Study design
We will conduct a parallel group randomised controlled design in which half our participants receive Group Mindfulness Based Cognitive Therapy and half receive Group Supportive Therapy. We will implement a computerised block randomisation procedure to ensure equal distribution of participants across conditions (Fig. 2). As far as practically possible participants will be blind to group allocation. Self-report, behavioural and physiological measures will be administered at pre-treatment, mid-treatment, post-treatment, and 3 and 6-month follow-up. To maximise external validity, a minimum follow-up period of 6 months is recommended and will be adopted in the present study [26]. Baseline assessments and follow up assessments will be conducted by a researcher blind to group allocation.
Participants and recruitment
We will recruit 96 participants (48 per group) via our existing referral networks of GPs, mental health professionals, and university counselling services, as well as through advertising in print media and social network sites. Inclusion criteria include: 1) Aged 18–25 years, and 2) meet proposed DSM-5 criteria for NSSI [8]. Participants will be excluded if they: 1) are currently receiving psychological treatment, 2) have attempted suicide in the previous 12 months, 3) exhibit acute psychosis, 4) have a diagnosis of borderline personality disorder (BPD), or 5) have prior experience of MBCT. As NSSI is one diagnostic criterion for BPD, young people diagnosed with BPD will be excluded to avoid confounding diagnosis and behaviour. Individuals presenting with suicidal behaviour, psychosis or BPD will be appropriately referred.
Intervention
A trained psychologist will conduct telephone interviews to assess eligibility with the single-item Clinician-Rated Severity of Non-Suicidal Self-Injury [27] and 6-item Mini International Psychiatric Interview [28]. Both treatment conditions will comprise 8 weekly group sessions of 2 h duration with up to 12 participants in each group. All group sessions and assessments will occur at the Curtin University Psychology Clinic. As per standard practice, two trained therapists, blind to the results of baseline and follow-up assessments, will facilitate each group. Suicidality will be monitored weekly throughout treatment (both conditions) using the 3-item Self-Monitoring Suicide Ideation Scale [29].
To ensure treatment fidelity, each therapist will receive training and weekly supervision. All treatment groups will be audiotaped with 10 % checked to ensure adherence to manualised treatment protocols, using the Mindfulness Based Cognitive Therapy Adherence Scale for MBCT group [30]. At the first treatment session all participants will complete the 6-item Credibility/Expectancy Questionnaire [31] to assess treatment face validity. We will assess treatment acceptability using Dear et al’s [32] 4-item acceptability rating scale. Upon completion of treatment, participants will also provide open comments about acceptability, and specify intended NSSI prevention strategies.
Mindfulness-based cognitive therapy (MBCT)
We will utilise the standard MBCT treatment protocol [21]. Each session combines key elements of cognitive therapy with training in mindfulness meditation. Participants are taught skills designed to foster present moment awareness which include practising mindfulness meditation, body scan, mindful walking and stretching. Cognitive therapy techniques include education about the role of negative thoughts and how rumination, avoidance, suppression, and struggling with unhelpful cognitions and emotions can perpetuate distress rather than resolve it. Participants learn to identify patterns of emotional response and negative thinking that act as warning signals for NSSI and help one another to develop crisis plans and actions to take in the event of future NSSI urges.
Supportive therapy (ST)
Supportive therapy is a widely used active control condition in psychotherapy outcome studies, as it controls for both the non-specific effects of any psychological intervention (i.e. therapeutic relationship) and the unique aspects of group therapy (i.e. social support). Additionally, from an ethical perspective, the group support condition ensures that no participants are left without an intervention. We will use Borkovec and Costello’s [33] manualised protocol. In ST, the therapist provides empathy, fosters a supportive environment, and facilitates discussions among group members around NSSI and other life issues. No MBCT techniques are taught by the therapist.
Ethical issues
After initial telephone screening to assess eligibility, during which participants are verbally informed of the study requirements, all participants will be provided with a detailed information sheet which outlines the aims and participation requirements of the study, informs participants of the confidential and voluntary nature of participation, and outlines how data are to be collected, used and stored in accordance with relevant Privacy Legislation. All participants will be afforded the opportunity to ask questions about the study and will provide signed consent to participate in the research project. Participants are free to withdraw from the research project at any time.
There is some concern regarding the potential for social contagion when discussing NSSI in a group setting. Our experience and a growing evidence base, suggests that when appropriately addressed, iatrogenic effects are rarely observed [34]. Conversely, research indicates that participants benefit from being asked about their NSSI [35, 36]. We will minimise risk of iatrogenic effects by following established guidelines to reduce social contagion [37], including discouraging sharing of explicit details, NSSI images or scars in the group setting. This conduct of this trial has been approved by the Curtin University Human Research Ethics Committee (Ref number: 4884).
Clinical outcome measures
Self-injury monitoring diary:
Participants will complete the Self-Injury Monitoring Diary, developed for this study, across the course of the trial, to assess ongoing frequency and medical severity of NSSI. Each day participants will indicate whether they had an urge to self-injure, the strength of this urge, whether they did self-injure, and the severity of the injury.
Clinician-rated severity of Non-suicidal self-injury
[27]: This single item scale assesses the severity of NSSI on a scale of 0 = None; 1 = Sub-threshold; 2 = Mild; 3 = Moderate; and 4 = Severe. The measure was designed to capture clinically meaningful changes in NSSI severity, based on the proposed DSM-5 criteria [8]. Therapists will rate each participant at the end of each therapeutic session.
Beck depression inventory
- 2nd Edition [38] (BDI-II): The BDI-II is a 21-item measures of depressive symptoms, each rated on a four-point scale. Both a continuous measure indicating severity of symptoms, and clinically meaningful cut-off scores can be obtained. The BDI-II is the gold-standard questionnaire assessment of depression and demonstrates acceptability as a screening tool in both healthy and clinical populations [39, 40].
Beck anxiety inventory
[41] (BAI): The BAI is a 21-item assessment of anxiety symptoms experienced in the last week, with each symptom rated on a four-point scale. Both cognitive and somatic symptoms are assessed. As the gold standard questionnaire assessment of anxiety, the BAI demonstrates excellent psychometric properties [41].
Mechanism of change measures
Cognitive and affective mindfulness scale - revised
[42] (CAMS-R): The CAMS-R is a brief (12 item) self-report measure designed to assess the capacity an individual has to be mindful. The measure demonstrates internal consistence and convergent validity with similar measures when administered to university students [42].
Ruminative thought style questionnaire
[43] (RTSQ): The RTSQ is a 20-item measure describing positive, negative and neutral facets of global rumination (e.g., “I can’t stop thinking about some things” or “I have never been able to distract myself from unwanted thoughts”). Respondents rate each statement on a 7-point Likert scale (1 = not at all descriptive of me, 7 = describes me very well). The RTSQ has demonstrated good convergent validity with the Response Style Questionnaire, the Global Rumination Scale and the Beck Depression Inventory, adequate test-retest reliability and high internal consistency [43].
Difficulty in emotion regulation scale
[44] (DERS): The DERS uses 36 items to tap into 5 aspects of emotion regulation: non-acceptance of emotional response, difficulty in goal directed behaviour, impulse control, emotional awareness, lack of emotion regulation strategies and emotional clarity. Participants respond to each item on a 5-point scale. The measure demonstrates acceptable reliability in university students as well as content and convergent validity [44].
Distress tolerance scale
[45]: This 14 item self-report measure assesses an individual’s ability to withstand feeling distressed. The scale assesses an individual’s ability to tolerate emotions, their appraisal of emotional situations, how absorbed they are by negative emotion and emotion regulation using 5-point Likert scales.
Perceived stress scale
[46]: The Perceived Stress Scale is a 10-item assessment designed to provide a global assessment of perceived stress. Items assess how unpredictable, controllable and overloaded individuals find their lives, without reference to specific events. The scale evidences discriminant validity with depression, and internal consistency [46].
Cortisol measurement:
Both cortisol awaking response (CAR) and daily slope (DS) will be assessed to obtain total daily cortisol output. Cortisol is a biomarker of stress and anxiety [47], and varies in response to stress among people who self-injure [48, 49]. Following best practice [50] we will collect saliva on two consecutive days at each data collection point. On each of these days, participants will collect saliva upon waking, 30 and 45 min after waking (CAR) and again at 4, 9, and 13 h after waking (DS). To increase adherence to the protocol participants who comply with at least 80 % of saliva collection will receive a $50 iTunes voucher. Participants will use Sarstedt Cortisol Salivettes® to collect saliva, which provides an easy and hygienic collection method, further encouraging compliance. Saliva is obtained by chewing on a synthetic swab which is then placed in the container for safe transportation to the laboratory for analysis. Importantly, the Salivettes® are designed to allow reliable analysis from small saliva volumes and low cortisol levels. Following collection, we will recover the saliva from the Salivettes®, centrifuge and store in aliquots at −80 °C until analysis using an enzyme-linked immunosorbent assay (ELISA). Participants will record waking time and all sampling times, and bring their clearly labelled samples to the subsequent assessment session in the clinic.
Attentional bias
The differential allocation of attention to emotional stimuli can be assessed in simple reaction time tasks that are presented using a computer. One of these tasks is the so called ‘dot probe’, which assesses the effect of emotional cues on the detection or identification of a probe stimulus [51]. In this task, two pictures or two words that differ in emotional valence (cues; pleasant and neutral or unpleasant and neutral) are presented simultaneously for a short period of time in different locations on a computer screen, left and right half or upper and lower half. After 500 ms, a probe stimulus is presented in the location that was previously occupied by one of the cues. The probe used in our task are two dots either arranged horizontally ‘..’ or vertically ‘:’. The participant’s task is to indicate the arrangement of the dots by pressing one of two buttons. Participants will be faster to respond to the probe if their focus of attention is at the location at which the probe is presented. If participants are faster to respond to probes that replace unpleasant cues, we conclude that attention is biased towards the unpleasant stimuli indicating preferential processing. If participants are faster to respond to probes that replace neutral cues, we conclude that attention is biased away from unpleasant stimuli indicating avoidance. Pleasant stimuli are included to assess whether the differential allocation of attention is driven by cue valence (differs for pleasant and unpleasant cues relative to neutral cues) or emotional arousal (is similar for pleasant and unpleasant cues relative to neutral cues).
Cues will be pictures (neutral, pleasant, unpleasant, NSSI related) drawn from the standardised International Affective Picture System [52] and normed words (neutral, pleasant, unpleasant, NSSI related) [53]. Cue location will be counterbalanced (e.g. pleasant cues will appear equally often in each possible location) as will be relation of probe to cue (i.e., the probe will follow an emotional cue equally often than a neutral cue - so cues are not predictive of probe position). This results in 24 different stimulus configurations (trials) for a fully counterbalanced design (3 × 2 × 2 × 2; cue valence [pleasant, unpleasant, NSSI related] x cue location [top vs. bottom/left vs right] x probe location [emotional, neutral] x stimulus material [words, pictures]). These 24 stimulus configurations will be repeated 10 times in a random order yielding a total of 240 trials. Including practise trials, this procedure will take approximately 10 min.
Sample size calculation and statistical analysis
We will perform both Intention to Treat and Per Protocol analyses. We will conduct a sensitivity analysis [54], and compare complete cases versus cases lost to follow-up on baseline characteristics and scores on clinical measures, by randomization group [55]. Mixed Model Repeated Measures analyses will compare results on outcome measures across conditions and time points, while controlling for random effects (e.g. age, gender). Using medium effect sizes [56] 40 per group are required for power = .80 and α = .05 (G*Power) [57]. Recruiting an additional 20 % (n = 48 per group) will allow for typical dropout [58, 59]. Mechanisms of action (see Fig. 1), will be tested using multiple mediational analysis with 1000 bias-corrected bootstrap samples in MPlus, to test total and specific indirect effects. This powerful test of mediation requires a sample of 71 participants to detect a medium effect [60]. Participants will be classified into outcome categories (Recovered, Improved, Unchanged, Deteriorated) [61] according to Reliable Change Index and Clinical Significance of change at post-treatment and follow-up, to determine whether MBCT and ST differ in clinically significant reductions in NSSI and associated symptoms of anxiety and depression over time.