The effectiveness and cost-effectiveness of mindfulness-based cognitive therapy compared with maintenance antidepressant treatment in the prevention of depressive relapse/recurrence: results of a randomised controlled trial (the PREVENT study)

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 08/56/01. The contractual start date was in January 2010. The draft report began editorial review in October 2014 and was accepted for publication in February 2015. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Willem Kuyken declares that he is a codirector of the Mindfulness Network Community Interest Company.

Copyright statement

© Queen’s Printer and Controller of HMSO 2015. This work was produced by Kuyken et al. under the terms of a commissioning contract issued by the Secretary of State for Health. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

Background

Depression is a major public health problem that, like other chronic conditions, tends to run a relapsing and recurrent course,1 producing substantial decrements in health and well-being. 2 The World Health Organization (WHO) predicts that by 2020 depression will be the second leading cause of disability in the world. 3 The cost of mood disorders in the UK has been calculated at 7% of national income with a direct cost to health services of £3B. 4 More than 50% of patients experience at least two episodes of depression. 5 Moreover, without ongoing treatment, people suffering recurrent depression suffer relapse/recurrence at rates as high as 80%, even after successful acute treatment. 5,6 Thus, most of the prevalence, burden and cost of depression is a consequence of relapse/recurrence and the majority of the burden attributable to depression could be offset through interventions aimed at preventing depressive relapse/recurrence. 7

Current treatments for depression in primary care

Currently, the majority of depression is treated in primary care and treatment with maintenance antidepressant medication (m-ADM) is the mainstay approach to preventing relapse/recurrence. 8–11 To stay well the National Institute for Health and Care Excellence (NICE) recommends that people with a history of recurrent depression continue m-ADM for at least 2 years. 8 However, many patients experience unpleasant side effects, rates of m-ADM adherence tend to be low and patients often express a preference for psychosocial interventions. 12–14 Service user organisations, such as Depression Alliance, therefore advocate greater availability of psychosocial therapies. Government advisors Layard and Clark4 recommend that there should be parity of esteem for mental and physical health. That is, today, evidence-based treatment should be as available for mental illness as for physical illness. They provide a compelling narrative that this would be a cost-effective approach to enhancing the mental health of the nation. In line with this, significant government initiatives, such as Improving Access to Psychological Therapies, aim to offer accessible, acceptable and cost-effective psychosocial models of care and the last 10 years has seen significant progress in the accessibility of evidence-based mental health services in the UK. 4,15,16

Psychosocial approaches to prevent depressive relapse/recurrence

There is a strong evidence base for psychosocial treatments for recurrent depression, most notably cognitive–behavioural therapy (CBT) and interpersonal therapy, and promising evidence for several other therapies. 8,17 Policy initiatives, user groups and professional consensus recommend as priorities for future research the development of psychosocial interventions to prevent depressive relapse/recurrence and the use of non-traditional delivery systems, such as group interventions, to maximise accessibility and cost-effectiveness. 18,19 Mindfulness-based cognitive therapy (MBCT) is a psychosocial group-based relapse prevention programme. It was developed from translational research into mechanisms of depressive relapse/recurrence. 20 It is recommended by NICE as a psychological approach to relapse prevention for people who are currently well but who have experienced three or more previous episodes of depression. 8 There is much clinical enthusiasm for MBCT, as evidenced by the high rates of patient engagement, the development of MBCT therapist training programmes in the UK at the Universities of Bangor, Exeter and Oxford and more recently the development of an All Party Parliamentary Group on Mindfulness. 21 Implementation of MBCT throughout the NHS is patchy and variable but becoming more widespread. 22,23 In summary, MBCT shows the potential to contribute significantly to reducing the prevalence of depression in UK primary care settings.

Review of the evidence for mindfulness-based cognitive therapy up to the trial start date

Exploratory pilot trial

Through a Medical Research Council (MRC)-funded pilot trial27–29 we sought to address the gaps in the evidence base outlined in the previous section by comparing MBCT with the current mainstay approach to relapse prevention, m-ADM. In this pilot trial 123 patients who were currently taking m-ADM to manage recurrent depression were randomised to either continue m-ADM or take part in a MBCT course and then taper and discontinue their m-ADM. The findings suggested that MBCT may not only provide an alternative to m-ADM (relapse/recurrences at 15 months: MBCT 47%, m-ADM 60%), but also, in an adequately powered definitive trial, produce superior outcomes (Figure 1). 28 Finally, the study suggested that MBCT may also be superior to m-ADM in terms of improved quality of life, reduced residual depressive symptoms and reduced psychiatric comorbidity. The study’s stated aims and outcomes were:

1. Examine feasibility. The achievement of all study milestones within budget and on time demonstrated feasibility.

2. Establish recruitment methods. An acceptable and effective recruitment methodology was developed as evidenced by our over-recruitment by 54%.

3. Establish a training model for MBCT therapists. We developed a model of training with input from John Teasdale, one of the developers of MBCT.

4. Cost MBCT. The MBCT intervention was estimated to cost circa £200 per participant,28 which compares favourably with an estimated cost of £750 for individual CBT. 16

5. Elicit service user feedback. At the end of the pilot trial all participants were interviewed and asked about their experience of MBCT as a relapse prevention programme. 30 These studies informed the PREVENT study to maximise MBCT’s acceptability and application to supporting patients taper and discontinue their m-ADM.

FIGURE 1.

Kaplan–Meier survival curve in the exploratory trial. Survival estimates by intention to treat.

Costs and cost-effectiveness

To inform commissioning of care for people with recurrent depression we must establish MBCT’s cost-effectiveness over an adequate follow-up period. In our pilot trial28 we found that the per-person cost for the MBCT group was £274 more than that for the m-ADM group but this difference was not significant. Cost-effectiveness analyses suggested that the additional cost of MBCT may be justified in terms of improvement in the proportion of patients who undergo relapse/recurrence, but only if willingness to pay for such improvements is ≥ £600. In terms of depression-free days, the incremental cost-effectiveness ratio (ICER) of MBCT is comparable with that in similar studies, with a ratio of £30 per depression-free day for total costs and £14 per depression-free day for health service costs. (Please note that these costs were reported in reference 28 in international dollars using a conversion rate of 0.6.) Recent ICERs for collaborative care programmes include £20 in terms of total outpatient costs31 and £12 in terms of total inpatient and outpatient costs. 32 Estimates of £8–14 have been reported for a depression relapse prevention programme. 33 Exploration of costs over time suggested that differences in cost converge and MBCT becomes cheaper than m-ADM over the final 3-month period of the study. If this trend were to continue, the relative cost-effectiveness of MBCT may increase over time. Future studies should consider longer follow-up periods to test this hypothesis.

Process studies to examine acceptability and mechanism of change

As with the cost-effectiveness of MBCT, the evidence for the mechanism of change in outcomes with MBCT therapy is limited. Coelho et al. 26 found no evidence of the ‘specific effectiveness of MBCT, despite this being the logical progression from the current research’ (p. 1004). In other words, no trials had shown whether MBCT works through its specific hypothesised mechanisms and/or through non-specific cognitive–behavioural, psychoeducation and group/therapist support components. They therefore recommended ‘the need for randomised controlled trials to compare MBCT with other non-pharmacological approaches’ (p. 1005) and the inclusion of tests of the specific and non-specific mechanisms of change.

Proposed mechanisms of action

Mindfulness-based cognitive therapy’s theoretical premise is that depressive relapse/recurrence is associated with the reinstatement of negative modes of thinking and feeling that contribute to depressive relapse and recurrence. 34 This ‘reactivated’ network of negative thoughts and feelings can perpetuate into a depressive episode. Laboratory studies support this model by showing that recovered depressed patients revert to a depressive information processing style following a sad mood induction (for a review see Segal et al. 35). Following successful treatment for depression, those patients showing greater reactivation of dysfunctional thinking styles in response to a sad mood provocation are at the highest risk of relapse/recurrence over an 18-month period. 35 Moreover, patients who recovered with CBT showed significantly less cognitive reactivation than those who recovered with ADM. Attenuating the reactivation of dysfunctional thinking styles may therefore represent one mechanism by which CBT helps prevent depressive relapse/recurrence. Mindfulness skills are taught as a means to note distressing thoughts and feelings, hold such experiences in awareness and cultivate acceptance and self-compassion to break up associative networks and offset the risk of relapse/recurrence. 34 This dimension of mindfulness, which involves meeting distressing thoughts and feelings with kindness, empathy, equanimity and patience, is woven into mindfulness-based applications and is thought to be crucial to the change process. 36 Intentional attention is learned in the first three MBCT sessions using a range of core mindfulness practices (the body scan, mindful movement and mindfulness of the breath). As well as developing attention, these early sessions highlight habitual patterns of reactivity that arise during meditation (e.g. intrusive negative thoughts) and the associated aversion and judgements (e.g. ‘I am no good at this, I am just more aware of how badly I feel’). As the person learns mindfulness skills, he or she learns to give less authority to self-judgement and blame – the fuel for depressive thinking – and to respond to these states with compassion, in short to step out of habitual unhelpful patterns of thinking. 36 Elucidating these putative mechanisms of action of MBCT will improve theoretical understanding of how this relatively new treatment works and provide the opportunity to enhance efficacy through emphasis of these mechanisms.

Our previous mechanisms study, which was embedded in the pilot trial, demonstrated that, consistent with MBCT’s theoretical premise, increases in mindfulness and self-compassion across treatment mediated the effect of MBCT on depressive symptoms at 15 months’ follow-up. 37 Furthermore, MBCT changed the relationship between post-treatment cognitive reactivity and depressive outcome. In patients receiving m-ADM, greater reactivity predicted poorer outcome, replicating previous findings. 35 However, following MBCT there was no support for this toxic relationship between reactivity and outcome, with an indication that enhancement of self-compassion had nullified this relationship. These findings were consistent with an evidence synthesis arguing that MBCT works through a ‘retraining of awareness and non-reactivity, allowing the individual to more consciously choose those thoughts, emotions, and sensations, rather than habitually reacting to them’ (p. 569). 38 The results are also in line with data using a self-report measure of cognitive reactivity showing that MBCT attenuates reactivity and its impact on depression. 39 This is reflected in distinct neural responses to sad mood in people who had undergone MBCT, suggesting a neural basis for these findings. 40 The findings suggest that, whereas negative mood may reactivate dysfunctional thinking patterns in people who have participated in a MBCT class, it is their response to these dysfunctional thoughts that is altering their impact at follow-up.

Rationale for the research

In summary, the evidence for MBCT indicates that it is more effective than usual care in preventing depressive relapse/recurrence in people with a history of three or more episodes. Moreover, there is preliminary evidence from our pilot trial to indicate that it may be cost-effective and that it works through its hypothesised mechanism. An editorial published in 2012 in the British Medical Journal41 concluded that key remaining uncertainties include (1) whether or not MBCT provides an alternative for people wishing to discontinue m-ADM, (2) how acceptable MBCT is and (3) what mechanism MBCT works through.

Aims and objectives

The overarching policy aim and research question of the PREVENT trial was to establish whether MBCT with support to taper and/or discontinue antidepressant medication (MBCT-TS) is superior to and more cost-effective than an approach of m-ADM in a primary care setting for patients with a history of recurrent depression followed up over a 2-year period.

The specific objectives of the PREVENT trial were to compare MBCT-TS and m-ADM for patients with recurrent depression in terms of:

• time to depressive relapse/recurrence over 2 years (primary outcome)

• depression-free days, residual depressive symptoms and health-related quality of life at 1 month post treatment and 9, 12, 18 and 24 months post randomisation and psychiatric and medical comorbidity at 12 and 24 months post randomisation (secondary outcomes)

• costs and cost-effectiveness as assessed by incremental cost per relapse/recurrence prevented and incremental cost per quality-adjusted life-year (QALY) at 1 month post treatment and 9, 12, 18 and 24 months post randomisation.

In addition, we asked whether or not an increase in mindfulness skills is the key mechanism of change of MBCT.

To address these policy and explanatory questions patients were recruited through primary care and treated in accessible primary care or community settings. This was a single-blind, parallel RCT examining MBCT-TS compared with m-ADM. The m-ADM was constant over the 2 years of the study and the psychosocial intervention was a front-loaded, 8-week relapse/recurrence prevention programme. Patients in the MBCT-TS arm received support to taper their ADM. The process studies employed mixed methods.

Mindfulness-based cognitive therapy

Mindfulness-based cognitive therapy is an 8-week, group-based programme (12–15 participants per group) designed to teach skills that prevent the recurrence of depression. 20 It is derived from both mindfulness-based stress reduction (MBSR), a programme with demonstrated efficacy in ameliorating distress in people suffering chronic disease,42 and CBT for acute depression,43 a programme with demonstrated efficacy in preventing depressive relapse/recurrence. 44 MBCT is based on theoretical and empirical work showing that depressive relapse/recurrence is associated with the reinstatement of automatic modes of thinking, feeling and behaving that are counterproductive in contributing to and maintaining depressive relapse and recurrence (e.g. self-critical thinking and behavioural avoidance). 45

Participants learn to recognise these ‘automatic pilot’ modes and respond to them in more functional ways by employing complementary cognitive–behavioural and mindfulness practices. This involves recognising early warning signs of depressive relapse/recurrence, being able to ‘turn towards’ them with kindly interest and decentring in ways that ‘nip them in the bud’. The cognitive–behavioural component involves responding to negative thinking and behavioural activation. In the latter stages of the course participants develop a ‘response/action plan’ that sets out strategies for responding when they become aware of early warning signs of relapse/recurrence. The mindfulness component involves extensive practice of mindfulness skills (e.g. meditation practices) designed to improve participants’ attentional control, ability to decentre from negative thinking and emotion regulation and increase self-compassion. There is a movement towards meeting difficulty with curiosity, a quality of allowing and self-compassion before deciding on skilful ways of responding. Participants are encouraged to bring these skills in to all aspects of their lives.

Adaptations made to the mindfulness-based cognitive therapy programme for the PREVENT trial

In the PREVENT trial delivery of MBCT followed the manual as described by Segal, Williams and Teasdale34 with a few adaptations based on (1) the need for therapists to support MBCT participants in tapering their ADM and (2) experience from the pilot trial. 28 Participants attended a one-to-one orientation session with the therapist followed by group sessions lasting 2.25 hours over 8 consecutive weeks. Session content included guided mindfulness practices (i.e. body scan, sitting meditation, movement); inquiry into participants’ experience of these practices; weekly review of home practice (i.e. 40 minutes of mindfulness practice per day with the guidance of a CD, bringing mindfulness into everyday life); and teaching of/dialogue around cognitive–behavioural skills. MBCT-TS patients also received an additional four group reunion sessions during the first year of follow-up to provide ongoing support and rehearse the key components of the interventions. The first of these booster sessions occurred within 3–5 weeks after the end of the 8-week MBCT-TS programme as this was the time when patients would be tapering their ADM and may be in most need of support. Such booster sessions significantly enhance relapse/recurrence prevention46 and were a match for general practitioner (GP) attention, which was part of ongoing clinical management in the m-ADM group. A list of the individual adaptations and when they were made is provided in Table 1.

Study design

The PREVENT trial was a two-arm, multicentre, single-blind superiority trial randomly allocating patients in a 1 : 1 ratio to receive either MBCT-TS or m-ADM. The trial included a parallel economic evaluation to examine the cost-effectiveness of MBCT-TS compared with m-ADM and a mixed-methods process evaluation that used qualitative methods to assess the acceptability of MBCT-TS from the perspectives of patients and included a quantitative analysis of potential mediators.

Setting, participants and recruitment

We recruited participants from 95 general practices in urban and rural settings in four UK centres: Bristol, Exeter and East Devon, North and Mid Devon and South Devon. Inclusion and exclusion criteria for patient participation were refined through the pilot trial28 to maximise real-world applicability to the population of people in primary care with recurrent depression who are treated with ADM and who are interested in considering a psychological approach to relapse/recurrence prevention. They are also based on our experience of running a Devon NHS Primary Care Trust-commissioned depression service (see www.centres.ex.ac.uk/mood/).

Inclusion criteria

Participants were considered for inclusion if they:

• had a diagnosis of recurrent major depressive disorder in full or partial remission according to the Diagnostic and Statistical Manual of Mental Disorders-Fourth Edition (DSM-IV)47

• had had three or more previous major depressive episodes in which depression was the primary disorder and it was not secondary to substance abuse, bereavement or a general medical condition

• were aged ≥ 18 years

• were on a therapeutic dose of ADM in line with the British National Formulary (BNF)48 and NICE guidance8

• were open either to continue taking antidepressants for 2 years or to take part in a MBCT class and consider stopping their ADM.

Exclusion criteria

Participants were considered unsuitable for inclusion if they:

• were currently depressed, as assessed using the Structured Clinical Interview for DSM-IV (SCID)47

• had a comorbid diagnosis of current substance abuse (patients with previous substance abuse were eligible for inclusion as long as they were in sustained full remission)

• had organic brain damage

• had current/past psychosis, including bipolar disorder

• displayed persistent antisocial behaviour

• engaged in persistent self-injury that required clinical management/therapy

• were undergoing formal concurrent psychotherapy.

Recruitment procedure

The recruitment strategy for the PREVENT trial built on the protocol used in the pilot trial, which proved acceptable and effective. 29 This is summarised in the following pathway:

1. General practice searches identified patients who had been prescribed ADM at a therapeutic dose in the last 3 months.

2. GPs were then asked to screen this list to exclude any patients who they knew met the exclusion criteria.

3. Letters were sent to the remaining patients enclosing an information pamphlet and reply form.

4. Interested patients were telephoned to discuss the study and a short eligibility screening interview was conducted over the telephone. Information about the study and MBCT was available to help people to begin to make an informed decision about participation. This included the timings and locations of the MBCT groups. The vast majority of exclusions were identified at this stage.

5. Patients who met the telephone screening criteria were invited to attend a face-to-face baseline interview.

6. Consenting patients who met the PREVENT inclusion criteria joined the trial during the baseline assessment.

7. Within a month of the start of the next MBCT-TS group a current GRID-Hamilton Depression Rating Scale (GRID-HAMD)49 score was obtained for each participant so that randomisation could occur.

Although the majority of referrals were through GP surgeries, interested patients were also able to self-refer into the study. We employed a number of different strategies to advertise the trial including placing posters in carefully targeted sites, developing a website, regional media coverage and leaflet dropping in local chemists.

Patients were recruited in cohorts during recruitment ‘time slices’ that corresponded to the 6–8 weeks before the next MBCT-TS group was due to start. Baseline assessments were conducted as close as possible to the start of the MBCT-TS group as residual depressive symptoms are a powerful predictor of relapse/recurrence. 50 On average, each researcher recruited six patients per month (Figure 2).

FIGURE 2.

Summary of the PREVENT recruitment process and follow-up assessments. BDI-ll, Beck Depression Inventory, second edition; EQ-5D, European Quality of Life-5 Dimensions; MSCL, Medical Symptom Checklist; WHOQOL, WHO Quality of Life.

Randomisation and concealment

Patients were randomised in a 1 : 1 ratio to either MBCT-TS or m-ADM using computer-generated random permuted blocks and stratified by recruitment locality (four sites) and symptomatic status (asymptomatic: GRID-HAMD score < 8 vs. partially symptomatic: GRID-HAMD score ≥ 8). Stratification by locality enabled a proportionate workload for research staff and therapists. Residual depressive symptoms are a powerful predictor of relapse/recurrence. 50 Moreover, the pilot trial suggested the importance of this stratification variable in predicting outcome. 27 To ensure concealment randomisation was conducted using a password-protected trial website that was maintained by the Peninsula Clinical Trials Unit (CTU), a UK Clinical Research Collaboration (UKCRC)-accredited CTU. Participants were informed of the outcome of randomisation in a letter sent from the trial administrator. Research assessors were blind to treatment allocation; in cases in which blindness was broken participants were assigned to another research assessor. The fidelity of this masking was moderate, with assessors correctly guessing allocation for 56% of assessments. Given the nature of the interventions, patients and clinicians were aware of treatment allocation.

Patients were randomised in the month before the next MBCT-TS group began. If a participant’s baseline assessment occurred more than a month before the next MBCT-TS group began he or she received a brief ‘randomisation assessment’ before being randomised. During this assessment the patient was asked to give a verbal reaffirmation of his or her wish to take part in the trial and researchers ensured that the patient’s situation had not changed significantly and also obtained a current GRID-HAMD score.

Health technologies assessed

Data collection

Baseline assessments did not occur more than 3 months from the start of the next MBCT-TS group, with the start defined as the first orientation session. Following randomisation, participants were assessed at five time points: 1 month after the end of the 8-week MBCT-TS programme (or the equivalent time in the m-ADM arm), which varied between 12 and 24 weeks post randomisation, and 9, 12, 18 and 24 months post randomisation.

Outcomes

Sample size

Different relapse prevention interventions with different populations produce different absolute rates of depressive relapse/recurrence. Therefore, we based our sample size on estimated HRs for MBCT-TS compared with m-ADM rather than estimated absolute relapse/recurrence rates. We canvassed service users and clinicians who concurred that a relative reduction in relapse/recurrence of 10% would be clinically important. We used the systematic review of MBCT compared with usual care for patients with recurrent depression that reported HRs of 0.28–0.47 for relapse/recurrence. 26 We applied several conservative assumptions (first row of Table 3). First, even though the pilot trial data suggest that the HR was improving in favour of MBCT as the length of follow-up increased,28 we assumed a HR of 0.63 at 15 months to power the trial at 24-months’ follow-up. Second, even though attrition from MBCT trials to date is consistently < 15%, we assumed an attrition rate of 20% over the 24 months of follow-up. Finally, in spite of evidence to the contrary, we assumed that there may be a small clustering effect (ICC = 0.01). The resultant sample size calculation assumptions and outputs are shown in Table 3, based on 90% power and significance set at 0.05. This led to a total sample size of 420 across the two groups.

For the secondary outcomes, meta-analyses of generic mindfulness approaches suggest medium effect sizes in terms of changes in depressive symptoms72,73 and our pilot trial suggested medium effect sizes for the secondary outcomes of residual depressive symptoms, psychiatric comorbidity and quality of life. 28 The sample size estimate for our policy question enabled us to detect a medium between-groups effect size (standardised mean difference or Cohen’s d = 0.40) for the main secondary outcomes.

Statistical analysis

Statistical analyses were conducted in accordance with International Conference on Harmonisation (ICH-9) statistical guidelines for clinical trials74 and updated Consolidated Standards of Reporting Trials (CONSORT) guidelines for trials. 75,76 All statistical analyses were undertaken in Stata version 13 (StataCorp LP, College Station, TX, USA) following a predefined analysis plan agreed with the Trial Steering Committee (TSC).

Baseline equivalence was assessed descriptively by comparing the summary baseline characteristics and outcome values in both groups (MBCT-TS and m-ADM). Although there was a difference in gender between groups (see Table 8), as we know of no strong evidence that gender moderates MBCT treatment outcomes28 it was decided not to include and adjust for this variable in statistical models.

The primary analysis model for all primary and secondary outcomes was conducted on an intention-to-treat (ITT) basis, that is, between-group comparison based on the random allocation of patients and using complete data sets. All models adjusted for the stratification variables (centre and severity of depression). For the primary outcome the primary analysis included all patients and censored for missing data at 24 months’ follow-up. To examine the sensitivity of the results to missing data, secondary outcomes models were also run following multiple imputation (using the ‘ICE’ and ‘MIM’ Stata commands and 10 imputation cycles) and based on the assumption that data were missing at random.

The primary outcome (time to relapse/recurrence) was analysed using a Cox regression proportional hazards model including treatment condition (MBCT-TS/m-ADM). We assessed the proportionality of hazards over time by plotting −ln[−ln(survival)] against ln(analysis time) and tested this using Schoenfeld residuals. 23,24 We found no major violations of the proportional hazards assumption. Secondary outcomes were compared using hierarchical repeated measures regression models adjusting for outcome at baseline.

To explore potential treatment moderator effects, interaction terms were included in the Cox regression model for the primary outcome at 24 months for three predefined subgroups: the two stratification variables (centre and severity of depression) and participants in two groups characterised by how abusive their childhood had been. Participants with a more abusive childhood reported experiencing childhood physical or sexual abuse and/or scored above the median score for the Measure of Parenting Scale (MOPS)77 abuse subscale. Participants completed the MOPS at baseline as part of an embedded process–outcome study. 78 The abuse subscale asks participants to indicate how true they felt certain statements about their parents’ behaviour was, for example ‘parent was physically violent or abusive to me’, ‘parent made me feel unsafe’. Participants were categorised either in the lower abusive childhood group (i.e. scored below the median score for the MOPS abuse subscale and did not report childhood physical or sexual abuse) or in the higher abusive childhood group (i.e. scored above the median score for the MOPS abuse subscale or did report childhood physical or sexual abuse).

Given the variable levels of patient adherence to the stated treatment protocol in both the ADM-m and the MBCT-TC groups, we sought to undertake predefined secondary (per-protocol) analyses to examine the potential impact on the primary outcome at 24 months. We first described adherence in the PREVENT trial according to the principles set out by Dodd et al. 79 in their systematic review exploring adherence reporting in RCTs by reporting the rates of adherence in the format displayed in Box 1.

We then undertook two secondary analyses of the primary outcome comparing groups according to whether participants had (1) received an adequate dose of treatment and (2) adhered to treatment as invited. Definitions of these groups are displayed in Table 4.

Data management

Ethical approval and research governance

Multicentre ethical approval for the study was given by the South West Research Ethics Committee (reference number 09/H0206/43) and local research governance approval was obtained for all sites (NHS Devon covering Exeter and Mid and North Devon – PCT0739; NHS Bristol, covering the Bristol site – 2010–004; NHS Plymouth and NHS Torbay, covering the South Devon site – PLY-TOR001). The trial was registered with the International Standard Randomised Controlled Trial Register with the reference number ISRCTN26666654. This trial was classed as a Clinical Trial of an Investigational Medicinal Product (CTIMP) as the MBCT-TS arm randomised patients to alter their standard ADM and as such approval to commence the trial was received from the MHRA (EudraCT number 2009–012428–10). A summary of the changes made to the original protocol is given in Table 5.

The University of Exeter sponsored the trial, which was hosted in the Mood Disorders Centre [see www.centres.ex.ac.uk/mood/ (accessed 16 March 2015)], a clinical research setting specialising in translational research hosting current MRC-, Wellcome Trust-, National Institute of Mental Health- and National Institute for Health Research Health Technology Assessment (HTA) programme-funded trials. The trial was overseen by the independent TSC (Chris Leach – Chairperson, Richard Moore and Glenys Parry) and the DMC (Paul Ewings – Chairperson, Andy Field and Joanne MacKenzie).

Unexpected serious adverse events

We followed the guidelines laid down by the MHRA for identifying, acting on and reporting adverse events [see www.mhra.gov.uk/Howweregulate/Medicines/Licensingofmedicines/Clinicaltrials/Safetyreporting-SUSARSandASRs/index.htm (accessed 16 March 2015)]. We adopted its definitions of adverse events and serious adverse reactions, reporting a total of 10 serious adverse events, none of which were classified as SUSARs by the TSC or the DMC.

Patient and public involvement

The PREVENT trial has benefited from the expertise of many people with lived experience of mental health difficulties including members of a locally organised voluntary group called the Lived Experience Group (LEG). The LEG has assisted the PREVENT trial at every stage of its development and the following sections detail a few examples of the ways in which the PREVENT trial has been shaped by patient and public involvement.

Participant flow

During the 19 months of recruitment, between 23 March 2010 and 21 October 2011, we identified 28,597 patients from GP practice searches. GPs excluded 8989 patients as unsuitable and invitation letters were sent to the remaining 19,608 patients. Although we asked all GPs to provide reasons for excluding patients, in 56% of cases this information was missing. For the remaining exclusions the most common reasons given were dementia (2%), psychosis (2%) and substance abuse (1%). In total, 3060 patients responded positively to our invitation and a further 89 patients self-referred. Full telephone screens were completed for 2188 patients, which resulted in 498 patients attending for a baseline assessment, of whom 424 patients were randomised. The most common reasons why patients were not eligible for the PREVENT trial were that they had recently stopped taking ADMs or wanted to reduce them (30%), they had not had three previous episodes of depression (19%) or they were currently depressed (17%). The reasons given for not wishing to take part in the PREVENT trial are explored in Chapter 7. Primary outcome data were collected for 90.3% (383/424) of participants and the remaining participants’ data were censored at their last assessment. We retained 86.3% (366/424) of participants over the 24-month follow-up period, with 4.7% (20/424) lost to contact, 8.0% (34/424) withdrawing consent for further follow-up and 0.9% (4/424) dying during the trial; the pattern of missing data was similar across interventions. The flow of participants through the trial is depicted in Figure 3.

FIGURE 3.

The CONSORT flow chart illustrating the flow of participants into the PREVENT trial.

Missing data

The missing data rates for each time point are shown in Table 6.

Baseline characteristics

Of the 424 randomised participants, 212 were allocated to receive MBCT-TS and 212 were allocated to receive m-ADM. As indicated in Table 7, baseline characteristics were balanced between the two groups with the possible exception of gender. As we know of no strong evidence that gender moderates MBCT treatment outcomes28 we did not add gender to the primary analysis model. It is interesting to note that the psychiatric history of these participants differs in a number of ways from the history of those randomised in the pilot trial (Table 8),28 with PREVENT participants reporting lower BDI scores and fewer comorbidities and a smaller proportion previously attempting suicide. A much larger percentage of the PREVENT participants also previously accessed psychiatric treatment, which is likely to be the result of the recent significant progress made in improving the accessibility of evidence-based mental health services in the UK. 4,15,16

Treatment adherence in each trial arm and the extent to which patients followed invitations to (dis)continue m-ADM are reported in Table 9; > 75% of patients adhered to treatment as intended. Details of the ADM that was taken are provided in Appendix 2.

Primary outcome

Intention-to-treat analysis

We observed little or no clustering in primary or secondary outcomes by therapist. As model results accounting for clustering by therapist were identical to those obtained for the primary ITT analysis, outcome findings without consideration of therapist clustering are reported.

With respect to the primary outcome, the primary ITT analysis showed no evidence of a reduction in the hazard of relapse/recurrence with MBCT-TS compared with m-ADM (HR 0.89, 95% CI 0.67 to 1.18; p = 0.43), with 44% (94/212) of the MBCT-TS patients relapsing compared with 47% (100/212) of the m-ADM patients (Figure 4).

FIGURE 4.

Survival (non-relapse/recurrence) curves comparing relapse/recurrence of major depression for the MBCT-TS and m-ADM groups over the 24-month follow-up period for ITT participants.

Secondary analyses

Two secondary analyses of the primary outcome were undertaken to explore the impact of variations in intervention adherence in the MBCT-TS and m-ADM groups.

There was a non-significant reduction in the hazard of relapse/recurrence with MBCT-TS compared with m-ADM at 24 months in those participants who received an adequate dose of treatment (HR 0.79, 95% CI 0.58 to 1.08; p = 0.14), with 46% (81/176) of the MBCT-TS patients relapsing compared with 49% (80/162) of the m-ADM patients (Figure 5). In this subgroup, the m-ADM group included more women and participants with a greater number of comorbidities than the MBCT-TS group (see Appendix 3).

FIGURE 5.

Survival (non-relapse/recurrence) curves comparing relapse/recurrence of major depression for the MBCT-TS and m-ADM groups over the 24-month follow-up period for those participants who received an adequate dose of treatment.

There was a non-significant reduction in the hazard of relapse/recurrence with MBCT-TS compared with m-ADM at 24 months in participants who followed the invited treatment with respect to ADM use (HR 0.77, 95% CI 0.56 to 1.06; p = 0.10), with 46% (70/153) of the MBCT-TS patients relapsing compared with 49% (80/162) of the m-ADM patients (Figure 6). In this subgroup, the m-ADM group included younger participants, lower earners, more women and participants with a greater number of comorbidities than the MBCT-TS group (see Appendix 4). Given their non-randomised nature, these secondary analyses are prone to selection bias and confounding.

FIGURE 6.

Survival (non-relapse/recurrence) curves comparing relapse/recurrence of major depression for the MBCT-TS and m-ADM groups over the 24-month follow-up period for those participants who followed the invited treatment with respect to ADM use.

There was no difference in treatment effect on the primary outcome across either stratification variable subgroup (severity of depression at baseline or centre) (Table 10). However, there was evidence of a significant interaction between the severity of reported childhood abuse and treatment group (HR 0.53, 95% CI 0.29 to 0.95; p = 0.03). Specifically, compared with ADM-m, MBCT-TS reduced the risk of relapse/recurrence for participants with a higher severity of reported childhood abuse (MBCT-TS 47% vs. ADM-m 59%) whereas there was a slightly higher risk of relapse/recurrence with MBCT-TS in the lower severity of reported childhood abuse subgroup (MBCT-TS 42% vs. ADM-m 35%) (see Table 10). Comparing the baseline characteristics of those with high and low severities of reported childhood abuse we find a number of differences. Participants who reported a more abusive childhood had had more previous psychiatric treatments, including more hospitalisations, had experienced more previous episodes of depression and made more suicide attempts, had a greater chance of a familial history of both suicide and mental illness and were more likely to smoke (see Appendix 5). Given their non-randomised nature, these secondary analyses are prone to selection bias and confounding

TABLE 10 - Treatment effect subgroup analyses

HRSD, Hamilton Rating Scale for Depression.

Reference subgroup.

p-value for treatment–centre interaction across centres.

Subgroup	Stratified HR (95% CI)	Interaction HR (95% CI), p-value
Depression severity
Asymptomatic (HRSD score < 8)a	0.83 (0.60 to 1.15)	1.27 (0.68 to 2.39), 0.46
Symptomatic (HRSD ≥ 8)	1.06 (0.62 to 1.18)
Centre
South Devona	0.61 (0.33 to 1.13)	1.75 (0.70 to 4.39)
Bristol	1.60 (0.54 to 2.12)	1.81 (0.83 to 3.96)
Exeter and East Devon	1.10 (0.68 to 1.81)	1.37 (0.61 to 3.08), 0.47b
North and Mid Devon	0.84 (0.49 to 1.43)
Childhood abuse
Lower riska	1.31 (0.83 to 2.04)	0.53 (0.29 to 0.95), 0.03
Higher risk	0.69 (0.47 to 1.00)

Secondary outcomes

With respect to the secondary outcomes, there was again no evidence of MBCT-TS’s superiority over m-ADM (Table 11). Furthermore, none of the secondary outcome treatment effects at any follow-up point exceeded a standardised mean difference of 0.4. The health economics outcomes are reported in full in Chapter 5.

Adverse events

A total of 10 serious adverse events were reported and are summarised in Table 12. Following discussion with the centre principal investigators and confirmation from the DMC we concluded that there was no reason to believe that any of the serious adverse events were intervention or trial related.

Over the 24 months there was no difference in the total number of suicidal ideations reported between the MBCT-TS group (n = 44 events) and the m-ADM group (n = 46 events) (rate ratio 0.94, 95% CI 0.64 to 1.38; p = 0.75).

Introduction

Methods

Results

Total costs

Changes over time

Changes in total cost per week over follow-up are summarised in Figure 7 (NHS/PSS perspective) and Figure 8 (societal perspective). The charts demonstrate that costs were fairly consistent between groups and over time, suggesting that randomisation to MBCT-TS or m-ADM had little impact on the cost of health and social care services and productivity losses.

FIGURE 7.

Average cost per week at each follow-up using a NHS/PSS perspective.

FIGURE 8.

Average cost per week at each follow-up using a societal perspective.

Total costs at follow-up

Total costs over follow-up are summarised in Table 16, including a breakdown of costs by service-providing sector. From a NHS/PSS perspective, the mean total cost over 24 months’ follow-up was £2360 in the m-ADM group and £2485 in the MBCT-TS group; the mean difference in costs was not statistically significant (p = 0.80). From a societal perspective the mean total cost over 24 months’ follow-up was £2755 in the m-ADM group and £3204 in the MBCT-TS group; again, the mean difference in costs of £449 was not statistically significant (p = 0.68).

TABLE 16 - Total costs (£) and outcomes over 24 months’ follow-up by randomised group

Cost category	MBCT-TS (n = 181)		m-ADM (n = 180)		Mean difference	95% CI	p-value
	Mean	Mean	Mean	SD
MBCT-TS	112.00	112.00	0.00	0.00
Antidepressants	40.10	72.13	69.79	168.48
Hospital and community services	2332.43	4065.88	2290.62	4190.65
Total health-care costs (NHS/PSS)	2484.52	4077.31	2360.41	4205.58	124.11	–749.98 to 972.57	0.80
Out-of-pocket costs to patients	56.76	168.29	83.33	283.12
Productivity losses (n = 265)	504.26	1881.49	310.54	761.06
Societal costs (n = 252)	3204.05	4011.91	2754.92	4465.07	449.14	–842.18 to 1286.26	0.68
Relapse/recurrence (%) (n = 402)	47	47	50	50	–3
QALYs (n = 213)	1.49	1.49	1.53	0.35	–0.04

Outcomes

Health-related quality of life

Mean EQ-5D health state scores are summarised in Figure 9. The graph shows that the average health state score was between 0.7 and 0.8 for both groups for the entire period between baseline and 24 months’ follow-up and that there is very little difference between the groups. The resultant QALYs over follow-up are summarised in Table 16; the difference of –0.04 between the groups was not statistically significant (p = 0.42, adjusted by stratification variables).

FIGURE 9.

Mean EQ-5D health state scores over follow-up.

Relapse/recurrence

Detailed information on the relapse/recurrence outcome measure is provided in Chapter 4. For the purpose of the economic evaluation we used the percentage of individuals who relapsed in each of the randomised groups. In the m-ADM group 50% had a relapse/recurrence over follow-up and in the MBCT-TS group this figure was 47% (see Table 16).

Cost-effectiveness

Relapse/recurrence

In terms of relapse/recurrence, costs were higher in the MBCT-TS group and outcomes better, generating an ICER (the additional cost of one intervention compared with another divided by the additional effects) of £4955 from the NHS/PSS perspective and £17,930 from the societal perspective, including productivity losses and patient costs. The ICER indicates that an additional £4955 (from the NHS/PSS perspective) or £17,930 (from the societal perspective) would need to be invested to generate a unit reduction in the percentage of participants who relapse.

Non-parametric bootstrapping from the cost and effectiveness data was used to generate a joint distribution of incremental mean costs and effects for the treatments under comparison to explore the probability that each is the optimal choice, subject to a range of maximum values (λ) that a decision-maker might be willing to pay for improvements in outcomes. Figures 10 and 11 illustrate the scatterplots of the bootstrapped cost and effectiveness pairs for MBCT-TS compared with m-ADM, from the NHS/PSS and societal perspectives respectively. The points in the scatterplot fall in all four quadrants of the cost-effectiveness plane suggesting that there is no clear conclusion to be made regarding cost-effectiveness in terms of relapse/recurrence. It is important to note that, as positive values of relapse/recurrence correspond with worse outcomes, the usual interpretation of the quadrants of the cost-effectiveness plane is reversed with respect to the x-axis so that points that fall to the left of 0 denote better outcomes for MBCT-TS and those that fall to the right denote worse outcomes. Statistical uncertainty around the ICER was explored through the calculation of CEACs, shown in Figure 12, which demonstrate that the probability of MBCT-TS being more cost-effective than m-ADM does not rise above 43%.

FIGURE 10.

Scatterplot showing the bootstrapped mean differences in costs and effects of MBCT-TS compared with m-ADM using relapse/recurrence and service costs.

FIGURE 11.

Scatterplot showing the bootstrapped mean differences in costs and effects of MBCT-TS compared with m-ADM using relapse/recurrence and societal costs.

FIGURE 12.

Cost-effectiveness acceptability curves showing the probability that MBCT-TS is cost-effective compared with m-ADM for different values that a decision-maker is willing to pay for a unit reduction in the percentage of patients who undergo relapse/recurrence.

Quality-adjusted life-years

In terms of QALYs, costs were higher in the MBCT-TS group and outcomes slightly worse, suggesting that MBCT-TS was dominated by m-ADM. The scatterplots in Figures 13 and 14 show points falling mainly in the north-west quadrant (outcomes worse and costs higher) and south-west quadrant (outcomes worse and costs lower). The probability that MBCT-TS is more cost-effective than m-ADM, shown in the CEACs in Figure 15, does not rise above 52%.

FIGURE 13.

Scatterplot showing the bootstrapped mean differences in costs and effects of MBCT-TS compared with m-ADM using QALYs and service costs.

FIGURE 14.

Scatterplot showing the bootstrapped mean differences in costs and effects of MBCT-TS compared with ADM-T using QALYs and adjusted societal costs.

FIGURE 15.

Cost-effectiveness acceptability curves showing the probability that MBCT-TS is cost-effective compared with m-ADM for different values that a decision-maker is willing to pay for a QALY.

Sensitivity analysis

Imputation of missing data and variation in the discount rate from 0% to 6% made no difference to the results, as summarised in Table 17.

TABLE 17 - Total costs and QALYs with discount rate varied from 0% to 6%

Cost category and discount rate	MBCT-TS (n = 181)		m-ADM (n = 180)		Mean difference	95% CI	p-value
	Mean	SD	Mean	SD
Total NHS/PSS costs (£), discount rate 0%	2439.98	3992.68	2319.75	4120.03	107.96	–735.63 to 951.55	0.80
Total societal costs (£), discount rate 0%	3007.99	3970.62	2747.60	4468.00	214.72	–844.97 to 1274.42	0.69
Total NHS/PSS costs (£), discount rate 6%	2516.34	4137.95	2389.46	4267.05	113.68	–760.28 to 987.64	0.80
Total societal costs (£), discount rate 6%	3035.53	4041.78	2760.14	4468.15	227.27	–840.25 to 1294.78	0.68
QALYs, discount rate 0%	1.48	0.48	1.52	0.35	–0.05	–0.16 to 0.07	0.42
QALYs, discount rate 6%	1.46	0.47	1.5	0.35	–0.05	–0.16 to 0.06	0.42
Multiple imputation of missing data, NHS/PSS costs (£)	2423.68	3780.54	2348.59	3903.41	67.13	–668.39 to 802.65	0.858
Multiple imputation of missing data, societal costs (£)	2934.68	3177.99	2692.53	3595.67	229.16	–417.93 to 876.25	0.487

Introduction

Establishing mediation

To establish mediation requires attention to several key aspects of study design103,105,106 that have been instantiated in the PREVENT trial. First, MBCT is compared with a treatment that works but not through the same mechanism of action (m-ADM),107 thus allowing a test of effects specific to MBCT. Second, assessment of change in the hypothesised mediator occurred (1) during MBCT and (2) before the assessment of outcome. For the current approach we assessed change in mindfulness at baseline and immediately following MBCT and depressive relapse/recurrence over 24 months of follow-up was the dependent variable in our mediation analyses. Finally, the design requires that all those in the intervention arm received an adequate dose of the intervention to properly test the hypothesis that MBCT’s impact on the hypothesised mechanism (mindfulness) mediates outcome. Therefore, only patients who attended four or more sessions were included in the mediational analyses.

Hypothesis

Based on our previous work37 and on the theoretical and clinical rationale of MBCT, we predicted that enhanced mindfulness over the active treatment period would significantly mediate outcome, with those participants evidencing larger gains in mindfulness faring better. We did not predict an interaction with treatment arm (see Mediation approach).

Methods

Analytic approach

Results

Discussion

The analyses presented in this chapter sought to address the following explanatory question: ‘Is an increase in mindfulness skills the key mechanism of change of MBCT-TS?’ We approached this question by examining whether or not change in mindfulness from baseline to 1 month post MBCT-TS (or the equivalent time point in the m-ADM arm), as measured by the self-report FFMQ, was a significant mediator between trial arm (MBCT-TS or m-ADM) and outcome as assessed by depressive relapse/recurrence over 24 months.

We found no support in the data for either main or interactive mediation effects, somewhat contrasting with our earlier finding from the PREVENT pilot trial37 that change in mindfulness (assessed using a different measure) mediated effects on residual symptoms of depression at 12 months.

Supplementary analyses examined whether or not change in mindfulness was significantly associated with clinical outcome (time to relapse/recurrence) within the MBCT-TS group alone. Again, we found no support for this prediction.

Future work will systematically explore whether or not there are any mediating effects of mindfulness on secondary trial outcomes or whether or not any putative mediating effects are moderated by key demographic variables (e.g. exposure to childhood abuse).

Introduction

Qualitative research within RCTs can provide a patient-centred and in-depth perspective on the implementation of complex interventions and on trial processes and outcomes. 110 The PREVENT trial offers an opportunity to build on previous qualitative work on MBCT in three important ways: (1) the follow-up period (2 years) is longer than that in many other studies; (2) participant experiences can be explored across larger numbers than in previous qualitative studies; and (3) these large numbers provide opportunities for purposive sampling. Previous qualitative work on MBCT has included very small numbers111,112 and sociodemographically homogeneous samples. 30 As a consequence, analysis has tended to focus on commonalities, with little exploration of variations in experiences of MBCT. The large number of participants in the PREVENT trial allows exploration of both similarities and variations in their experiences.

One area of focus for the qualitative exploration was the acceptability of MBCT-TS (other qualitative work within the PREVENT trial that is not reported here is listed at the end of the chapter). There are several reasons for interest in acceptability issues: (1) attrition between possible case identification and recruitment in MBCT trials is typically high;28 (2) between 7% and 25% of people do not complete the full MBCT course and a 15% dropout rate from MBCT is typical;24,25,28 and (3) there are wide variations in the extent to which people engage with mindfulness practices that are assumed to be the vehicle of change. 112 Existing evidence about the acceptability of MBCT is limited111 and there is no published evidence on the acceptability of MBCT-TS. Little is known about several elements of acceptability: why people choose not to participate in MBCT; reasons for dropout from treatment; and why people fail to engage with or continue mindfulness practices. The research reported here focuses on the first of these issues, namely barriers to participation in MBCT-TS within the PREVENT trial, at the point of initial invitation into the trial (other acceptability questions are addressed by data not reported here).

Methods

Development of data collection tools drew on the perspectives of both PREVENT MBCT therapists and members of the trial LEG (see Chapter 3). This helped to enhance the relevance and acceptability of data collection to patients and ensured that it was informed by the clinical judgements of MBCT therapists.

Results

Reasons for declining participation at initial contact

From a total of 19,608 invitations sent, 2157 people (11%) indicated that they were not interested in participating in the PREVENT trial either by declining any further contact or after further discussion with a member of the research team. Within this sample, 1535 (71%) participants were female and the average age was 58.6 years (range 20–91 years; missing data for 25 cases). The sample was spread across the different geographical areas: South Devon n = 490 (23%); North and Mid Devon n = 788 (37%); Exeter and East Devon n = 580 (27%); and Bristol n = 299 (14%). These characteristics are similar to those of the final sample of PREVENT trial participants (see Table 7) with the exception that the PREVENT trial population was younger on average (mean age 49 years). Education and marital status are not reported as fewer than half of the respondents provided this information.

Because a minority of people gave more than one reason for not participating, responses added up to an overall number of 2402 instances of coding. Of these, 120 codings (5%) were coded not eligible for participation (mainly when participants did not fulfil trial inclusion criteria such as having three episodes of depression or a diagnosis other than depression). Hence their responses were excluded from further analysis. Furthermore, some 1109 (46%) codings did not reveal any reasons, yielding a final number of valid codings of 1173 (49%).

Content analysis produced 10 categories of reasons, the distribution of which within the data is shown in Figure 16. We further categorised these reasons into four broad areas as follows:

1. intervention-related issues – these are about use of ADM, therapy-related issues and the group-based nature of the intervention

2. personal circumstances – three categories of reasons relating to personal circumstances were found: lifestyle issues, time issues and medical issues

3. symptom-related issues – reasons relating to current symptoms and feelings of being in recovery

4. research-related and other issues.

FIGURE 16.

Frequencies of reasons for non-participation provided in written responses on invitation reply forms.

Intervention-related issues

The most commonly cited reasons for non-participation were related to the treatment interventions provided in the PREVENT trial. Together these accounted for 40% of all reasons given. Within this, the largest category related to use of ADM (19% of all responses). Most commonly, people reported that they did not want to stop taking ADM (49% of ADM reasons). Other reasons were that people were no longer taking ADM (24%), were currently coming off ADM (9%) and were happy with their current ADM use (11%). The second most common category of reasons overall related to therapy issues (14% of all responses). In total, 23% of responses in this category related to people having previously tried talking therapies and a further 17% of responses related to reports that past therapy had not helped. Some reported that they were currently receiving a different kind of therapy (19%) and others reported wanting to avoid bringing up bad memories (16% of responses in this category). Related to this, a third category of responses was about reluctance to participate, or anxiety about participating, in group-based therapy (7% of all responses).

Personal circumstances

In total, 35% of responses described personal circumstances as a barrier to participation. The category of lifestyle issues accounted for 13% of responses. Specific reasons within this category included feeling too old (23%), moving house or relocating (17%), transport difficulties (14%) and feeling that participation would mean taking on too much (15%). Health-related barriers were also common (given in 12% of responses overall). Most frequently, people reported their own physical or health difficulties as a reason preventing participation (47%), but caring for others (21%) and illness or death of a family member (15%) were also cited. In total, 10% of responses overall cited lack of time, with one-quarter of these participants specifically stating that they would be unable to take time off work.

Symptom-related issues

Symptom-related reasons for non-participation were cited in 15% of responses. Specific reasons within this included considering that they suffered from anxiety more than depression (18%), feeling that they did not suffer from depression (14%), that they were currently depressed (17%), that they were currently feeling well or recovering (27%) or that they already had strategies in place to manage their depression (12%).

Research-related and other issues

The final 10% of responses referred to research-related issues (3%) and miscellaneous other reasons that could not be categorised in other ways (7%). Just over half of the research-related reasons were that people did not want to be involved in research or had concerns about this. Another common reason was already being involved in research, currently or in the past.

Discussion

Principal findings

The overarching aim of the PREVENT trial was to establish whether or not MBCT-TS provides an effective alternative relapse/recurrence prevention approach to m-ADM in primary care settings for patients with a history of recurrent depression. Specifically, we asked a primary policy research question: ‘Is MBCT-TS superior to m-ADM in terms of a primary outcome of preventing depressive relapse/recurrence over 24 months?’ Secondary outcomes were depression-free days, residual depressive symptoms, psychiatric and medical comorbidities, quality of life and cost-effectiveness. The study was conducted in line with both CONSORT guidance74–76 and our published protocol. 78,83 Participant recruitment, flow and data completeness were above estimated thresholds. We are, therefore, able to answer the key research questions.

There was no evidence for the superiority of MBCT-TS over m-ADM for patients with recurrent depression in terms of either the primary outcome of time to depressive relapse/recurrence over 24 months or any of the secondary outcomes (see Chapter 4). The cost-effectiveness analysis does not support the hypothesis that MBCT-TS is more cost-effective than m-ADM, in terms of either relapse/recurrence or QALYs (see Chapter 5).

In a predefined subgroup analysis,83 we found that time to relapse/recurrence at 24 months for participants with a higher severity of reported childhood abuse was delayed in the MBCT-TS arm, compared with the m-ADM arm.

Finally, the first results from our process studies suggest that, although changes in mindfulness were greater in the MBCT-TS group than in the m-ADM group, they did not mediate time to relapse/recurrence at 24 months (see Chapter 6). In terms of acceptability, the qualitative analyses suggest that people have views about (dis)/continuing their ADM that can serve as a facilitator and barrier to taking part in a trial that requires either continuation for 2 years or discontinuation (see Chapter 7).

Research findings in context

Relapse/recurrence rates in people with three or more previous episodes are as high as 80% over 2 years. 6 Moreover, meta-analyses consistently suggest that m-ADM reduces the relative risk of relapse/recurrence by two-thirds compared with placebo, a halving of the absolute risk. 9 Therefore, it is likely that MBCT would provide benefits over and above either no treatment or pill placebo.

Outcomes were comparatively good across both treatment arms in the PREVENT trial in terms of relapse/recurrence over the 2 years of follow-up compared with those in previous trials (i.e. MBCT-TS 44%, m-ADM 47%). Moreover, PREVENT trial participants reported residual depressive symptoms within the minimal range and quality of life in the range ‘good’ at each follow-up point. This suggests that participants in both arms experienced relatively good mental health and quality of life over the follow-up period.

Before the PREVENT trial, only two small studies had compared MBCT-TS with m-ADM. In our pilot trial with 123 participants, MBCT-TS (n = 62) was compared with m-ADM (n = 61) over 15 months’ follow-up. 28 The relapse/recurrence rate was 47% for MBCT-TS and 60% for m-ADM. In a second study, 84 patients with recurrent depression who had remitted on ADM were randomised to MBCT-TS, m-ADM or pill placebo. 51 Relapse/recurrence rates observed over 18 months of follow-up did not differ between MBCT-TS (28%) and m-ADM (27%; p = 0.93) and both MBCT-TS and m-ADM were superior to placebo (71%; p = 0.007).

Since the start of the PREVENT trial, a key systematic review and meta-analysis has been published that included six RCTs comparing MBCT with usual care, m-ADM or placebo (including the two previously described trials) (n = 593). The pooled estimate from this review shows that MBCT reduces the risk of relapse/recurrence compared with usual care or placebo (risk ratio 0.66, 95% CI 0.53 to 0.82). 116

Since this systematic review/meta-analysis, several additional large-scale trials of MBCT have been published117–119 or are due to be published soon. 120 The Staying Well After Depression (SWAD) trial compared MBCT with cognitive psychoeducation plus usual care and usual care alone in 274 participants currently in remission but reporting at least three previous episodes of depression. 117 The SWAD trial reported no significant effect of treatment on risk of depressive relapse/recurrence over the 12-month follow-up (MBCT vs. psychoeducation: HR 0.88, 95% CI 0.58 to 1.35; MBCT vs. usual care: HR 0.69, 95% CI 0.42 to 1.12). In another recent RCT, 203 non-depressed adults with a history of three or more previous episodes of depression were randomised to MBCT plus depression relapse active monitoring (i.e. training in self-management of depression and monthly monitoring of symptoms) (n = 101) or active monitoring alone (n = 102). 119 The DARE trial found no significant difference between the two treatment arms for the primary outcome of time to first relapse/recurrence over 24 months, although there was some evidence that proportionally fewer people relapsed in the MBCT arm than in the active monitoring alone arm (odds ratio 0.68, 95% CI 0.38 to 1.23). It is noteworthy that the absolute rates of relapse/recurrence in the usual care arms of these two more recent trials are lower than those seen in earlier RCTs. In total, 66% relapsed in the usual care arms in the Piet and Hougaard systematic review116 compared with approximately 50% in the usual care arms in the DARE119 and SWAD117 trials at 12 months’ follow-up.

Economic evaluation found group-based MBCT-TS to be a relatively inexpensive intervention compared with individual therapies. In line with previous results from our pilot study,28 there was little difference in the use of other health and social services between the groups, resulting in only a small difference in the total costs of care over 24 months between the two groups. Coupled with small, non-significant differences in outcome, MBCT-TS failed to demonstrate cost-effectiveness compared with m-ADM.

Consistent with an emergent pattern of findings,117 MBCT may confer most benefit to patients at greatest risk of relapse/recurrence. Earlier trials24,25 and NICE guidance8 suggest that MBCT is indicated for patients who have a history of three or more previous episodes of depression. In the SWAD trial,117 a subgroup analysis compared participants reporting greater or lesser childhood adversity. For participants reporting greater childhood adversity MBCT conferred better protection against relapse/recurrence than both psychoeducation and usual care (MBCT vs. psychoeducation: HR 0.61, 95% CI 0.34 to 1.09; MBCT vs. treatment as usual: HR 0.43, 95% CI 0.22 to 0.87). The PREVENT trial results replicate the SWAD finding and converge with those of earlier studies suggesting that patients at greatest risk of depressive relapse/recurrence may benefit most from MBCT. Trials of other psychosocial approaches have shown that more intensive psychosocial treatments result in greater protection for those most at risk. For example, in a two-arm RCT with 21 months’ follow-up, relapse/recurrence rates were 51% for maintenance CBT and 60% for psychoeducation but, among those at greatest risk, CBT conferred greater protection than psychoeducation. 121 A reported history of abuse and adversity is associated with worse outcomes among people who suffer depression. 122 Perhaps MBCT confers resilience in this group at highest risk because patients learn skills that address some of the underlying mechanisms of relapse/recurrence, a question that we will seek to explore in a subsequent publication from this trial. Future trials of MBCT need to focus on the question of the effectiveness and mechanism of action of MBCT for those at differing risk of relapse/recurrence, with broad and robust measures of risk.

Strengths and limitations

Strengths of the PREVENT trial include the following.

• The pragmatic main study question is of high relevance to the NHS. The study answered an important clinical question of high relevance to GPs and patients at risk of depressive relapse/recurrence.

• The trial was a multicentre RCT conducted in accord with both CONSORT guidance and our published protocol/statistical analysis plan.

• This is the largest trial of MBCT for recurrent depression to date. Good recruitment/retention rates and low levels of missing data mean that the trial was adequately powered and able to answer the primary research question.

• The study benefited from a comprehensive economic evaluation, providing evidence to support resource allocation decision-making of both clinical and policy relevance.

• Treatment fidelity with respect to both arms of the trial was designed into the trial and high levels of fidelity ensured high internal validity.

• The 2-year follow-up allowed assessment of the impact of MBCT-TS over a time period that was comparable to or longer than that in all MBCT trials completed to date.

• A parallel process evaluation was undertaken to provide contextual understanding and examine the mechanism of action of MBCT.

Limitations of the PREVENT trial include the following.

• Our recruitment strategy involved searching primary care databases and inviting patients who were currently taking m-ADM rather than recruiting patients who were discussing their options with their GP for preventing relapse/recurrence.

• The design included neither a usual care nor an attention control arm. The absence of an attention control arm means that any effects of MBCT-TS or m-ADM cannot be inferred to be specific to these treatments.

• The m-ADM arm included active monitoring of adherence by the research team and in this sense might best be represented as enhanced m-ADM.

The pragmatic nature of the trial resulted in a proportion of patients in both arms not complying with the invitation to (dis)continue ADM. We undertook a per-protocol analysis to examine the impact on the primary outcome inference compared with ITT analysis. This is both a strength (pragmatism and generalisability) and a limitation (the ADM was not completely controlled). Finally, the sample consisted of a group at high risk of depressive relapse/recurrence,123 currently taking ADM, who were open both to considering a group-based psychosocial treatment and to (dis)continuing their ADM. This is both a strength and limitation of the study. The findings of the PREVENT trial are therefore generalisable to only those individuals who are in equipoise about the type of preventative treatment that they choose, that is, m-ADM or switch to psychosocial intervention and reduce their ADM.

Implications for practice

Depression is a major public health problem that tends to run a recurrent course, producing substantial decrements in health and well-being. The cost of mood disorders to the UK economy is substantial. Most of the prevalence, burden and cost of depression is a consequence of recurrent depression and could be offset through providing a range of psychosocial low- and high-intensity treatments. 4

Mindfulness-based cognitive therapy with support to taper/discontinue ADM may confer ongoing protection for patients who would like an alternative to m-ADM. Many patients with recurrent depression may have been on m-ADM for many years and for a variety of reasons would prefer to learn skills that they can use to stay well. The results further suggest that psychosocial treatments such as MBCT and CBT117,121,124 offer added value for patients who need them most, that is, those at highest risk of depressive relapse/recurrence. The investment of effort by these patients in learning the skills taught in psychosocial treatments such as CBT and MBCT pays off. However, for patients at low risk, treatments such as psychoeducation or m-ADM, which require less patient commitment and cost, may be indicated. This has significant potential to improve prevention by maximising the delivery of treatments through stratified approaches, which also have the potential to improve patient choice. As patients with recurrent depression select options to stay well in the long term, these findings taken together with the broader emergent literature suggest a number of alternatives. It is clear that m-ADM is an effective mainstream approach. For patients at low risk of relapse low-intensity interventions may be indicated and for patients at high risk high-intensity treatments such as MBCT may be indicated.

Future research

There are several key areas for future research:

• Given the completion of the PREVENT trial and the recent publication of other large RCTs, an updated meta-analysis is indicated to re-examine evidence for MBCT as a strategy for the prevention of depressive relapse/recurrence in the context of alternative treatments that include usual care, other active treatments such as m-ADM and additive therapy approaches (e.g. MBCT plus m-ADM). An alternative strategy would be to conduct a large cohort study or effectiveness study to examine the effectiveness of MBCT and m-ADM in various real-world scenarios.

• Furthermore, an individual patient data meta-analysis of published trials would help address the question of which specific patient subgroup MBCT is best indicated for. There is emerging evidence that MBCT is most effective for those at highest risk of relapse/recurrence and an individual patient data analysis could provide the empirical evidence to address this question.

• Studies have tended to operationalise risk in somewhat different ways (e.g. early adversity, unstable remission, a greater number of previous episodes, early age of onset) and, although these risk factors overlap, future research should examine how and through what mechanism risk is conferred and resilience learned.

• In addition to further analysis of the process data from the PREVENT trial and recently completed large MBCT RCTs, future trials need to incorporate process studies and consider dismantling designs to further unpack MBCT’s mechanism of action. This could enhance the specificity and efficacy of MBCT.

• Given the current NICE guidance that MBCT should be offered to patients with three or more previous episodes of depression, and the limited availability of MBCT within the NHS, research is needed to explore the facilitators and barriers to implementation. This National Institute for Health Research-funded work is in progress. 22

Conclusions

In a large, rigorous yet pragmatic RCT we have demonstrated that MBCT-TS is not superior to m-ADM over 2 years of follow-up for patients with recurrent depression. Benchmarked against epidemiological data, both treatments were associated with enduring positive outcomes in terms of relapse/recurrence, residual depressive symptoms and quality of life. This study provides important evidence that MBCT-TS may confer ongoing protection for patients who would like an alternative to m-ADM. The results further suggest that psychosocial treatments, such as MBCT and CBT,117,121,124 offer added value for patients who need them most, that is, those at highest risk of depressive relapse/recurrence. However, studies have tended to operationalise risk in somewhat different ways (e.g. early adversity, unstable remission, a greater number of previous episodes, early age of onset) and, although these risk factors overlap, future research should examine how and through what mechanism risk is conferred and resilience learned. In the interim, the implication is that, for patients at low risk, treatments such as psychoeducation or m-ADM, which require less patient commitment and cost, may be indicated, whereas for patients at highest risk more intensive treatments such as MBCT may be indicated. This has significant potential to improve prevention by maximising the delivery of treatments through stratified approaches, which also have the potential to improve patient choice.

Contributions of authors

Willem Kuyken, Sarah Byford, Richard Byng, Tim Dalgleish, Glyn Lewis, Rod Taylor, Edward Watkins, David Kessler and Nicola Morant were responsible for the original proposal, securing funding for the trial and drafting the original protocol.

Willem Kuyken as chief investigator had overall responsibility for the management of the study and the Exeter site and Glyn Lewis and David Kessler as co-investigators had responsibility for the Bristol site.

Willem Kuyken provided training and supervision for the trial therapists.

Rachel Hayes, Willem Kuyken, Rod Taylor and Sarah Byford wrote the statistical analysis plan.

Rod S Taylor conducted the main analyses and Sarah Byford and Barbara Barrett the health economics analyses. Richard Byng provided support for intervention development and delivery.

Rachel Hayes, Sarah Byford (see Chapter 5), Tim Dalgleish (see Chapter 6), Willem Kuyken, Nicola Morant (see Chapter 6) and Rod Taylor (see Chapter 4) wrote the initial draft of the report.

All authors contributed to, and approved, the final manuscript.

Data sharing statement

Non-identifiable data will be made available following an application to the corresponding author Professor Willem Kuyken.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health.

Serious adverse event form

(PDF download)

Suicidal thoughts protocol

(PDF download)

FIGURE 17.

Researcher action required – including suggested scripts.

(PDF download)