Comparing three forms of early intervention for youth with borderline personality disorder (the MOBY study): study protocol for a randomised controlled trial

This study is a parallel group, single-blind RCT comparing three groups: HYPE + CAT, HYPE + BEF, and YMHS + BEF. A full factorial design ( that is, including a fourth treatment arm comprising YMHS + CAT) was not implemented in order to meaningfully differentiate HYPE from YMHS. The HYPE service model incorporates the theory and principles of CAT, without CAT necessarily being used as an individual therapy. This is a key point of differentiation from usual youth mental health service models. If CAT were introduced at the YMHS site it would change usual YMHS practice (that is, the YMHS would represent a specialised rather than generalised service), thus limiting the external validity of this intervention.

The study design was developed in accordance with Good Clinical Practice (GCP) Guidelines and Standard Protocol Items; Recommendations for Interventional Trials (SPIRIT [30]). Figure 1 summarises the trial design. The trial is registered with the Australian New Zealand Clinical Trial Registry (ACTRN12610000100099) and has been approved by the Melbourne Health Human Research Ethics Committee (HREC2010.055).

The HYPE + CAT and HYPE + BEF treatments are being delivered at the HYPE Clinic at Orygen Youth Health [24]. Orygen Youth Health is the state-funded specialist mental health service for 15- to 25-year-olds living in western metropolitan Melbourne, Australia. YMHS + BEF is being conducted at headspace in western metropolitan Melbourne [31], which is part of a national network of federally funded youth (12-to 25-year-olds) mental health clinics that is managed independently of Orygen Youth Health. It has a similar multidisciplinary outpatient staffing profile and services a similar catchment area to Orygen Youth Health. Therefore, the clientele of the two services are socio-demographically similar. Should the need arise, headspace patients have access to Orygen Youth Health’s after-hours crisis and inpatient services.

All participants provide written informed consent. Participants comprise help-seeking young people who either self-refer or are referred to the Orygen Youth Health or headspace intake systems. Broad inclusion criteria are used to maximise the external validity of the trial:

Exclusion criteria are:

Patients at Orygen Youth Health are routinely screened using the 15 self-report borderline personality disorder items from the SCID-II Personality Questionnaire (SCID-II PQ) [32]. Patients at headspace are optionally screened using the SCID-II PQ. Published data from a previous study at Orygen Youth Health [34] indicate that a cut score of >12 has adequate sensitivity and good specificity and negative predictive value for BPD in youth. All patients scoring >12 are considered for eligibility, along with other trial inclusion criteria. In addition, patients who might screen negative for BPD (SCID-II PQ score ≤12) or who eluded screening but are identified by clinicians as meeting criteria for BPD using the SCID-II borderline personality disorder module [32] are also considered for eligibility, along with other trial inclusion criteria.

Participants are invited to participate by the trial coordinator, who contacts the young person (and their parent or legal guardian for those under 18 years) to introduce the study and to offer an appointment to obtain written consent. After the trial coordinator obtains written consent, a research assistant or the trial coordinator conducts the baseline assessment. All participants receive assertive case management, crisis care and general psychiatric care during the pre-screening and assessment phase. Once eligibility for the trial is confirmed, the trial coordinator randomly and consecutively assigns participants to one of the three interventions in a ratio of 1:1:1 using a password-protected computer program with a randomisation sequence that was computer-generated by an independent statistician. Treatment allocation uses randomised permuted blocking, and participants are stratified by age (cut point of age 18 years), sex and the Centre for Epidemiological Studies Depression Scale Revised (CESD-R) [35] score (cut point = 37, which is the mean score for borderline personality disorder patients on entry to HYPE), thus minimising these potentially confounding factors. Following randomisation to a treatment group, the trial coordinator then randomly assigns participants to the next available trial clinician using a separate randomisation sequence that was also computer-generated by an independent statistician. The trial coordinator contacts the participant and the treating service to inform them of the treatment group and clinician allocation. The clinician then contacts the participant to arrange the first appointment. Thus, the trial coordinator conducts the consent and randomisation procedures separately from the blinded research assistants, using independently derived, password-protected, computer-generated randomisation sequences.

The research assistants conduct all follow-up assessments. The study design and the treatment group assignment are concealed from the research assistants conducting the assessments. Blinding of the research assistants is achieved and maintained by having a specially written version of the trial protocol that removes all references to the three-group design and to any form of treatment. Most research assessments are conducted in the community or at an Orygen Youth Health assessment clinic where no community treatment takes place. Research assistant contact with clinical staff is minimised, and all trial participants, trial clinicians and other staff are regularly reminded of the importance of maintaining the blinding of the research assistants. The research assistants are aware that they are blind to elements of the study. Blinding of the research assistants will be tested at the end of their employment or the end of the study using an online self-report instrument designed specifically for the study. The study’s biostatistician is also blinded with respect to the randomisation.

The HYPE + CAT group receives up to 16 × 50 minute CAT sessions. Both the HYPE + BEF and YMH + BEF groups receive up to 16 × 50 minute sessions of befriending. Clinicians aim to deliver the 16 CAT or befriending sessions weekly, but the disorganised behaviour evident in many in this patient population means that it usually takes approximately 26 weeks to deliver them. Completion of 16 sessions marks the completion of all trial treatment. At the end of treatment or at any stage thereafter, participants can be referred for or seek additional treatment.

All participants receive case management and general psychiatric care as needed, and crisis and inpatient services accessed from the state mental health system, where indicated.

All treatment sessions are audio-recorded to enable treatment integrity checks. CAT and befriending are both manualised treatments. Integrity (adherence, competency and separation) will be managed via regular expert supervision for CAT and befriending. Each CAT therapist will have two randomly selected CAT session recordings rated for adherence and competency, using a scale designed specifically for CAT [38]. A random selection of 50 % of befriending clinicians will have two randomly selected sessions rated for adherence and competency, using a scale designed specifically for befriending [37], and for treatment separation using the CAT scale. The befriending scale measures whether the therapist: (i) redirects from unresolved conflicts to a neutral topic, (ii) redirects from discussion about symptoms to a neutral topic, (iii) chooses the most neutral line of questioning, (iv) reacts minimally to loaded speech (for example, symptoms, conflicts), (v) redirects from identity issues to a neutral topic and (vi) predominantly engages the client in neutral conversations on day-to-day subjects.

The HYPE service model is manualised, but YMHS is not. Integrity of HYPE and YMHS will be maintained through the structure of the services and clinician supervision. HYPE also uses checklists of the tasks of care. Measures have been implemented to reduce ‘contamination’ of treatments and thereby to ensure treatment separation. HYPE and YMHS are governed and administered by separate organisations. YMHS clinicians have never worked in HYPE and do not have regular clinical contact with HYPE staff. Clinicians conducting the befriending have no training in CAT and are excluded from conducting further befriending if they commence CAT training at any time during the trial. CAT practitioners are permitted to have exposure to other clinical intervention techniques.

A random selection of 50 % of case managers will have two randomly selected sessions rated for treatment separation (to ensure that CAT is not being done covertly), using the above-mentioned CAT scale [38].

The primary endpoint of the trial is 12 months after baseline assessment. Assessments occur at 3, 6, 12 and 18 months after baseline. Table 1 lists the primary and secondary outcomes, along with subsidiary measures. Multiple methods are being employed to assess outcomes. These include independent observer rating, self-report, and Mobiletype© [39], a novel method of in vivo ecological momentary assessment.

The primary outcome is adaptive functioning, defined by measures of interpersonal problems [40] and social adjustment [41]. This was chosen in light of the evidence that impairments in adaptive functioning among individuals with borderline personality disorder tend to be persistent and refractory to current treatments [14]. The Inventory of Interpersonal Problems Circumplex Scales (IIP-C) [40] has 64 self-report items, rated using a five-point scale ranging from 0 (not at all) to 4 (extremely). It contains eight scales (Domineering, Vindictive, Cold, Socially Inhibited, Overly Accommodating, Non-assertive, Self-Sacrificing, Intrusive). The Social Adjustment Scale Self-Report (SAS-SR) [41] has 54 self-report items, scored on a five-point scale, measuring instrumental and expressive role performance over the previous 2 weeks. It yields scores in six domains (Primary Relationship, Work, Social and Leisure, Extended Family, Parental (own children) and Family Unit (partner or children)) and a total score, calculated by averaging all applicable items. The SAS-SR contains ‘skip-outs’, so that non-applicable items are omitted, making it appropriate for the age range of participants in the proposed study.

Secondary outcomes measure client satisfaction [42], borderline personality disorder features [43], suicidal ideation and deliberate self-harm [39, 44, 45], affective instability [39, 46], depression [35, 47] and substance use [48, 49]. Subsidiary outcome measures include emotion regulation [50], quality of life [51], a summary measure of social and occupational functioning [52], therapeutic alliance [53] and demographic, treatment and diagnostic information [32, 54].

Clinical governance for trial participants rests with the service to which they are allocated (Orygen Youth Health or headspace). Treatment continues for approximately 26 weeks, unless the participant meets the trial’s withdrawal criteria. Withdrawal from the trial’s treatment program will occur if:

The research assistants who conduct follow-up assessments are psychology graduates who are trained and supervised by the chief investigator/consultant psychiatrist and the trial coordinator/clinical psychologist. Assessment interviews are audio recorded, and ten percent of assessments at each time point are randomly selected for inter-rater reliability.

To determine whether there are group differences on the primary and secondary outcome measures at 12 months, a series of mixed effects model repeated measures analysis of variances models (MMRM) will be employed. The within groups factor will be time, and the group will serve as the between subjects factor. MMRM is a preferred method for analysing clinical trial data in psychiatry in comparison with the more traditional repeated measures analysis of variance and analysis of covariance models [55]. MMRM is advantageous because it includes all existing data in the model, there is no imputation or substitution of missing data, dispersion and correlation can occur at all time points, and sensitivity analysis can be successfully conducted. Within these models, planned comparisons will be used to determine whether HYPE + CAT is superior to the other two interventions, and whether the HYPE service model leads to better outcomes compared with YMHS.

Power and sample size calculations have been based on a one-way analysis of covariance model with baseline values as the covariate, and the outcome measures as the dependent variables. It is assumed that the baseline covariate will account for 50 % of the variance in an outcome measure. Based on our published data [29, 56], we expect to find medium treatment effects on the primary outcome measures of interpersonal problems and social adjustment. Alpha has been set at 0.05 and power (1-β) for the study at 0.80. Given these parameters, we require 40 participants per group (120 in total) with at least two data points. Based upon the 14 % attrition rate in our previous randomised control trial we intend to recruit at least 45 cases per group.