Effect of 3 Forms of Early Intervention for Young People With Borderline Personality Disorder: The MOBY Randomized Clinical Trial

Key Points

Question  What combination of treatment components is sufficient for early intervention for young people with borderline personality disorder (BPD)?

Findings  In this randomized clinical trial with 139 youth with BPD, a dedicated BPD service model and a specialized BPD psychotherapy were associated with superior retention in care but not a superior rate of change in psychosocial functioning by 12 months, compared with general youth mental health care and a psychotherapy control condition.

Meaning  Effective early intervention for BPD is not reliant on availability of BPD psychotherapy.

Importance  Clinical trials have neither focused on early intervention for psychosocial impairment nor on the contribution of components of borderline personality disorder (BPD) treatment beyond individual psychotherapy.

Objective  To evaluate the effectiveness of 3 early interventions for BPD of differing complexity.

Design, Settings, and Participants  This single-blinded randomized clinical trial recruited young people between March 17, 2011, and September 30, 2015, into parallel groups. The study took place at 2 government-funded mental health services for young people in Melbourne, Australia. Inclusion criteria were age 15 to 25 years (inclusive), recent DSM-IV-TR BPD diagnosis, and never receiving evidence-based BPD treatment. A total of 139 participants were randomized (pool of 876; 70 declined, 667 excluded), balanced for sex, age, and depressive symptomatology. Data analysis completed May 2020.

Interventions  (1) The Helping Young People Early (HYPE) dedicated BPD service model for young people, combined with weekly cognitive analytic therapy (CAT); (2) HYPE combined with a weekly befriending psychotherapy control condition; and (3) a general youth mental health service (YMHS) model, combined with befriending. Therefore, the 3 treatment arms were HYPE + CAT, HYPE + befriending, and YMHS + befriending. Participants were randomly assigned both to 1 treatment arm (in a 1:1:1 ratio) and to a clinician.

Main Outcomes and Measures  Psychosocial functioning, measured with the Inventory of Interpersonal Problems Circumplex Version and the Social Adjustment Scale Self-report.

Results  One hundred twenty-eight participants (104 [81.3%] were female; mean [SD] age, 19.1 [2.8] years; HYPE + CAT: 40 [31.3%]; HYPE + befriending: 45 [35.2%]; YMHS + befriending: 43 [33.6%]) who provided postbaseline data were included in the intent-to-treat analysis. Regardless of group, from baseline to 12 months, there was a mean of 19.3% to 23.8% improvement in the primary outcomes and 40.7% to 52.7% for all secondary outcomes, except severity of substance use and client satisfaction. The latter remained high across all time points. Planned comparisons (YMHS + befriending vs HYPE; HYPE + CAT vs befriending) showed that neither the service model nor the psychotherapy intervention was associated with a superior rate of change in psychosocial functioning by the 12-month primary end point. The HYPE service model was superior to YMHS + befriending for treatment attendance (median [IQR], 22 [19] vs 3 [16] contacts; median duration, 200 [139.5] vs 94 [125] days) and treatment completion (44 of 92 [47.8%] vs 9 of 47 [19.2%]). HYPE + CAT was superior to befriending for treatment attendance (median [IQR], 12 [16.5] vs 3 [9.8] sessions) and treatment completion (24 of 46 [52.2%] vs 29 of 93 [31.2%]).

Conclusions and Relevance  In this randomized clinical trial of 3 interventions for young people with BPD, effective early intervention was not reliant on availability of specialist psychotherapy but did require youth-oriented clinical case management and psychiatric care. A dedicated early intervention BPD service model (HYPE), with or without individual psychotherapy, achieved greater treatment attendance and completion, making it more likely to meet service user, family, and community expectations of care.

Trial Registration  anzctr.org.au Identifier: ACTRN12610000100099

Borderline personality disorder (BPD) is associated with harmful long-term outcomes that include premature mortality^1,2 and persistent and severe impairments in psychosocial functioning,^3,4 which are largely responsible for the high burden of disease.¹ BPD has its clinical onset between puberty and emerging adulthood, but the opportunity to reduce or avoid such poor outcomes is routinely missed through delay in diagnosis and treatment.¹

The past decade has seen significant progress, with 8 published randomized clinical trials comparing structured psychological interventions with active comparators in young people with BPD, to our knowledge.^5-12 In most trials, specialized, structured interventions have outperformed comparators with regard to the rate^5,7-9,11 and extent^7-9,11 of improvement on the primary outcome. Differences have been clinically modest and have not been sustained at 12-month¹¹ or 3-year follow-up.¹³ Furthermore, methodological issues to be addressed include improving on treatment as usual as a comparator^6-9,12 and consistently reporting treatment fidelity.^6-10,12

There is international consensus that further progress requires clarity regarding the target population, developing more widely accessible interventions, and defining meaningful outcomes.¹ While 3 trials focused on early-stage disorder^5,6,12 and 1 enrolled only young people with the full BPD syndrome,¹⁰ no trial has targeted early intervention with recently diagnosed disorder^14-16 across the peak period of onset for the major mental disorders.¹⁷ All trials have provided individual and/or group psychotherapies, usually with high training needs and limited capacity for scaling up throughout health care systems.¹⁸ Finally, no trial has stipulated psychosocial functioning as its primary outcome and most have excluded participants with the most severe clinical presentations or with typical, real-world, co-occurring problems, such as substance dependence,^6-8,11,12 limiting generalizability of findings. A trial is warranted to investigate the role of more widely available components of treatment (eg, service model, clinical case management, psychiatric care) that often accompany psychotherapy programs among adolescents and young adults with BPD.

Quiz Ref IDThe Monitoring Outcomes of Borderline Personality Disorder in Youth (MOBY) randomized clinical trial aimed to clarify the degree of complexity of treatment required for effective early intervention for BPD by examining 3 combinations of individual psychotherapy and clinical service model: (1) the Helping Young People Early (HYPE) dedicated BPD service model for young people combined with weekly cognitive analytic therapy (CAT); (2) HYPE combined with a weekly befriending psychotherapy-control condition; and (3) a general youth mental health service (YMHS) model combined with befriending. Therefore, the 3 treatment arms were HYPE + CAT, HYPE + befriending, and YMHS + befriending. Hypotheses were that at the 12-month primary end point, (1) participants receiving individual psychotherapy (HYPE + CAT) would have a faster rate of change on the primary and secondary outcomes than those receiving befriending (HYPE + befriending and YMHS + befriending) and (2) participants receiving the HYPE service model (HYPE + CAT and HYPE + befriending) would have faster rate of change on the primary and secondary outcomes than those receiving the generalist mental health service model for young people (YMHS + befriending).

This single-blind, parallel groups randomized clinical trial was approved by the Melbourne Health Human Research Ethics Committee. The trial protocol is published¹⁹ and available in Supplement 1.

Minimal exclusions ensured the trial was open to most young people with BPD referred to either the Orygen or Headspace government-funded youth mental health services, located in western and northwestern metropolitan Melbourne, Australia. Key inclusion criteria were (1) age 15 to 25 years (inclusive) and (2) Structured Clinical Interview for DSM-IV-TR Axis II Disorders²⁰ diagnosis of BPD (identical to DSM-5 criteria). Key exclusion criteria were (1) Structured Clinical Interview Axis I Disorders Patient Edition²¹ diagnosis of psychotic disorder within the past 12 months; (2) lifetime diagnosis of a schizophrenia spectrum or bipolar I or II disorder; and (3) prior evidence-based treatment for BPD. All participants (and a parent/legal guardian for minors) gave written informed consent, with participant enrollments occurring between March 17, 2011, and September 30, 2015.

Using computer-generated randomization sequences, prepared by an independent statistician, participants were randomly assigned both to 1 treatment arm (in a 1:1:1 ratio) and to a clinician. Permuted blocking was used, and participants were stratified by age (cut point of age 18 years), sex assigned at birth, and Center for Epidemiological Studies Depression Scale–Revised²² score (cut point of 37). Participants were informed of their treatment assignment by the trial coordinator. Assessors were blind to treatment assignment and to the study design. Testing of assessors at the end of the data collection phase indicated that blinding was maintained throughout the trial.

The 3 interventions, HYPE + CAT, HYPE + befriending, and YMHS + befriending, are described in detail elsewhere.¹⁹ The trial intervention ended once 16 CAT or befriending sessions were offered or attended or when a participant did not attend any CAT or befriending session for 6 consecutive weeks. Treatment completion was defined as attending 8 or more CAT or befriending sessions.

Orygen’s HYPE model is detailed elsewhere.^23,24 Key elements include a CAT-based relational model (relational clinical care), integrated clinical case management and general and acute psychiatric care, an explicitly collaborative approach, assertive engagement, outreach care in the community when needed, active inclusion of families, and a focus on functional recovery.

The YMHS model was implemented at Headspace by mental health clinicians who had expertise in young people (but who were not BPD treatment specialists), working within a multidisciplinary group practice setting. Key elements included a focus on engagement and young people–friendly practice, diagnosis, and treatment of mental disorders, clinical case management when needed, medical or psychiatric consultation when clinically indicated, and access to acute crisis care.

The YMHS intervention was created for the purpose of this randomized clinical trial and required a specific agreement with Headspace to accept acutely unwell clients with BPD and to refrain from any training in CAT. Routine Headspace care normally comprises time-limited, government-funded, fee-for-service individual psychotherapy.²⁵ For some, it includes a psychosocial intake assessment by an intake worker. Unlike YMHS, it does not include clinical case management or routine access to psychiatric consultation and/or acute crisis care.

CAT is a time-limited psychotherapy that focuses on understanding problematic self-management and interpersonal relationship patterns and the thoughts, emotions, and behavioral responses that result from these patterns.^26-28 CAT is collaborative, integrative, practical, and transdiagnostic.

Quiz Ref IDBefriending has been used in trials to match for common factors of psychotherapy, including time in therapy, participant expectations, and the client-therapist relationship.²⁹ Befriending consists of talking about neutral topics of interest to the participant, such as sport, music, social, and vocational activities, while avoiding or redirecting the participant from emotionally loaded topics, such as symptoms or interpersonal problems.

Nine clinicians (7 clinical psychologists, 1 social worker, 1 occupational therapist) with a mean (SD) of 8.22 (5.13) years of clinical experience delivered HYPE + CAT and 28 trainee psychologists (25 clinical, 3 educational and developmental) with a mean (SD) of 0.37 (0.91) years of experience delivered befriending.

Management of treatment integrity is detailed elsewhere.¹⁹ Briefly, in addition to manuals, guidelines, and clinician supervision, YMHS was structurally separated from HYPE by location and different clinicians. Randomly selected audio recordings were rated with the CAT competence³⁰ and ACE Treatment Integrity Measure.³¹ CAT adherence and competency were excellent. The mean (SD) CAT competence domains was 3.4 (0.6) (subscale range, 0-4, with 4 indicating highest competence). The mean (SD) CAT competence items was 1.6 (0.4) (subscale range, −2 to 2, with −2 indicating missed opportunity and/or not good enough CAT, 0 indicating absence of CAT, 2 indicating CAT was present and well demonstrated). Adherence to befriending was very good, with the mean (SD) ACE Treatment Integrity Measure befriending subscale of 4.1 (1.1) (range, 0-6, with 6 indicating the highest adherence). Separation of CAT individual psychotherapy tools and techniques from befriending was excellent (mean [SD] CAT competence items, 0.04 [0.2]) and from HYPE and YMHS was very good (mean [SD] CAT competence items, 0.5 [0.5]).

Graduate research assessors, trained and supervised by a psychiatrist (A.M.C.) and psychologist (J.K.B.), conducted assessments at baseline (before randomization) and 3, 6, 12 (primary end point), and 18 months (secondary end point). Interrater reliability was excellent (eg, depression³² ranged from intraclass correlation coefficient [ICC] [2,1] = 0.94; P < .001 to ICC [2,1] = 0.98; P < .001; BPD features³³ ranged from ICC [2,1] = 0.98; P < .001 to ICC [2,1] = 1.0; P < .001).

The primary outcome was psychosocial functioning, jointly measured by the Inventory of Interpersonal Problems Circumplex Version^34,35 and the Social Adjustment Scale Self-report.³⁶ Secondary outcomes included DSM-IV-TR BPD features,³³ suicidal ideation,³⁷ suicide attempts and nonsuicidal self-injury,³⁸ depression,^22,32 substance use,^39,40 and client treatment satisfaction.⁴¹ Subsidiary measures included mental state and personality disorder diagnoses,^20,21 social and occupational functioning,⁴² quality of life,⁴³ and demographic and treatment information.

Research retention rates and treatment attendance were examined with independent-samples t tests, independent-samples nonparametric Kruskal-Wallis tests, and χ², with α set at .01. A statistician, blinded to treatment allocation, conducted an intent-to-treat analysis using participants with 1 or more data point postbaseline and a per-protocol analysis with participants who completed treatment. A series of mixed-effects model repeated-measures analysis of variance models examined group differences in the rate of change, with α set at .025 for the primary outcomes and .01 for the secondary outcomes. The models contained group (HYPE + CAT, HYPE + befriending, and YMHS + befriending), time (0, 3, 6, 12, and 18 months), and the interaction of group and time as fixed factors, as recommended.⁴⁴ Covariates were the strata in the randomization and the covariance matrix was modeled as unstructured. Because testing whether there were differences between the 3 groups over time (eg, interaction group × time) would not address the study’s hypotheses, we conducted a series of planned comparisons within the mixed-effects model repeated-measures analysis of variance models. The first involved comparing the rate of change between baseline and each follow-up point between HYPE + CAT vs the combined befriending (ie, HYPE + befriending and YMHS + befriending) conditions. The second tested whether the combined HYPE groups (ie, HYPE + CAT and HYPE + befriending), compared with YMHS + befriending, had a superior rate of change between baseline and each point. Means and standard errors related to the change from baseline to the follow-up time points were calculated. Suicide attempts and nonsuicidal self-injury were examined with χ² with participants who had completed the majority of follow-up assessments (ie, ≥2 of 3 by 12 months or ≥3 of 4 by 18 months). The power and sample size calculations for the study were based on a 1-way analysis of covariance model, with baseline values as the covariate, and the outcome measures as the dependent variables. It was assumed that the baseline covariate would account for 50% of the variance in outcome measure. Based on published data,^5,45 we expected a medium treatment effect on the primary outcome measures, and α was set at .05 and power at .80. Given these parameters, 40 participants per group would be required (120 in total) with at least 2 data points. Allowing for a 14% attrition rate, as per our previous randomized clinical trials,⁵ at least 45 cases per group would be required.¹⁹ Data analysis was completed in May 2020.

A total of 139 young people (112 female individuals [80.6%]; mean [SD] age, 19.1 [2.8] years; 123 [88.5%] were born in Australia) were randomized.⁴⁶ Of these, 128 participants (92.1%) were included in the intent-to-treat analysis and 53 (38.1%) in the per-protocol (Figure 1 and Table 1). Retention rates did not differ significantly between the 3 groups at any time point.

Participants assigned to HYPE + CAT attended significantly more individual psychotherapy sessions than those assigned to befriending (Kruskal-Wallis χ² = 13.6, P < .001) (Table 2) and completed treatment more than those assigned to befriending (χ² = 8.3; P = .004). Compared with those assigned to YMHS + befriending, participants assigned to HYPE had significantly more treatment contacts (Kruskal-Wallis χ² = 21.1; P < .001), spent more days in treatment (Kruskal-Wallis χ² = 24.0; P < .001), and completed treatment more often (χ² = 11.2; P = .001). Two participants died by suicide in the follow-up period and 1 died by suicide while receiving the study intervention (Figure 1).

Quiz Ref IDWith regard to the rate of change between baseline and 12 months, HYPE + CAT was not superior to befriending for either interpersonal problems (Inventory of Interpersonal Problems Circumplex Version, t_99.73 = −0.87; P = .39) or social adjustment (Social Adjustment Scale Self-report, t_110.36 = −0.753; P = .45) (Table 3 and Figure 2). HYPE was not superior to YMHS + befriending for interpersonal problems (Inventory of Interpersonal Problems Circumplex Version, t_96.12 = 0.80; P = .43) or social adjustment, (Social Adjustment Scale Self-report, t_106.68 = 0.06; P = .95). Notably, all groups improved significantly over time on these variables (eTable in Supplement 2). From baseline to 12 months, there was a mean 28.91-point (23.8%) drop in Inventory of Interpersonal Problems Circumplex Version scores and a mean 0.55-point (19.3%) drop in Social Adjustment Scale Self-report scores.

With regard to the rate of change between baseline and 12 months, befriending was superior to HYPE + CAT for suicidal ideation (Beck Scale for Suicide Ideation, t_90.96 = −2.75; P = .007), and HYPE was not superior to YMHS + befriending. YMHS + befriending was superior to HYPE with regard to the rate of change in suicidal ideation from baseline to 6 months (Beck Scale for Suicide Ideation, t_91.64 = 8.82; P = .006).

Regardless of group, from baseline to 12 months, mean change ranged from 40.7% (17.64 points) for the Center for Epidemiological Studies Depression Scale–Revised to 52.7% (6.22 points) on the Beck Scale for Suicide Ideation for all secondary outcomes (Table 3 and eTable in Supplement 2), except severity of substance use and client satisfaction. The latter remained high across all time points.

This yielded the same pattern of findings, except that the differences in rate of change in suicidal ideation between HYPE + CAT compared with befriending and HYPE compared with YMHS + befriending no longer remained significant.

Quiz Ref IDThe marked and sustained improvement in the primary outcomes across all 3 treatments suggests that early intervention for BPD is effective. This is likely to represent a true treatment effect, as recent data from 701 young people with BPD, or BPD features, across 76 Australian Headspace centers showed only a 3% to 6% improvement in mean levels of quality of life, distress, and social and occupational functioning scores with treatment (D. Rickwood, PhD, email, July 22, 2021) and that 69%, 60%, and 45% of participants either did not improve or deteriorated on each outcome, respectively.⁴⁷ Moreover, studies of the natural history of the primary outcome variables in young people with BPD suggest continuing, stable, and severe psychosocial impairment into adulthood.^48,49

For the secondary outcomes, the relatively rapid rate of change on the symptom severity measures, consistency of findings, and abovementioned Headspace data support true treatment effects. Nonetheless, treatment effects are more difficult to reliably discern for some outcomes, as the natural history of self-harm⁵⁰ and BPD features⁵¹ in young people is toward improvement and regression to the average cannot be ruled out. Notably, substance use did not change over time, while satisfaction with treatment remained good throughout.⁵²

Taken together, the intent-to-treat and per-protocol analyses revealed that both the HYPE model (compared with YMHS + befriending) and individual therapy (HYPE + CAT compared with befriending) achieved greater treatment attendance and treatment completion but appeared to confer no additional benefit in terms of the rate of psychosocial improvement, questioning the purpose of such treatment elements. YMHS + befriending does not represent routine Headspace care and was designed specifically for this trial. Despite this, 4 YMHS + befriending participants were discontinued from the trial intervention and referred to Orygen/HYPE because their clinical risks and/or needs exceeded Headspace’s capacity. Data from 76 Australian Headspace centers show that, despite a clinical prevalence of 11% to 22%, only 701 of 74 804 young people (0.9%) received a primary diagnosis of BPD or borderline traits.⁴⁷ The mean (SD) number of sessions attended was 3.4 (2.6), similar to YMHS + befriending. The most common intervention reported was cognitive behavior therapy, delivered to only 28% of clients by session 3. These data suggest that usual Headspace practice is insufficient for this patient group. Like YMHS, the HYPE model fosters a service culture that is welcoming, nonstigmatizing, and nonharmful for young people with BPD. Unlike YMHS, HYPE is also able to retain young people with BPD in care, fulfilling the expectations of young people, family, and the community that mental health services meet the needs of this patient group,⁵³ regardless of the young person’s inclination to enter into individual psychotherapy. Uncoupling individual psychotherapy from the HYPE service model is likely to allow for more cost-effective generalization of HYPE to Headspace and similar platforms internationally, taking early intervention for BPD out of the wasteland of niche services for few young people.¹⁸ Improving access to services is a specific need expressed by service users⁵³ that will require structural changes, including funding to support clinical case management, irregular attendance, community outreach, and crisis care. Such skills are already commonplace and can be more rapidly and widely implemented than training in specialized BPD psychotherapies.

The MOBY trial suggests that individual psychotherapy is not essential for successful BPD early intervention, which concurs with evidence that structured, high-quality, generalist treatments perform similarly to specific BPD psychotherapies in young people⁵ and in adults.^54,55 However, unlike the MOBY trial, the comparator in previous trials contained psychotherapeutic elements (problem solving^5,55 or psychodynamic therapy⁵⁴), whereas befriending contains none of these elements and is designed to control only for time in treatment, participant expectations, and the client-therapist relationship. We cannot rule out that these befriending components contributed to patient outcomes. Nonetheless, it does not follow that a model of BPD is unnecessary or that anything goes, as routine Headspace outcome data suggest that usual care is inadequate.⁴⁷

The precise role of individual psychotherapy in acute-phase early intervention for BPD remains unclear. CAT might be uniquely ineffective among the available psychotherapies for BPD. This seems unlikely, given CAT’s effectiveness in other clinical trials in young people⁵ and adults⁵⁶ and evidence that a wide variety of other psychotherapies are similarly and modestly effective.^57-59 Alternatively, the timing of when to introduce psychotherapy might need to be reconsidered. In practice, clinicians usually associate a diagnosis of BPD with referral to an individual psychotherapy program. Yet, access to, and the uptake of, psychotherapy occurs in only a small minority and treatment completion is low.¹⁸ The MOBY trial suggests that an overriding focus on psychotherapy, instead of more fundamental aspects of clinical care, might be misplaced in early-stage BPD. Individual psychotherapy might be more suited to nonacute or milder BPD, to later stages of the disorder (when the recurrent nature of problems is more evident), or to people with BPD who have more developed (or less impaired) self-regulatory capacities.

The MOBY trial’s strengths include being the largest BPD trial cohort of young people to date and to our knowledge, the first to focus exclusively on early intervention for people with recently diagnosed BPD who had never received evidence-based BPD treatment,^14,16 and the first to span the age range of young people. It intentionally included acute, severe, and complex young people, typical of frontline services.⁴⁶ The design enabled both an examination of the effectiveness of service model over and above the background individual psychotherapy/control condition and the effectiveness of individual psychotherapy over and above the service model. The MOBY trial had a high consent rate, demonstrated adherence and competency of interventions, diversity of clinicians delivering each treatment, and good researcher interrater reliability.

Quiz Ref IDLimitations include uneven randomization for antisocial personality disorder, with nearly double the proportion in the HYPE + CAT arm, which might have reduced the effectiveness of CAT. Not being a noninferiority design, it is impossible to say whether all 3 treatments were equally effective. The geographical separation of the HYPE and YMHS sites might have contributed to some young people being unable to access care. The posttreatment phase for each group was uncontrolled and some participants received further treatment, including HYPE + CAT.

In this randomized clinical trial of 3 early interventions for young people with BPD, regardless of group, from baseline to 12 months, there was a mean improvement of 19.3% to 23.8% in the primary outcomes of psychosocial functioning. The MOBY trial provides welcome news for early intervention for BPD and has the potential to improve access to care. Early intervention is likely to be effective if it includes a service culture that is nonstigmatizing toward BPD, youth-friendly, and oriented to early detection and treatment for BPD. Essential elements of care appear to include a model for understanding BPD, clinical case management, and treatment of co-occurring functional, physical, and psychopathological difficulties. How these elements might work remains an open question for further investigation. Early intervention for BPD does not appear to be reliant on availability of specialist psychotherapy and the role and timing of such treatments needs further study. With appropriate funding, services could be scaled up immediately, via existing national mental health platforms for young people or similar services globally.⁶⁰

Corresponding Author: Andrew M. Chanen, MBBS(Hons), PhD, 35 Poplar Rd, Locked Bag 10, Parkville, VIC 3052, Australia (andrew.chanen@orygen.org.au).

Accepted for Publication: October 22, 2021.

Published Online: December 15, 2021. doi:10.1001/jamapsychiatry.2021.3637

Author Contributions: Profs Chanen and Cotton had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Chanen, Jackson, Cotton, Gleeson, Davey, McCutcheon.

Acquisition, analysis, or interpretation of data: Chanen, Betts, Cotton, Davey, Thompson, Perera, Rayner, Andrewes.

Drafting of the manuscript: Chanen, Betts, Jackson, Cotton, Gleeson, Davey, Andrewes.

Critical revision of the manuscript for important intellectual content: Jackson, Cotton, Gleeson, Davey, Thompson, Perera, Rayner, McCutcheon.

Statistical analysis: Betts, Cotton, Andrewes.

Obtained funding: Chanen, Jackson, Cotton, Gleeson, Davey, McCutcheon.

Administrative, technical, or material support: Chanen, Betts, Cotton, Gleeson, Davey, Perera, Rayner, McCutcheon.

Supervision: Chanen, Betts, Jackson, Gleeson, Davey, McCutcheon.

Conflict of Interest Disclosures: Prof Chanen reported grants from the Australian government’s National Health and Medical Research Council (NHMRC) during the conduct of the study and other support from the Helping Young People Early (HYPE) translational program outside the submitted work. Profs Chanen and McCutcheon cofounded and lead the HYPE clinical program, a government-funded program with continuous funding, and the HYPE translational program, which is a not-for-profit training program. No other disclosures were reported.

Funding/Support: This trial was funded by the National Health and Medical Research Council (NHMRC) (grant GNT0628739). NHMRC Career Development Fellowship supported Prof Cotton (APP1061998) and Prof Davey (APP1061757).

Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Data Sharing Statement: See Supplement 3.

Additional Contributions: The authors wish to thank the clients, families, and staff of the Helping Young People Early program at Orygen and Headspace Sunshine, and also Sarah Bendall, PhD, PGDipClinPsych, MA, Orygen and the Centre for Youth Mental Health, University of Melbourne, for training and supervising clinicians performing the befriending intervention and Patrick McGorry, MD, PhD, Orygen and the Centre for Youth Mental Health, University of Melbourne, for feedback on an earlier version of this manuscript. Drs Bendall and McGorry did not receive compensation for their contribution. This study is dedicated to the 3 young participants who died during the course of this study and to their families and friends.