mood registration was high. There were no expected or unexpected adverse events related to the study procedure.
Sequence analysis resulted in three different clusters of patients with a similar mood pattern and course during treatment, suggesting that participants experiencing few mood symptoms prior to the start of intervention remained stable, and participants experiencing more mood symptoms showed a decrease, especially in depressive symptoms. Overall, mood instability diminished over the course of the intervention, and there was a change from depressive states to more euthymic states. Although participants with more lifetime depressive episodes showed more variation in mood states, they too showed a decrease in variation in mood states during treatment. Scott and colleagues (Scott et al., Reference Scott, Paykel, Morriss, Bentall, Kinderman, Johnson and Hayhurt2006) found that support for CBT was less effective in bipolar patients with more than 12 previous episodes. However, these results have been contradicted by Lam and colleagues (Lam et al., Reference Lam, Burbeck, Wright and Pilling2009) who found no evidence for a moderating effect of the number of previous episodes in preventing or delaying relapses in bipolar disorder by psychological therapy. Although in our study we measured variation in mood states during and shortly after intervention and no full recurrences, our results suggest that possibly bipolar patients with more lifetime depressive episodes do profit from CBT, but experience more variations in mood states to start with. Monitoring of fluctuations in mood symptomatology for a substantial period of time, as we did in this study, may reveal a more subtle change in mood symptoms in patients with BD, which cannot be shown in more classic designs.
Results of more traditional repeated measures analysis revealed a decrease in symptomatology; in depressive symptoms (as measured with the BSI), symptoms of somatisation and symptoms common in a withdrawn lifestyle as seen in more psychiatric disorders. There was an increase in overall psychosocial functioning and self-reported psychological health. Improvement appeared to persist post-treatment until 1-year follow-up for several outcomes, except for somatic symptoms and self-rated psychological health. Although this contrasts with results of previous research showing that the benefits of CBT in bipolar patients were strongest during active treatment (Ball et al., Reference Ball, Mitchell, Corry, Skillecorn, Smith and Malhi2006; Lam et al., Reference Lam, Burbeck, Wright and Pilling2009), it seems plausible since CBT aims to teach patients to change their dysfunctional thoughts and behaviours and consequently affect their psychological symptoms and psychosocial functioning in a positive way. No significant changes in manic symptoms were seen. Participants tolerated the intervention without experiencing an increase in manic symptoms, and there were no drop-outs as a result of manic symptomatology. No changes were revealed in dysfunctional attitudes, which contradicts our expectations. Dysfunctional attitudes even showed a slight non-significant increase in mean scores. There are several possible explanations for this. First, the lack of change in our data of dysfunctional attitudes may be due to our method of analysing the data. We only used the total scores of the DAS-A-NL, as its subscales have not been validated in research. Still, Johnson and Tran (Johnson and Tran, Reference Johnson and Tran2007), and Lam and colleagues (Lam et al., Reference Lam, Wright and Smith2004) suggested that patients with BD possibly have higher scores on the subscale goal attainment. Consequently, using total scores of the instrument may do no justice to this aspect of BD. Secondly, possibly different cognitive styles are visible in depressive and manic phases of the disorder, something we did not separate in our study. Thirdly, illness history may influence dysfunctional attitudes, as stated by Alloy and colleagues (Alloy et al., Reference Alloy, Reilly-Harrington, Fresco and Zechmeister1999) who found that patients with a lifetime history of hypomania, but no prior history of depression, had cognitive styles that were similar to those of healthy controls, whereas participants with a lifetime history of both depression and hypomania had more negative cognitive styles. Last, possibly participants became more aware of their dysfunctional attitudes but were not able to change these in the timeframe of the intervention. Despite our findings, the level and content of dysfunctional attitudes may be of relevance in treatment, as extreme, rigid attributions may be associated with a more severe course of depression (Stange et al., Reference Stange, Sylvia, Magalha, Miklowitz, Otto, Frank and Deckersbach2013). New protocols may need to be updated using tailored interventions targeting bipolar-specific cognitive styles. Another unexpected finding was the lack of change in sense of mastery. One possible reason we did not find any changes in sense of mastery might be that we assessed it with the PMS. No overt tests have been done to evaluate the validity of this scale, but strong face validity was suggested because the instrument is widely used and translated into multiple languages (Brady, Reference Brady2003). However, in hindsight we think the instrument may have limited construct validity as some items seem to reflect feelings of responsibility instead of feelings of mastery. To our knowledge, there is currently no better alternative instrument to assess this construct.
Limitations of our study are the small sample size and lack of a control group. Generalisation of the results may be limited as the participants were euthymic or only mildly depressed or hypomanic at inclusion, and because of the large percentage of female participants. In this design, several confounding variables are possibly of influence. We controlled for other psychotherapeutic interventions parallel to the intervention, but not for changes in medication as it seemed unethical to request medication to remain stable during and after intervention. Intragroup processes can influence the individual experiences of participants, but in this study observations appeared to be independent of treatment location. There were a large number of outcome measures. Due to small sample size, this can increase the chances of error. Non-specific effects of the intervention, such as peer support and contact with a therapist, and the influence of life events on symptoms or input of other professionals, cannot be ruled out. It is possible that mood monitoring as such has a positive additional effect on the life chart outcomes scores: regression to the mean of the scores over time may have occurred. However, recording mood with daily life charts was not new to all participants, which makes occurrence of regression to the mean as a consequence of the measurements less likely.
Despite growing awareness of the importance of non-medical treatments for BD, including CBT, research on the efficacy of individual or group CBT for BD has been relatively under-explored in comparison with research in pharmacology. Our relatively small study contributes to this emerging field, using an upcoming method of monitoring lability and intensity of mood states in time. Our results suggest that CBT may be effective in reducing the lability and intensity of mood symptoms characteristic for people suffering from BD, especially depressive symptoms. Patients with more reported lifetime depressive episodes showed more variation in mood states, but also showed a decrease of variation in mood symptoms during the intervention. Moreover, CBT may be a promising treatment not only in reducing depressive symptoms and improving psychological health at the end of treatment, but also sustained at 1-year follow-up. The results of our study justify larger controlled studies, including comparative efficacy and cost-effectiveness testing.
Acknowledgements
We thank Ingrid Albers, psychologist (GZ), and Jules Smits, nurse specialised in bipolar disorder, for their participation in carrying out the intervention. We also thank Marcel van ‘t Veer, PhD, for his assistance in statistical analyses of the data.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Conflicts of interest
H.T. Henken, R.W. Kupka, S. Draisma, J. Lobbestael, K. van den Berg, S.M.A. Demacker and E.J. Regeer have no conflicts of interest with respect to this publication.
Ethical statements
The authors have abided by the Ethical Principles of Psychologists and Code of Conduct as set out by the APA. Ethical approval was granted by the medical ethical committees of all centres in this study (Medical Research Ethics Committees United: research number NL47889.060.14, research committee GGzE Institute for Mental Health Care: number ChvN/2014007.wk, research committee Altrecht Institute for Mental Health Care: number CWOnr 1422).
Comments
No Comments have been published for this article.