Journal of Affective Disorders 169 (2014) 21–29

Contents lists available at ScienceDirect

Journal of Affective Disorders journal homepage: www.elsevier.com/locate/jad

Research report

A web-based self-management intervention for Bipolar Disorder ‘Living with Bipolar’: A feasibility randomised controlled trial Nicholas J. Todd a,n, Steven H. Jones a, Anna Hart b, Fiona A. Lobban a a Spectrum Centre for Mental Health Research, Division of Health Research, School of Health and Medicine, Lancaster University, Lancaster LA1 4YT, United Kingdom b Combining Health Information, Computation & Statistics, Lancaster Medical School, Faculty of Health and Medicine, Lancaster University, Lancaster LA1 4YT, United Kingdom

art ic l e i nf o

a b s t r a c t

Article history: Received 17 February 2014 Received in revised form 15 July 2014 Accepted 16 July 2014 Available online 27 July 2014

Background: Bipolar Disorder (BD) is a severe mental health problem. Psychological interventions are recommended by the National Institute for Health and Care Excellence (NICE) but patients experience severe inequalities in access. This study assessed the feasibility and potential effectiveness of a recovery informed web-based self-management intervention for people with BD. Methods: An online randomised controlled trial (n ¼122) compared treatment as usual (TAU) plus the ‘Living with Bipolar’ (LWB) intervention with a waiting list control (WLC) group. Results: The study recruited to target and the retention rates were high. Participants engaged with the approach. Compared with the WLC, those receiving LWB showed the most robust improvement in psychological and physical domains of quality of life, wellbeing and recovery at the end of the intervention. Limitations: The trial was not definitive and requires further investigation. Conclusions: There is preliminary evidence that a web-based treatment approach in BD is feasible and potentially effective. Such interventions could form part of the Improving Access to Psychological Therapy (IAPT) initiative in severe mental health. & 2014 Elsevier B.V. All rights reserved.

Keywords: Bipolar Disorder Online Individual psychotherapy Randomised controlled trial Quality of life Recovery

1. Introduction Bipolar Disorder (BD) is characterised by recurrent periods of extreme mood including depression, mania and mixed affective states (Goodwin and Jamison, 2007). It has been identified in 2% of the population (Merikangas et al., 2007) and is estimated to cost England d5.2 billion per year (McCrone et al., 2008). Although BD is ranked by the World Health Organisation as one of the sixth most debilitating conditions (Murray and Lopez, 1997), outcome is highly variable and there are those who experience long periods of stability (Michalak et al., 2006). Psychological interventions for BD, recommended by the National Institute for Health and Care Excellence (NICE) (NICE, 2006a) have been shown to be effective at reducing relapse and hospital admission and improving functioning (Morriss et al., 2007), but significant problems exist with overall delivery across the National Health Service (NHS) (Bird, 2006) and there is substantial inequality in access (Shapiro et al., 2003). Improving access to psychological intervention is a UK government priority currently recently extended

n

Corresponding author. E-mail address: [email protected] (N.J. Todd).

http://dx.doi.org/10.1016/j.jad.2014.07.027 0165-0327/& 2014 Elsevier B.V. All rights reserved.

into severe mental health (DOH, 2011), but this requires the development of new accessible interventions. Web-based self-management interventions have the potential to improve equality of access in cost-effective ways (DOH, 2005). Such interventions offer round the clock access and are aimed at fostering existing coping skills and empowering people to take control of their own condition (DOH, 2001). Computerised interventions are recommended as an effective treatment in the NICE guidelines for mild to moderate anxiety and depression (NICE, 2006b). However, to date, only two trials have evaluated web based interventions for BD. ‘Beating Bipolar’ (Smith et al., 2011) and the ‘Bipolar Education Programme’ (Proudfoot et al., 2007) are web-based psycho-educational programmes. However, neither reported significant difference between active intervention and control groups (Smith et al., 2011; Proudfoot et al., 2012). Other approaches currently under development include ‘Online Relapse Prevention’ (Barnes et al., 2007); and ‘Mood Swings’ (Lauder and Castle, 2008). The growing interest in this area reflects the potential utility of developing effective web based interventions, but more work is needed to design approaches that significantly improve outcome for service users. Current web based approaches have focussed on reducing symptoms and relapse, and have controlled access to

22

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

modules directing service users through a prescribed programme. Here we describe the development of a new web based intervention ‘Living with Bipolar’ (LWB). In contrast to the emphasis of previous interventions on symptom reduction, LWB has a ‘recovery’ focus; defined as supporting people in a process of living fulfilling lives alongside symptoms (Anthony, 1993). Recovery' focussed outcomes are priorities for many service users (Mead and Copeland, 2000), and for people with Bipolar Disorder, include increased involvement in meaningful activity, improvement in self-efficacy, and reduced reliance on support networks, in addition to symptom management (Jones et al., 2010). Recoveryfocussed treatment approaches have been hailed as one of the key priorities for the modernisation of UK mental health care, replacing previous approaches outlined in the National Service Framework for Mental Health – New Horizons (HM Goverment, 2009). Recovery focussed interventions show promising results in severe mental health populations (McGuire et al., 2014) including in Bipolar Disorder (Jones et al., 2012). The LWB intervention encourages service users to take responsibility for their treatment accessing, the intervention flexibly to meet their individual needs. There is an emphasis on peer support through an online discussion forum with professional support limited to moderation of this forum only. This paper reports the results of a UK randomised controlled trial of a web-based self-management intervention ‘Living with Bipolar’, designed to evaluate the feasibility and potential effectiveness of this approach.

completion of a socio-demographic questionnaire, which was adapted by the authors from the Structured Clinical Interview (SCID) overview (APA, 2000). For the purposes of the intervention, participants needed to understand written English, have access to a computer, the internet, an email account and a printer. All participants gave informed consent. Ethical approval was obtained from the Lancaster University Faculty of Health & Medicine Ethics Committee (FHMREC1000005). 2.2.3. Procedure The trial was delivered online with no face to face interaction between the researcher and the participants. The trial assessed short term outcomes and follow-up assessments were taken at three and six months post-randomisation. Participants in the treatment group were given access to the intervention for six months, to give participants enough time to fully engage with the material. Participants were informed of their group allocation by email and prompted to complete baseline and follow-up assessments by email and telephone. Missing data at baseline was minimised in the treatment group by requiring participants to have completed all measures before they were able to access the intervention. Participants in the WLC group were incentivised by being offered access to the intervention at the end of the trial if they completed the measures. Participants who dropped out without explanation were asked but did not have to provide a reason for discontinuation. Participants were given detailed information about the study design, highlighting the importance of not sharing log on details to prevent contamination between arms.

2. Method 2.3. The ‘Living with Bipolar’ (LWB) intervention 2.1. Design This was an exploratory online randomised controlled trial (registration number: ISRCTN64826171) in which participants were allocated to either receive LWB or a wait list control group (WLC). All participants continued to receive all other treatment as usual. A simple randomisation procedure was performed using a computer generated random number sequence (www.random. org), carried out by a postdoctoral researcher independent from the research team. Participants were randomised together at the end of the 12 month recruitment window prior to the delivery of baseline assessments. A detailed protocol has been published elsewhere (Todd et al., 2012b). The primary aim of this trial was to assess the feasibility of delivering the LWB intervention; the secondary aim was to assess the potential clinical effectiveness, estimating effect sizes for a number of psychological outcomes. 2.2. Participants 2.2.1. Sample size The sample size was not based on a formal power calculation as the main aim of this trial was to assess feasibility. We aimed to recruit 100 participants in total which would be sufficient to assess feasibility and estimate treatment effect sizes. 2.2.2. Recruitment Participants were recruited via voluntary sector organisations and the internet, through verbal presentations and online advertisements. Participants were eligible if they were aged 18–65, resident in the UK, with a self-reported clinical diagnosis of Bipolar Disorder Type I or II, and scoring above a threshold sensitive to diagnostic criteria for Bipolar Disorder Type I and II on the Mood Disorders Questionnaire (MDQ) (Hirschfeld et al., 2000; Twiss et al., 2008). The sample was further described through the

LWB is an online interactive recovery informed self-management intervention, broadly based on the principles of Cognitive Behavioural Therapy (Lam et al., 2010) and psycho-education (Colom and Vieta, 2006). The intervention aims to help people to: (1) learn more about bipolar experiences and how these affect their life; (2) increase their self-esteem and self-efficacy around managing bipolar; (3) increase their knowledge of specific self-management techniques to effectively manage their condition in order to pursue personally meaningful recovery goals, and; (4) increase their knowledge of practical and interpersonal skills to live a fulfilling life alongside their condition. The content was the result of a systematic review (Todd et al., 2010) and extensive service user involvement at all stages of development via five face-to-face focus groups and an online service user consultancy group (Todd et al., 2012a, 2013). Ten interactive modules were developed: (1) Recovery & Me; (2) Bipolar & Me; (3) Self-management & Me; (4) Medication & Me; (5) Getting to Know Your Mood Swings; (6) Staying well with Bipolar; (7) Depression & Me; (8) Hypomania & Me; (9) Talking about my diagnosis; and (10) Crisis & Me. Worksheets were used to enhance learning and personalise the content, and could be downloaded or printed out. Case studies and worked examples, written by service users were used extensively to reduce perceived isolation through shared experience. A mood checking tool was available for participants to help them identify major changes in their mood. Participants receive information about the most appropriate modules, given their mood symptoms. In line with the recovery agenda participants were given access to all aspects of the intervention and encouraged to use it as and when they felt appropriate. 2.3.1. Intervention support The intervention can be defined as a human supported webbased intervention, which includes ‘minimal provision of human

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

support, largely consisting of reminders and programme usage support and or a bulletin board postings moderation.’ (Barak, 2009). Generic email prompts were sent to participants in the treatment group every fortnight by NT. These emails would remind participants how long they had left to access the intervention, the features it had to offer and report levels of activity on the forum. If a participant replied they were provided with limited individual support aimed at directing them back to the LWB intervention. All email support was intended to be motivational and involved encouraging participants to use the intervention as a whole, as well as indicating elements of the intervention that might be particularly relevant to individual participants. Participants were encouraged to make use of an asynchronous forum to discuss the modules and obtain peer support. This forum included the following boards: (1) Introduce Yourself; (2) General Discussion; (3) Module Discussion; (4) Off-Topic Discussion, and was moderated by NT. Forum moderation involved monitoring and responding to risk issues and answering questions about the module content. This support did not go beyond the content available on the LWB intervention. 2.4. The waiting list control group The WLC group received TAU which is classed as the usual treatment received by the participant within health care services; this could include a general practitioner (GP) and /or specialist mental health services. Service and medication use were measured in both groups. WLC participants were sent email prompts to maintain their engagement in the trial, and where questions were raised about questionnaires these were answered by email. 2.5. Outcome measures This paper focusses on reporting the feasibility of offering a webbased psychological intervention for BD, and on primary and secondary effectiveness outcomes, which were as follows: i) Primary; quality of life; ii) Secondary; recovery, symptom severity and social functioning. Feasibility is assessed by recruitment and retention rates, and use of the online intervention. The outcome measures were chosen in line with the theoretical underpinnings of the intervention. Quality of life outcome was chosen as the primary outcome rather than recovery outcome because there were no validated recovery measures at the time the trial was undertaken. The measures used were the best available to measure these constructs within a Bipolar Disorder population: reliable and valid measurements sensitive to clinical change and correlated with other measures of the same construct. All the measures were self-report and administered online. a. The Quality of Life in BD scale (Brief version) (QoL.BD-Brief) (Michalak et al., 2010) assesses 12 areas of QoL in BD which can be summed to give a total score. b. The World Health Organisation Quality of Life assessment tool, brief version (WHOQoL-BREF) (WHO, 1998) assesses four domains of QoL: physical health, psychological health, social relationships, and environment. c. The Bipolar Recovery Questionnaire (BRQ) (Jones et al., 2012, 2013) assesses recovery processes in BD and results in a total score. d. The Internal States Scale (ISS) (Bauer et al., 1991) assesses mood state severity and results in four sub-scales: wellbeing, depression, activation and perceived conflict. e. The Social Adaptation Self-Evaluation Scale (SASS) (Bosc et al., 1997) assesses social functioning in people with depression and results in a total score.

23

An increase in scores on the measures represents an improvement, apart from three of the four subscales of the ISS (perceived conflict, activation and depression) where reduction in scores represents improvement. Medication use was measured using the Stephenson Medication Adherence Interview, adapted for self-report (Stephenson et al., 1993), and service use was measured using a measure specifically designed for this study to record the services participants used during the trial. 2.6. Statistical procedure Analyses were conducted on all participants who completed the baseline data. When data from an entire measure was missing this was regarded as a missing value. There were no instances of single items missing from any measures. SPSS Statistics 19 was used to analyse the data and a trial statistician checked the data (AH). Analyses compared groups using t-test and ANCOVA controlling for baseline scores and key sociodemographic and clinical differences between groups. This comparison was made based at the six month time point which was deemed to be the most clinically important as participants continued to use the intervention up until the six month time point and participants who had improved at three months, continued to improve at six months. Effect sizes were calculated based on Cohen's d. In order to remain consistent with the protocol, random effects modelling was conducted to estimate mean changes of outcome scores over the two follow-up assessments, contrasting between arms. Participants were emailed after dropping out of the trial to ask for a reason for their discontinuation. Based on a review of these verbatim accounts, missing data could not be deemed ‘missing at random’ and therefore a sensitivity analysis was performed. An additional extreme ‘missing not at random’ analysis was conducted, where missing scores were imputed using simple imputation in which intervention drop outs had the worst score possible score and WLC had best observation carried forward, in this case last observation.

3. Results 3.1. Feasibility 3.1.1. Participants – recruitment and retention The recruitment target was 100 participants. 240 participants initially registered their interest in the project over a 12 month period (80% within the first six months), and 122 were randomised. The decision was made to involve more than the target number of participants due to the substantial interest from participants and the online methodology providing increased capacity. All 240 participants met the inclusion criteria, but following delay between registration and the start of the trial, 100 were un-contactable and 18 declined to participate at the time, the main reasons for declining to take part were lack of time and current acute symptoms (Fig. 1). Two participants in the intervention group and three participants in the WLC group dropped out immediately after randomisation, did not complete any baseline measures or receive access to the intervention and are therefore not included in the analyses. Excluding these five drop outs, 88% of participants in the intervention group and 92% of participants in WLC group remained in the trial right through to the 6 month time point (Fig. 1). The reasons for drop out were acute symptoms (n ¼3), lack of time (n ¼3), other life events (n ¼4), unable to contact (n ¼2). The randomised participants were from all areas of the UK. They had an average age of 43.44 years (11.25), were predominately female (88; 72%), White British (109; 89%), and were educated to

24

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

Fig. 1. CONSORT Diagram depicting participants flow through the study.

degree level (85; 70%). 62 participants (51%) were employed, 31 (25%) were unemployed. The remaining 29 (24%) did voluntary work, were students or had retired. Participants were diagnosed with Bipolar I (86; 70%), were taking medication (93; 76%), in secondary care (98; 80%) and rating their disorder as a ‘serious problem’ (85; 70%). Participants had a median of five years since diagnosis and 30 days since last bipolar episode. Forty seven participants (39%) had accessed psychological support previously. After randomisation, there was an imbalance on education and employment. 21% more of the intervention group were educated to degree level and 9% more were unemployed. These variables were controlled for as co-variates. These socio-demographics are further described in Table 1. After randomisation, there was a baseline imbalance on diagnosis. 15% more of the intervention group were diagnosed with Bipolar I Disorder, also, the use of medication and secondary care are higher in

the intervention group. Diagnostic status was controlled for as a covariate with BDI and Rapid Cycling as one group and BDII as another group. The clinical history is further described in Table 2. Logistic regression analyses indicated no clear predictors of drop-out. However, because of the small numbers of drop out it is difficult to be certain. There is some evidence that participants with more hospital admissions in the intervention group, those taking medication in both groups, and those with lower social relationships score in the WLC were more likely to drop out. These variables were considered associated with clinical diagnosis and were therefore not controlled for separately. However, social relationship baseline scores were controlled for as a covariate. The reasons given by almost all the participants were more informative but not predictable from the observed data, for instance, lack of time, bereavement, divorce, holiday, physical or mental ill health.

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

3.1.2. Use of the intervention On average, participants who remained in the trial completed 9/15 modules (60% of the programme). A module is complete once a participant has viewed every page within the module. 6% completed no modules, 29% completed 1–5 modules, 31% completed 6–14 modules and 34% completed 15 modules (100% of the programme). Participants who withdrew or were lost to follow-up completed between 0% and 7% of the programme. Of the completed modules, 74% were completed within the first 3 months of the trial. The most frequently used modules, taking into account re-visiting, were the ones dealing with acute symptoms: ‘Depression & Me’ and ‘Hypomania & Me’. The least popular modules were the ones dealing with maintaining stability: ‘Staying Well’ and ‘Getting to Know Your Mood Swings’.

Table 1 Socio-demographics. Entire group (n¼ 122) Age, years, mean 43 (11.25) & SD Gender Male 34 (28%) Female 88 (72%) Ethnicity White British 109 (89%) Education Higher 85 (70%) Further 26 (21%) Secondary 11 (9%) Employment Full time 40 (33%) Part time 22 (18%) Voluntary 14 (11%) Student 7 (6%) Retired 8 (7%) Unemployed 31 (25%)

Intervention group (n¼61)

WLC group (n¼ 61)

42 (10.35)

45 (11.97)

16 (26%) 45 (74%)

18 (30%) 43 (70%)

55 (90%)

54 (88%)

49 (80%) 8 (13%) 4 (7%)

36 (59%) 18 (30%) 7 (11%)

18 13 6 4 2 18

22 9 8 3 6 13

25

70% of participants in the intervention arm registered to use the forum. There were 1927 posts in 130 topics created by participants. 90% of registered participants contributed to discussions. 65% of registered participants made between 1 and 25 posts; 10% made between 25 and 100 posts; and 15% made over 100 posts on the forum. The median number of posts was 3, with an interquartile range of 36 (0.25–36.25). There was no clear association between the number of modules completed by participants and clinical or functional outcomes. All correlations were below 0.2 in magnitude with the exception of WHO-QoL social relationships ( 0.278, p ¼0.05). There was also no clear association between time spent or number of posts on the forum and outcome. All correlations were below 0.2 in magnitude. All participants who completed the entire intervention also used the forum, however, this was bi-modal, with completers either using it extensively or very little. The total time spent online by the moderator was 106 h which equates to around 30 min per day during the trial period and about two hours per participant who completed to follow-up. In addition, 151 emails were received from treatment group participants, which on average took five minutes to answer, totalling 12 h in total, equating to about 15 min per participant who competed to follow-up. 3.2. Impact of the intervention on psychological outcome

(30%) (21%) (10%) (7%) (3%) (29%)

(36%) (15%) (13%) (5%) (10%) (21%)

3.2.1. Primary and secondary outcomes Table 3 shows the baseline and follow-up scores on all primary and secondary outcome measures. These descriptive statistics indicate that participants who had access to the ‘Living with Bipolar’ intervention showed greater improvement on all outcomes at six months follow up, apart from activation, compared to participants receiving WLC. Table 4 shows the ANCOVA for the complete cases analysis. The complete case ANCOVA suggests that a medium effect was present on quality of life, in particular physical and psychological quality of life, recovery, perceived conflict, wellbeing, depression and social function, all improved significantly more in LWB than in WLC condition.

Table 2 Clinical history.

Diagnosis Bipolar I Bipolar II Rapid cycling Previous episodes: Depression 0 1–6 7–11 12–29 Z 30 Previous episodes: Mania 1–6 7–11 12–29 Z 30 Hospitalisations 0 1–6 7–11 12–29 Days last episode, median & interquartile range Years diagnosed, median & interquartile range Problem Moderate Serious Secondary care: Previous psychology

Entire group (n¼ 122)

Intervention group (n ¼61)

86 (70%) 30 (25%) 6 (5%)

47 (77%) 11 (18%) 3 (5%)

39 (64%) 19 (31%) 3 (5%)

1 30 22 25 44

(1%) (25%) (18%) (20%) (36%)

1 16 9 13 22

(2%) (26%) (15%) (21%) (36%)

0 14 13 12 22

(0%) (23%) (21%) (20%) (36%)

56 22 16 28

(46%) (18%) (13%) (23%)

28 11 7 15

(46%) (18%) (11%) (25%)

28 11 9 13

(46%) (18%) (15%) (21%)

49 66 4 3 30 5

(40%) (54%) (3%) (3%) (205, 5–210) (8.75, 2.25–11)

24 32 2 3 51 6

(39%) (53%) (3%) (5%) (200.50, 4.50–205) (8, 3–11)

37 85 98 47

(30%) (70%) (80%) (39%)

24 37 54 24

(39%) (61%) (87%) (39%)

WLC group (n¼61)

25 34 2 0 21.00 5 13 48 44 23

(41%) (56%) (3%) (0%) (235, 5–240) (9, 2–11) (21%) (79%) (72%) (38%)

26

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

Table 3 Descriptive statistics for primary and secondary outcomes. Scale

n

reduction indicates improvement

QoLBD (12–60) Treatment WLC WHO_Physical (0–100) Treatment WLC WHO_Psychological (0–100) Treatment WLC WHO_Relationships (0–100) Treatment WLC WHO_Environment (0–100) Treatment WLC BRQ (0–3600) Treatment WLC ISS_Perceived Conflictn (0–500) Treatment WLC ISS_Wellbeing (0–300) Treatment WLC ISS_Activationn (0–500) Treatment WLC ISS_Depressionn (0–200) Treatment WLC SAS (0–60) Treatment WLC a

Baseline, Mean (SD) Treatment n¼ 59 WLC n ¼58

3 months follow-up, Mean (SD) Treatment n¼ 51 WLC n ¼53

6 months follow-up, Mean (SD) Treatment n ¼52 WLC n¼ 53

35.35 (10.10) 37.33 (7.65)

39.63 (10.06) 36.53 (8.97)

40.73 (10.27) 36.34 (10.02)

40.27 (16.11) 49.29 (18.10)

58.98 (20.62) 51.30 (18.62)

59.67 (18.84) 50.06 (18.19)

38.58 (19.61) 42.19 (20.33)

49.08 (22.28) 41.58 (19.61)

52.75 (20.93) 41.51 (22.62)

44.80 (24.75) 41.88 (26.64)

49.33 (27.39) 42.43 (24.08)

53.86 (21.10) 48.11 (25.14)

58.35 (21.81) 56.64 (19.50)

65.47 (19.10) 58.81 (19.08)

67.54 (18.36) 58.04 (21.39)

2055.00 (466.18) 2115.96 (470.22)

2311.92 (492.55) 2094.77 (426.93)

171.00 (102.92) 157.33 (102.43)

145.70 (117.21) 164.49 (101.70)

118.98 (107.43) 158.02 (87.59)

100.44 (78.22) 106.86 (75.84)

131.39 (84.53) 116.04 (72.27)

151.67 (84.64) 106.47 (76.99)

130.07 (123.00) 139.10 (127.33)

155.49 (149.82) 125.09 (114.15)

136.90 (132.59) 112.47 (112.82)

86.32 (61.57) 76.90 (52.35)

67.94 (61.23) 86.40 (62.53)

58.96 (59.76) 85.23 (59.18)

34.95 (9.23) 35.00 (8.21)

37.27 (9.11) 35.17 (7.62)

38.29 (8.95)a 33.75 (7.94)

a

2446.88 (465.70) 2111.33 (396.89)

a a

1 participant data missing.

Table 4 ANCOVA for complete cases on primary and secondary outcomes. Scale n reduction indicates Improvement

t-test (95% confidence intervals) of complete cases

ANCOVA (95% confidence intervals) controlling for co-variates ANCOVA effect size of of complete cases complete cases

QoLBD WHO_Physical WHO_Psychological WHO_Relationships WHO_Environment BRQ ISS_Perceived Conflictn ISS_Wellbeing ISS_Activationn ISS_Depressionn SAS

4.39 (0.46 to 8.32) 0.03 9.62 (2.25 to 16.98) 0.01 11.24 (2.80 to 19.68) 0.01 5.75 (  3.40 to 14.90) 0.22 9.50 (1.79 to 17.22) 0.02 335.55 (85.68 to 165.56) o 0.01  39.04 (  77.10 to  0.98) 0.04 45.19 (13.74 to 76.64) o 0.01 24.43 (  23.39 to 72.25) 0.31  26.26 (11.66 to  49.40) 0.03 4.54 (1.66 to 1.25) 0.01

3.24 (  0.92 to 7.39) 0.12 10.40 (3.12 to 17.66) 0.01 10.55 (1.78 to 19.32) 0.02 3.06 (  5.65 to 11.77) 0.49 4.32 (  2.32 to 10.96) 0.20 313.92 (145.43 to 482.40) o 0.01  43.69 (  82.93 to  4.45) 0.03 29.49 (  5.06 to 64.04) 0.09 18.96 (  31.61 to 69.54) 0.46  23.52 (  49.53 to 2.50) 0.08 4.45 (1.55 to 7.36) o0.01

0.4 0.6 0.5 0.1 0.2 0.7  0.4 0.4  0.2  0.4 0.5

The random effects results, taking into account both time points, are broadly similar and did not change the conclusions. Supplementary Table 6 shows the extreme ‘missing not at random’ (MNAR) analysis. The extreme MNAR ANCOVA indicates that the most robust potential treatment effects are physical and psychological quality of life, recovery and wellbeing. The effect sizes remained medium despite imputing the intervention drop outs as extreme treatment failures and WLC as non-failures.

increased use of services. Medication and service use have reduced on average during the trial period for the treatment group and increased on average for the WLC group. The only exception is Lamotrigine which increased in the intervention group and decreased in the WLC group. Each medication and service use was controlled for in the analysis as covariates where they had small and non-significant effect on clinical outcome.

3.2.2. Medication and service use Table 5 shows the baseline and follow-up scores on medication and service use for both arms of the trial. These descriptive statistics suggest that any treatment effects are not likely to be the consequence of increased medication dose or

4. Discussion To our knowledge, this is the first UK national feasibility randomised controlled trial of a web-based self-management intervention for Bipolar Disorder.

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

Table 5 Medication and service use. Scale

Baseline, Mean (SD) Treatment n¼ 59 WLC n ¼58

Anti-depressant Treatment WLC Lithium Treatment WLC Valproate Treatment WLC Carbamazepine Treatment WLC Lamotrigine Treatment WLC Benzodiazepine Treatment WLC Anti-Psychotic Treatment WLC Service Use (0–8) Treatment WLC a b

3 months follow-up, Mean (SD) Treatment n ¼51 WLC n¼ 53

6 months follow-up, Mean (SD) Treatment n ¼52 WLC n¼ 53

69.78 (103.38) 63.58 (95.06)

49.84 (90.95) 82.69 (109.31)a

53.78 (83.79)b 87.98 (120.10)b

247.46 (405.72) 89.66 (267.34)

119.61 (331.07) 225.48 (361.22)a

208.00 ( 381.64)b 249.52 (391.78)b

97.47 (347.45) 79.33 (303.05)

60.80 (282.71) 233.65 (537.12)a

39.00 (176.50)b 239.44 (598.71)b

22.88 (112.31) 0.00 (0.00)

13.72 (69.34) 23.08 (116.51)a

4.00 (28.28)b 23.08 (116.51)b

41.10 (104.51) 98.27 (320.77)

42.16 (107.99) 34.61 (84.33)a

105.50 (292.08)b 39.42 (87.63)b

0.34 (1.53) 4.72 (27.12) 96.86 (130.95) 46.83 (110.11) 3.31 (1.72) 3.00 (1.76)

0.20 (1.41)b 0.36 (1.57)b

0.20 (1.41) 0.23 (1.41)a 107.84 (144.28) 97.76 (146.77)a

107.84 (144.28)b 89.88 (148.87)b

Data not collected

2.50 (1.43)b 2.65 (1.67)b

1 participant data missing. 2 participant's data missing.

4.1. Main findings and comparison with other studies There is preliminary evidence that a web-based treatment approach is feasible and potentially effective in Bipolar Disorder.

4.1.1. Feasibility This study recruited to target, which is consistent with other online trials with similar inclusion criteria (Barnes et al., 2007; Proudfoot et al., 2012). Smith et al. (2011) only randomised 50 participants despite assessing 80 for inclusion, perhaps due to requiring participants to be in clinical remission. The socio-demographic data were comparable to those in other studies of web self-help for people with BD (Smith et al., 2011; Barnes et al., 2007; Proudfoot et al., 2012). This includes education levels which are significantly higher in these studies, consistent with the association between higher education and greater help seeking behaviour (Feinstein et al., 2006). The retention rates are amongst the highest in the online mental health intervention literature (Melville et al., 2010), and although this trial did not have a postintervention follow-up, higher than other BD web-based interventions (Smith et al., 2011). This suggests that high drop outs might not be ‘a natural and typical feature’ of internet interventions (Eysenbach, 2005) even amongst participants with a severe mental health problem. We suspect these retention rates are not only due to the acceptability of the intervention (Christensen et al., 2009), but also the positive motivational email relationship with participants, and the availability of an online community. In the WLC, we suspect participants remained in the study because they would receive access to the intervention at the end of the trial. Most of the participants who dropped out due to lack of time, acute symptoms and significant life events which were not significantly different between groups. These reasons are consistent with those identified in an attrition investigation of ‘The Bipolar Education Programme’ (Nicholas et al., 2010). In the current study, drop out was

27

highly correlated with non-adherence, with participants who did not access the material dropping out of the trial. Although it was not possible to identify predictors of drop out, the extreme sensitivity analysis presumed a worst case scenario. Adherence rates, measured in terms of completing every module were higher than other BD web-interventions (Smith et al., 2011; Proudfoot et al., 2012).

4.1.2. Potential effectiveness on psychological outcome The observed effect sizes are medium on a number of outcomes (even after extreme missing not at random analysis), in particular on physical and psychological quality of life, recovery and wellbeing outcomes. The only outcome not to improve in the complete case analysis was activation, although this is a debated sub-scale of the ISS (Udachina and Mansell, 2007). The potential treatment effects are not likely to be the consequence of increased medication use or service use as if anything such increases occurred in WLC. The findings observed were not explained by increased module or forum use, suggesting that participants don‘t need to complete all the modules to improve, and may instead be able to choose the modules most appropriate to their needs (Andersson et al., 2011), which is consistent with the recovery approach (Todd et al., 2012a). The existence of an online community appears to play a key role. The participants who used the forum tended to complete more modules, and all participants who completed the entire programme used the forum, albeit in different ways. However, engagement with the forum did not predict outcome. Further investigation of online community related mechanisms of change such as community involvement/inclusion or belonging is warranted. Participants with lower social relationships scores on the WHO QoL completed more modules in the intervention group, but there was some evidence that they were more likely to drop out in the WLC. This could mean that the intervention may be protecting those with poor or unsatisfied social relationships from dropping out, and although the intervention does not alter the social relationships score, it may be an important factor in engagement which warrants further investigation. The current findings provide an interesting contrast to the two other trials to report results for a web-based interventions for BD (Smith et al., 2011; Proudfoot et al., 2012). Both these trials found no evidence of a significant differential treatment effect on symptomatic or quality of life outcome, although Smith et al. (2011) did identify a medium effect size on psychological quality of life. The current trial replicates this effect size on psychological quality of life and demonstrated a further potential medium effect size on physical quality of life, recovery and wellbeing outcome. These encouraging results could be explained both by differences between the ‘Living with Bipolar’ (LWB) intervention and both ‘Beating Bipolar’ (BB) (Smith et al., 2011) and ‘The Bipolar Education Programme’ (BEP) (Proudfoot et al., 2012). BB is a structured therapeutic intervention focussed on relapse prevention and medication compliance, and BEP focuses primarily on providing information about Bipolar Disorder and is ‘not designed as a psychotherapeutic intervention’ (Proudfoot et al., 2007, p. 904). In contrast, LWB was a therapeutic intervention with a primary focus on improving quality of life outcome rather than symptomatic outcome. The content is informed by recovery principles where the focus is on building a personally meaningful life alongside symptoms. Service users are given complete access to the intervention with an emphasis on self-directed peer support. Both the LWB and BEP trials were significantly larger than that of Smith et al. (2011) and the majority of participants were recruited online, demonstrating a pre-existing interest in this mode of access. However, the LWB trial reports the highest retention and adherence rates across all the trials. Although, BEP and BB trials

28

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

had longer follow-up, assessing outcome six months after the end of the intervention. 4.2. Study limitations The findings suggest the LWB intervention could be effective at improving key outcomes for people with BD. However, as a feasibility study (Lancaster et al., 2004), the trial was not designed for hypothesis testing and requires further investigation. This trial made no attempt to examine post-investigation follow-up or economic impact. The diagnostic screen relied exclusively on self-report and was not independently verified. However, the MDQ is a validated screening measure (Twiss et al., 2008) that has been used in previous studies with community samples (Lobban et al. 2012), and which despite some contradictory evidence (Zimmerman et al., 2010), provides a highly practical approach to confirming diagnosis online, and in this study is consistent with self-reported diagnosis and medication use. No socio-demographic data was collected about the participants who registered their interest in the trial but were un-contactable, precluding conclusions being drawn about how representative the sample are of the larger population of people with Bipolar Disorder who may access online support. Participants were randomised prior to completing baseline questionnaires, thus it was not possible to undertake a true intention to treat analysis as some participants dropped out prior to baseline. Due to the small numbers of drop out prior to baseline it is unlikely this would have affected the findings. We cannot be sure that there were no predictors of drop out. Contamination between arms was possible without restricting participants based on their unique internet protocol (IP) address. There was no measurement of participants’ computer literacy which could have affected participants’ ability to use the intervention. However, these factors would be more likely to dilute than inflate the treatment effect. It is unclear how much effect the personal emails from the researcher and the personalised forum moderation had on outcome. The findings may not generalise to those who do not routinely use the internet to access self-help. The intervention was designed as a low intensity intervention to be useful for a large number of people with BD. However, there are those who need a greater level of input and this approach cannot replace high intensity, structured psychological interventions. The current intervention was a pilot intervention and lacked the interactivity expected by some participants. 4.3. Future directions A definitive trial is needed to assess the clinical and cost effectiveness of the LWB intervention and this paper provides sufficient data for such a trial to be designed. It should focus on the evaluation of quality of life and recovery as the primary outcomes over medium to long term. Dropouts are unlikely to be at random and so need careful consideration. Future studies should focus on the mechanisms of change and how important the online community and researcher prompts are to the success of this intervention. Mixed methods studies may be best placed to fully unpack the unique mechanisms of change in this intervention. The investigation of the generalisability of this approach for people in traditional NHS settings not routinely accessing the internet for self-help would support translation of the intervention into practice. Further recommendations for the design of online randomised controlled trials in severe mental health will be reported elsewhere. 4.4. Clinical implications The accessibility of evidence based psychological intervention using new modes of delivery has been recommended in order to

meet the increasing need for psychological services (Kazdin and Blase, 2011). LWB offers a potentially acceptable and effective recovery informed web-based self-management intervention for BD. This intervention could feasibly deliver an evidence based psychological intervention (NICE, 2006a) to a large number of people with BD and has the potential to be part of the stepped care pathway in the IAPT initiative for severe mental health problems (DOH, 2011).

5. Conclusion This trial has provided preliminary evidence that a web-based treatment approach in Bipolar Disorder is feasible and potentially effective.

Conflict of interest None declared from all authors. Funding source detailed in acknowledgements

Role of funding source The project was funded by a Mersey Care NHS Trust Research and Development Grant and sponsored by Lancaster University.

Acknowledgements The authors are grateful for the assistance of Dr Elizabeth Tyler with randomisation. The authors wish to thank all the participants who took part in the project.

Appendix A. Supporting information Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.jad.2014.07.027. References Andersson, G., Estling, F., Jakobsson, E., Cuijpers, P., Carlbring, P., 2011. Can the patient decide which modules to endorse? An open trial of tailored internet treatment of anxiety disorders. Cognit. Behav. Ther. 40, 57–64. Anthony, W.A., 1993. Recovery from mental illness: the guiding vision of the mental health service system in the 1990s. Psychosoc. Rehabil. J. 16 (4), 11–23. APA, 2000. Diagnostic and Statistical Manual of Mental Disordersfourth ed. American Psychiatric Association, Washington, DC. Barak, A., Klein, B., Proudfoot, J.G., 2009. Defining internet-supported therapeutic interventions. Annals of Behavioral Medicine 38 (1), 4–17. Barnes, C., Harvey, R., Mitchell, P., Smith, M., Wilhelm, K., 2007. Evaluation of an online relapse prevention program for bipolar disorder: an overview of the aims and methodology of a randomized controlled trial. Dis. Manag. Health Outcomes 15, 215–224. Bauer, M.S., Crits-Christoph, P., Ball, W.A., Dewees, E., Mcallister, T., Alahi, P., Cacciola, J., Whybrow, P.C., 1991. Independent assessment of manic and depressive symptoms by self-rating: scale characteristics and implications for the study of mania. Arch. Gen. Psychiatry 48, 807–812. Bird, A. 2006. We need to talk: the case for psychological therapy on the NHS. Accessed online: 〈http://www.mind.org.uk/campaigns_and_issues/report_an d_resources/899_we_need_to_talk〉. Bosc, M., Dubini, A., Polin, V., 1997. Development and validation of a social functioning scale, the Social Adaptation Self-evaluation Scale. Eur. Neuropsychopharmacol. 7, S57–S70. Christensen, H., Griffiths, K.M., Farrer, L., 2009. Adherence in internet interventions for anxiety and depression. J. Med. Internet Res. 11, e13. Colom, F., Vieta, E., 2006. Psychoeducation Manual for Bipolar Disorder. Cambridge University Press, Cambridge. DOH, 2001. The expert patient: a new approach to chronic disease management for the 21st century, National Institute of Clinical Excellence. Accessed online: 〈http://www.dh.gov.uk/en/Publicationsandstatistics/Publications/Publication sPolicyAndGuidance/DH_4006801〉. DOH, 2005. Self care - A real choice: Self care support – a practical option, National Institute of Clinical Excellence. Accessed online: 〈http://www.dh.gov.uk/en/ Publicationsandstatistics/Publications/PublicationsPolicyAndGuidance/ DH_4100717〉. DOH, 2011. Talking therapies: a four-year plan of action, National Institute of Clinical Excellence. Accessed online: 〈http://www.iapt.nhs.uk/silo/files/talk ing-therapies-a-four-year-plan-of-action.pdf〉.

N.J. Todd et al. / Journal of Affective Disorders 169 (2014) 21–29

Eysenbach, G., 2005. The law of attrition. J. Med. Internet Res.7, http://dx.doi.org/ 10.2196/jmir.7.1.e11. Feinstein, L., Sabates, R., Anderson, T.M., Sorhaindo, A., Hammond, C., 2006. What are the effects of education on health?. In: Desjardins, R., Schuller, T. (Eds.), Measuring the Effects of Education on Health and Civic/Social Engagement. CERI/OECD, Paris, pp. 171–354. Goodwin, F.K., Jamison, K.R., 2007. Bipolar Disorder and Recurrent Depression, 2nd ed. Oxford University Press, New York. Hirschfeld, R.M.A., Williams, J.B.W., Spitzer, R.L., Calabrese, J.R., Flynn, L., Keck JR, P. E., Lewis, L., Mcelroy, S.L., Post, R.M., Rapport, D.J., Russell, J.M., Sachs, G.S., Zajecka, J., 2000. Development and validation of a screening instrument for bipolar spectrum disorder: the mood disorder questionnaire. Am. J. Psychiatry, 157; , pp. 1873–1875. HM Goverment, H.M.S.G. 2009. New horizons: a shared vision for mental health. Accessed online: 〈http://webarchive.nationalarchives.gov.uk/20130107105354/ http://dh.gov.uk/en/publicationsandstatistics/publications/publicationspoli cyandguidance/dh_109705〉. Jones, S., Higginson, S., Murray, C. & Morrison, A.P. 2010. Recovery experiences in bipolar disorder. In: Proceedings of The British Association for Behavioural and Cognitive Psychotherapy Annual Conference, Manchester, UK, July 10. Jones, S., Mulligan, L.D., Higginson, S., Dunn, G., Morrison, A.P., 2013. The bipolar recovery questionnaire: psychometric properties of a quantitative measure of recovery experiences in bipolar disorder. J. Affect. Disord. 147 (1), 34–43. Jones, S., Mulligan, L.D., Law, H., Dunn, G., Welford, M., Smith, G., Morrison, A.P., 2012. A randomised controlled trial of recovery focused CBT for individuals with early bipolar disorder. BMC Psychiatry 12, 204. Kazdin, A.E., Blase, S.L., 2011. Rebooting psychotherapy research and practice to reduce the burden of mental illness. Perspect. Psychol. Sci. 6, 21–37. Lam, D.H., Jones, S.H., Hayward, P., 2010. Cognitive Therapy for Bipolar Disorder: A Therapist's Guide to Concepts, Methods and Practice, (2nd ed.) WileyBlackwell, Chichester. Lancaster, G.A., Dodd, S., Williamson, P.R., 2004. Design and analysis of pilot studies: recommendations for good practice. J. Eval. Clin. Practice 10, 307–312. Lauder, S.M.B., Castle, D.E.A., 2008. 〈www.moodswings〉: the highs and lows of an online intervention for bipolar disorder, preliminary findings. Aust. NZ J. Psychiatry 42, A83–A84. Lobban, F., Taylor, K., Murray, C., Jones, S., 2012. Bipolar Disorder is a two-edged sword: A qualitative study to understand the positive edge. J. Affect. Disord. 141 (2), 204–212. McCrone, P., Dhanasiri, S., Patel, A., Knapp, M. & Lewton-Smith, S. 2008. Paying the price, King’s fund. Accessed online: 〈www.kingsfund.org.uk/publications〉. McGuire, A.B., Kukla, M., Green, A., Gilbride, D., Mueser, K.T., Salyers, M.P., 2014. Illness management and recovery: a review of the literature. Psychiatr. Serv. 65 (2), 171–179. Mead, S., Copeland, M.E., 2000. What recovery means to us: consumers' perspectives. Community Ment. Health J. 36, 315–328. Melville, K.M., Casey, L.M., Kavanagh, D.J., 2010. Dropout from Internet-based treatment for psychological disorders. Br. J. Clin. Psychol. 49, 455–471. Merikangas, K.R., Akiskal, H.S., Angst, J., Greenberg, P.E., Hirschfeld, R.M., Petukhova, M., Kessler, R.C., 2007. Lifetime and 12-month prevalence of bipolar spectrum disorder in the National Comorbidity Survey replication. Arch. Gen. Psychiatry 64 (5), 543–552. Michalak, E., Yatham, L., Kolesar, S., Lam, R., 2006. Bipolar disorder and quality of life: a patient-centered perspective. Qual. Life Res. 15, 25–37. Michalak, E.E., Murray, G., Crest, B.D., 2010. Development of the QoL.BD: a disorderspecific scale to assess quality of life in bipolar disorder. Bipolar Disord. 12, 727–740.

29

Murray, C.J.L., Lopez, A.D., 1997. Global mortality, disability, and the contribution of risk factors: Global Burden of Disease Study. Lancet 349, 1436–1442. Morriss, R.K., Faizal, M.A., Jones, A.P., Williamson, P.R., Bolton, C., McCarthy, J.P., 2007. Interventions for helping people recognise early signs of recurrence in bipolar disorder. Cochrane Database Syst. Rev., Article No.: CD004854. http:// dx.doi.org/10.1002/14651858.CD004854.pub2. NICE, 2006a. Bipolar disorder: the management of bipolar disorder in adults, children and adolescents, in primary and secondary care. Accessed online: 〈http://guidance.nice.org.uk/index.jsp?action=byId&o=10989&history=t. National Institute of Clinical Excellence〉. NICE, 2006b. Depression and anxiety – computerised cognitive behavioural therapy (CCBT). Accessed online: 〈http://guidance.nice.org.uk/TA97〉. Nicholas, J., Proudfoot, J., Parker, G., Gillis, I., Burckhardt, R., Manicavasagar, V., Smith, M., 2010. The ins and outs of an online bipolar education program: a study of program attrition. J. Med. Internet Res. 12, e57. Proudfoot, J., Parker, G., Hyett, M., Manicavasagar, V., Smith, M., Grdovic, S., Greenfield, L., 2007. Next generation of self-management education: web-based bipolar disorder program. Aust. NZ J. Psychiatry 41, 903–909. Proudfoot, J., Parker, G., Manicavasagar, V., Hadzi-Pavlovic, D., Whitton, A., Nicholas, J., Smith, M., Burckhardt, R., 2012. Effects of adjunctive peer support on perceptions of illness control and understanding in an online psychoeducation program for bipolar disorder: a randomised controlled trial. J. Affect. Disord. 142 (1), 98–105. Shapiro, D.A., Cavanagh, K., Lomas, H., 2003. Geographic inequity in the availability of cognitive behavioural therapy in England and Wales. Behav. Cogn. Psychother. 31, 185–192. Smith, D.J., Griffiths, E., Poole, R., Di Florio, A., Barnes, E., Kelly, M.J., Craddock, N., Hood, K., Simpson, S., 2011. Beating bipolar: exploratory trial of a novel internet-based psychoeducational treatment for bipolar disorder. Bipolar Disord. 13, 571–577. Stephenson, B.J., Rowe, B.H., Haynes, R.B., Macharia, W.M., Leon, G., 1993. Is this patient taking the treatment as prescribed? JAMA: J. Am. Med. Assoc. 269, 2779–2781. Todd, N.J., Solis-Trapala, I., Jones, S.H. & Lobban, F.A. 2010. Computerised psychosocial interventions for mood and psychotic disorders: a systematic review. In: Proceedings of The British Association for Behavioural and Cognitive Psychotherapy Annual Conference, Manchester, UK, July 2010. Todd, N.J., Jones, S.H., Lobban, F.A., 2012a. “Recovery” in bipolar disorder: how can service users be supported through a self-management intervention? A qualitative focus group study. J. Ment. Health 21, 114–126. Todd, N.J., Solis-Trapala, I., Jones, S.H., Lobban, F.A., 2012b. An online randomised controlled trial to assess the feasibility, acceptability and potential effectiveness of ‘Living with Bipolar’: a web-based self-management intervention for Bipolar Disorder: trial design and protocol. Contemp. Clin. Trials 33, 679–688. Todd, N.J., Jones, S.H., Lobban, F.A., 2013. What do service users with bipolar disorder want from a web-based self-management intervention? A qualitative focus group study. Clin. Psychol. Psychother. 20 (6), 531–543. Twiss, J., Jones, S., Anderson, I., 2008. Validation of the Mood Disorder Questionnaire for screening for bipolar disorder in a UK sample. J. Affect. Disord. 110, 180–184. Udachina, A., Mansell, W., 2007. Cross-validation of the Mood Disorders Questionnaire, the Internal State Scale, and the Hypomanic Personality Scale. Personal. Individ. Differ. 42, 1539–1549. WHO, 1998. Development of the World Health Organisation WHOQOL-BREF quality of life assessment. Psychol. Med 28 (3), 551–558. Zimmerman, M., Galione, J.N., Ruggero, C.J., Chelminski, I., Young, D., Dalrymple, K., McGlinchey, J.B., 2010. Screening for bipolar disorder and finding borderline personality disorder. J. Clin. Psychiat. 71 (9), 1212–1217.