Short-Term versus Long-Term Mentalization-Based Therapy for Borderline Personality Disorder: A Randomized Clinical Trial (MBT-RCT)

Borderline personality disorder (BPD) is a mental disorder characterized by a pervasive pattern of symptoms, such as affective dysregulation, identity diffusion, interpersonal problems, self-harming behavior, and suicide-related mortality [1]. According to epidemiological studies, 1.6% of the general population suffers from BPD [2]. In clinical populations, it is the most common personality disorder [2], with a prevalence of between 9 and 22% of all psychiatric outpatients [3]. BPD causes a significant socioeconomic burden, not only because of direct health care costs but also because of indirect costs like social and occupational dysfunction [4]. These findings emphasize the need for efficacious and cost-effective treatments for this highly prevalent disorder. Currently, psychological interventions are considered the primary treatment of choice for BPD [5].

Traditionally, long-term comprehensive psychotherapy programs have been considered necessary for patients with BPD [6]. These include, but are not limited to, long-term psychodynamic therapy (for complex mental health disorders including BPD) lasting minimum a year [7], dialectical behavior therapy (DBT) up to 1 year, transference-focused psychotherapy (TFP) up to 3 years, schema-focused therapy (SFT) up to 3 years, and mentalization-based therapy (MBT), which was originally manualized and tested for 18 months but is sometimes delivered in longer durations in clinical practice, e.g., up to 3 years of outpatient treatment [8].

MBT is a psychological intervention rooted in psychodynamic and attachment theory, developed specifically to treat BPD [9]. MBT was originally designed as a long-term program, with 18 months of psychotherapy consisting of four components: (1) psychoeducation, (2) case formulation, (3) group therapy, and (4) individual therapy. Mentalization refers to the ability to understand one’s own and others’ mental states. The theoretical assumption is that BPD patients are more vulnerable to lose the capacity to mentalize when experiencing emotional distress. Increasing mentalization skills is assumed to minimize BPD symptoms such as emotional dysregulation, impulsivity, and suicidal ideation.

A systematic review with meta-analysis of all psychological therapies for BPD was published in 2020 [8]. The authors concluded that long-term MBT was more effective than treatment as usual in reducing self-harm, suicidality, and depression [8]. The original 18-month MBT program is prescribed by the MBT manual [9], but it is unclear if this is the most optimal treatment duration. We recently published a systematic review with meta-analysis assessing the beneficial and harmful effects of shorter-term versus longer-term psychotherapy for all adult mental health disorders [10], and we found no previous trial comparing the effects of different durations of MBT. The aim of the present trial was to assess the beneficial and harmful effects of short-term versus long-term MBT for outpatients with subthreshold or diagnosed BPD.

This trial is reported in accordance with the Consolidated Standards of Reporting Trials (CONSORT) 2010 statement [11] (online suppl. material pp. 1–2; for all online suppl. material, see https://doi.org/10.1159/000534289) and the methodological recommendations for trials of psychological interventions by Guidi et al. [12]. The trial was prospectively registered at ClinicalTrials.Gov (NCT03677037, registered on September 19, 2018) before the first participant was randomized. The methodology was predefined and described in detail previously [13, 14] (online suppl. material, pp. 3–20).

In this investigator-initiated, parallel group, single-centered, randomized clinical superiority trial, we consecutively screened patients from the Outpatient Clinic for Personality Disorders at Stolpegaard Psychotherapy Centre in the Mental Health Services of the Capital Region of Denmark. Eligible participants were adults (≥18 years) with subthreshold or diagnosed BPD according to the diagnostic criteria provided in the Diagnostic and Statistical Manual for Mental Disorders, 5th edition (DSM-5) [1] who provided written informed consent. Exclusion criteria were reduced intellectual functioning (IQ <75), a full diagnosis of schizotypal personality disorder or antisocial personality disorder according to DSM-5, a comorbid mental health disorder requiring specialist treatment, current (past 2 months) substance dependence assessed with the Mini International Neuropsychiatric Interview (MINI) [15], concurrent psychotherapeutic treatment, inability to understand Danish, and lack of informed consent.

The trial was overseen by a steering committee and by an independent data and safety monitoring committee who reviewed the data and performed one predefined, blinded interim analysis after half of the target sample size had been randomized. The trial was approved by the Danish Data Protection Agency (I-suite: 6553; August 30, 2018) and the Regional Ethics Committee of Health Research of the Capital Region of Denmark (H-18023136; August 27, 2018).

Copenhagen Trial Unit, a Danish center for clinical intervention research, was responsible for the central randomization. Participants were randomly assigned (1:1) to short-term MBT or long-term MBT by an independent investigator not otherwise involved in the trial according to a computer-generated allocation sequence with permuted blocks of various sizes. Randomization was stratified by sex (male/female) and low or high (0-11/12-36) BPD symptom severity at baseline assessed with the Zanarini Rating Scale for Borderline Personality Disorder (ZAN-BPD) [16].

Outcome assessors, the data safety and monitoring committee, data managers, statisticians, and the steering committee were blind to treatment allocation [13, 14]. Trial participants and therapists were not blind to treatment allocation due to the difficulties of implementing efficient blinding procedures in trials assessing psychological interventions [17]. Blinding was broken on December 18, 2022, after the statistical report was finalized and after the steering group had agreed on two blinded abstracts.

Patients were initially screened for eligibility during their routine intake interviews at the Outpatient Clinic for Personality Disorders. A psychiatrist or consultant psychiatrist performed a psychiatric assessment, assessed patient eligibility, provided the patient with a trial-specific information booklet, and obtained written informed consent. Eligible and consenting patients were then seen by a researcher who confirmed eligibility with the MINI interview [15] and the Structured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD) [18], collected demographic information, and performed baseline assessments before randomization.

Three external MBT experts, who were blind to treatment allocation, assessed adherence to the MBT manual in both short-term and long-term MBT using the Mentalization-Based Treatment Adherence and Competence Scale [19]. All raters were certified by one of the developers of the scale (A.B.). A total of 270 videos were obtained. From these videos, 59 videos (22%) from different treatment stages were randomly chosen for adherence ratings.

Short-term MBT was organized as a 20-week program of therapy in closed groups, commencing with five sessions of introduction to MBT followed by 15 weekly sessions of group MBT. Group therapy was accompanied by individual sessions every second week. The participants were treated by two group therapists, one of them also being the individual therapist, i.e., conjoined therapy. The MBT therapists followed the manual developed by Bateman and Fonagy [9].

Long-term MBT was organized as a 14-month program. All participants randomized to long-term MBT initially entered a 6-week introduction to MBT. When the introduction group finished, participants were allocated to one of eight slow-open MBT treatment groups. Treatment was then organized as 12 months of weekly group therapy sessions combined with individual therapy every second week. The long-term MBT group consisted of a total of 13.5 months of active treatment, but since there was some waiting time from the MBT intro group to the beginning of the MBT treatment group, we pragmatically defined the long-term MBT intervention as 14 months of duration. The participants were treated by two group therapists and a third individual therapist, i.e., combined therapy. Both group and individual therapies were manualized by Bateman and Fonagy [9]. More information about the trial interventions can be found in the online supplementary material, p. 7.

The primary outcome was severity of borderline symptomatology assessed with the ZAN-BPD interview by a blinded assessor [16]. Secondary outcomes were functional impairment assessed with the Work and Social Adjustment Scale (WSAS) [20], quality of life assessed with the Short-Form Health Survey (SF-36) mental component [21], global functioning assessed with the Global Assessment of Functioning scale (GAF) [22], and severe self-harm defined as the proportion of participants with one or more deliberate act of self-harm resulting in visible tissue damage (not including less severe self-harm, e.g., head-banging, superficial cutting, or hitting oneself without tissue damage).

All outcomes were assessed at 8 months, 16 months, and 24 months (will be presented in a separate publication) after randomization. We predefined the 16-month time point as the assessment time point of primary interest to ensure that the participants were assessed at similar time points after randomization regardless of the length of their intervention.

Safety was assessed as serious adverse events. Two blinded outcome assessors independently went through patient medical records to assess serious adverse events using the International Conference on Harmonization of technical requirements for registration of pharmaceuticals for human use – Good Clinical Practice (ICH-GCP) definition of a serious adverse event (any untoward medical occurrence that resulted in death, was life-threatening, required hospitalization or prolonging of existing hospitalization and resulted in persistent or significant disability or jeopardized the participant) [23].

We calculated that 166 participants were needed to provide 80% power at a 5% significance level for the primary outcome ZAN-BPD. We defined 3.5 points as a minimal important difference for ZAN-BPD.

The analyses followed the statistical analysis plan, which was published before unblinding of the data [14]. All continuous outcomes were analyzed using linear regression and dichotomous outcomes were analyzed using logistic regression. The analyses were based on an intention-to-treat population and were primarily adjusted for the baseline value of the outcome of interest and the stratification variables used in the randomization.

We used a five-step procedure [24] to assess if the thresholds for statistical and clinical significance were crossed, and we handled missing data according to the procedure suggested by Jakobsen et al. [25]. A more detailed explanation of our statistical methodology is available in our published and predefined statistical analysis plan [14].

Between October 4, 2018, and December 3, 2020, 207 patients were initially screened for eligibility to the trial. We randomly assigned 166 participants to short-term MBT (n = 84) or long-term MBT (n = 82) (Fig. 1).

In the short-term MBT group, five participants did not receive the allocated intervention, and 12 participants discontinued the intervention prematurely for various reasons. One participant randomized to short-term MBT received two consecutive short-term MBT group interventions. One participant randomized to short-term MBT crossed over to long-term MBT. According to our protocol, they were both included in the primary analysis per intention-to-treat principle. In the long-term MBT group, four participants did not receive the allocated intervention, and 25 discontinued the intervention prematurely for various reasons (Fig. 1).

Baseline sociodemographic and clinical characteristics can be found in Table 1. The psychopharmacological co-interventions were delivered similarly in both groups across the trial period (online suppl. material, p. 22). Overall, the two groups were balanced across baseline characteristics.

Participants allocated to short-term MBT attended a mean (SD) of 12 (6.2) individual therapy sessions and 14 (6.9) group sessions over a period of 5 months. Participants allocated to long-term MBT attended a mean of 17 (11) individual therapy sessions and 24 (14) group sessions over a period of 14 months. Attendance data were missing for 21 participants. These data are based on the intention-to-treat population. Attendance data for the protocol population can be found in the online supplementary material, p. 22.

Independent and certified coders rated a random selection of 35 individual sessions and 24 group sessions with the Mentalization-Based Treatment Adherence and Competence Scale [19]. The mean (SD) score was 4.46 (0.94) (n = 30 videos) in the short-term MBT and 4.79 (1.1) in the long-term MBT (n = 29 videos) (≥3.5 was considered adherent) with no evidence of a difference between short-term and long-term MBT (t test: p = 0.22). Overall adherence to MBT was good for both individual therapy (4.3, 0.9) and group therapy (4.7, 0.8) in the short-term group and individual therapy (4.6, 1.1) and group therapy (5.0, 1.1) in the long-term group. Table 2 summarizes primary and secondary outcome data in the intention-to-treat population.

Linear regression showed no evidence of a difference between short-term and long-term MBT at 16 months when assessing borderline symptoms (ZAN-BPD MD 0.99; 95% CI: −1.06 to 3.03; p = 0.341). 14 (16.7%) of 84 participants in the short-term MBT group were lost to follow-up compared with 22 (26.8%) of 82 participants in the long-term MBT group. Sensitivity analysis showed that missing data had the potential to alter the results (online suppl. material, pp. 23–107). None of the preplanned subgroup analyses (baseline severity of borderline symptoms; sex; age; baseline global functioning; baseline proportion of participants with severe self-harm; proportion of participants who had their group therapy temporarily paused due to COVID-19) showed evidence of a difference (Fig. 2). The outcome assessed at 8 months after randomization showed similar results (online suppl. material, pp. 23–107). Inter-rater reliability for ZAN-BPD was high (intraclass correlation: 0.92) based on the ratings of three independent assessors of a random subsample of 40 participants.

Linear regression showed no evidence of a difference between short-term and long-term MBT at 16 months when assessing the level of functioning (WSAS MD 1.44; 95% CI: −1.43 to 4.32; p = 0.321). The outcome assessed at 8 months after randomization showed similar results (online suppl. material, pp. 23–107).

Linear regression showed no evidence of a difference between short-term and long-term MBT at 16 months when assessing quality of life (SF-36 MD −0.91; 95% CI: −4.62 to 2.79; p = 0.626). The outcome assessed at 8 months after randomization showed similar results (online suppl. material, pp. 23–107).

Linear regression showed no evidence of a difference between short-term and long-term MBT at 16 months when assessing global functioning (GAF MD −2.25; 95% CI: −6.70 to 2.20; p = 0.318). The outcome assessed at 8 months after randomization showed similar results (online suppl. material, pp. 23–107).

At 16 months, 13 of 68 (19.1%) participants performed self-harm in the short-term MBT group compared with 16 of 61 (26.2%) participants in the long-term MBT group (RR 1.37; 95% CI: 0.70–2.84; p = 0.335). The outcome assessed at 8 months after randomization showed similar results (online suppl. material, pp. 23–107).

At 16 months after randomization, 15 of 84 (17.9%) participants experienced a serious adverse event in the short-term MBT group compared with 24 of 82 (29.3%) of the participants in the long-term MBT group (RR 1.63; 95% CI 0.94–3.07; p = 0.088) (online suppl. material pp. 23–107). Logistic regression showed no evidence of a difference but indicated a harmful effect of long-term MBT primarily driven by a difference in psychiatric hospitalizations: 9 of 84 (10.7%) participants had a psychiatric hospitalization in the short-term MBT group compared with 18 out of 82 (22%) in the long-term MBT group (RR 2.03; 95% CI: 0.99–5.09; p = 0.056) (online suppl. material, p. 108). A table of individual serious adverse events can be found in the online supplementary material, p. 110.

This is the first randomized clinical trial assessing short-term versus long-term MBT for adult outpatients with BPD. Our analyses showed that at 16 months after randomization, no evidence of a difference was observed either when assessing BPD symptomatology, level of functioning, quality of life, global functioning, and self-harm. Serious adverse events occurred more often in the long-term MBT group primarily driven by more psychiatric hospitalizations, but this result was not statistically significant. The therapists in the trial showed good adherence to the MBT manual, and were found to be delivering MBT as intended.

Our trial has important practical implications for the treatment of BPD, as the high prevalence of the disorder combined with high costs of existing treatments causes a significant barrier to accessing evidence-based psychotherapies for BPD in many parts of the world [26]. Given that access to evidence-based psychotherapy for BPD is a problem for many patients due to scarce resources in the mental health systems, shortening the overall length of psychotherapy in outpatient mental health settings may increase access to evidence-based care. Our trial challenges the prevailing opinion that people with personality disorders require long-term psychotherapy and supports the use of briefer psychotherapy programs for BPD. Based on our results, administrators and clinicians in the mental health system may give serious consideration to short-term MBT as a desirable option for outpatients with BPD.

In our trial, more patients in the long-term MBT group experienced a serious adverse event primarily driven by a difference in psychiatric hospitalizations. As the follow-up period was identical for the compared groups (i.e., 16 months after randomization), our analyses might indicate a “true” difference in the risk of serious adverse events, because the participants in both groups were observed in comparable follow-up time periods. While this difference was not statistically significant, it is still worth commenting on, as the confidence interval (0.99–5.08) showed that benefit could almost be excluded and an increased risk of psychiatric hospitalizations in the long-term MBT group was possible. Furthermore, thresholds for statistical significance should be less conservative when considering increased risks of harmful effects of experimental interventions [27]. The difference in serious adverse events might be a chance finding (“random error”); MBT may overwhelm BPD patients and lead to psychiatric hospitalizations; and/or being in MBT may increase odds that patients contact acute psychiatric care in times of crisis.

This trial has several strengths. First, we used a rigorous trial methodology to reduce the risks of bias, and our methodology was predefined in our published protocol and detailed statistical analysis plan prior to unblinding the results [13, 14]. Second, we assessed both beneficial and harmful effects and included objective assessments of serious adverse events, which is rare in psychotherapy trials for BPD [8]. Third, our trial has a high degree of external validity, as it was conducted in a real-world outpatient setting, and the excellent compliance with MBT suggests that this approach could be acceptable for implementation in routine clinical use with appropriate training and supervision. Fourth, our analyses showed limited signs of heterogeneity as all subgroup analyses showed no evidence of a difference, and our results were consistent across most outcomes. Fifth, all participants were assessed at the same time points regardless of their allocated intervention taking into account “regression toward the mean” potentially influencing the present results. Sixth, the psychopharmacological treatment was comparable between groups at all time points, reducing the risk of differences in psychopharmacological treatment influencing the present results.

This trial also has limitations. First, consistent with other trials assessing participants with BPD [8], we experienced a large proportion of missing outcome data. Large proportions of missing data should always warrant cautious interpretations of results [25]. Second, we included participants with subthreshold or diagnosed BPD. Hence, our trial population was heterogeneous and, on average, showed lower baseline BPD symptoms, including lower proportions of self-harm and higher levels of global functioning, as compared to previous trials assessing the effects of MBT for BPD [28, 29]. Our analyses did not indicate that the effects of MBT were influenced by baseline severity, but subthreshold BPD is not a diagnosis defined in either DSM-5 [1] or the International Classification of Diseases – eleventh version (ICD-11) [30] which is a limitation of the present trial. Our results may only be generalized to the population we included in the trial, i.e., participants with either subthreshold or diagnosed BPD. This should be considered, when interpreting our results. We chose this approach for pragmatic reasons, as many outpatients in everyday clinical practice do not meet the criteria for a full BPD diagnosis but are still functionally impaired, and we wished to include these patients to increase the external validity of the trial. Similar inclusion criteria have been used in other trials assessing psychotherapy for BPD [31, 32]. Third, the long-term MBT group in this trial was 14 months of MBT consisting of psychoeducation followed by weekly group therapy combined with individual therapy every second week. Our long-term MBT intervention is not identical to the original, manualized long-term MBT program, which is 18 months of treatment with both weekly group and individual therapies [9]. Hence, the evidence base for the long-term MBT group in this trial compared with treatment as usual is unclear. Fourth, clinicians and participants were not blind to the treatment allocation, which could potentially bias the results [17]. Fifth, we predefined the minimal clinically important difference (MCID) for the primary outcome as a 3.5-point superiority margin. However, this estimation was based on clinical experience with the population and an approximation of the expected standard deviation (SD/2) from previous trials [13]. Hence, the estimation of the MCID in this trial is uncertain and might be smaller or larger than 3.5. This has to be considered when interpreting the results. Furthermore, some of the items in the ZAN-BPD scale might reflect more severe problems than others, and the MCID might depend on the baseline severity, further leading to problems interpreting our present results. Finally, due to the COVID-19 pandemic, we experienced protocol deviations during the trial period, as group therapy in both trial interventions was temporarily paused for 2 months in 2020 and again in 2021. According to our subgroup analyses, these pauses did not seem to influence the outcomes, and they were identical in both groups.

Different short-term psychotherapies exist for patients with BPD, including brief dialectical behavior therapy (DBT) for 20 weeks [33], emotion regulation group therapy (ERGT) for 14 weeks [32], and systems training for emotional predictability and problem solving (STEPPS) for 20 weeks [34]. However, only one previous randomized clinical trial has assessed the effects of different durations of psychotherapy for BPD. The results of our trial are similar to a recent noninferiority trial assessing the effects of 6 versus 12 months of DBT for BPD [35], but this trial assessed the effects of a different intervention and included a narrower participant group, assessing only acute participants (eligibility criteria were one or more suicide attempts or acts of self-harm within the past 2 months). The authors concluded that the noninferiority hypothesis was supported for the primary outcome, total self-harm, and for several secondary outcomes.

In conclusion, long-term MBT did not lead to lower levels of BPD symptoms, nor did it influence any of the secondary outcomes compared with short-term MBT. In light of scarce treatment resources for BPD, briefer psychotherapy (5 months instead of 14 months) may be considered a desirable option for outpatients with BPD.

We thank the trial participants and the trial clinicians from the Outpatient Clinic for Personality Disorders at Stolpegaard Psychotherapy Centre, Mental Health Services in the Capital Region of Denmark; the independent members of the Data Safety and Monitoring Committee (Prof. Ole Jakob Storebø and Prof. Mickey Kongerslev); our student assistants (Mathilde Hasselby-Andersen, Anders Sonne Munch, Amanda Ark Søndergaard, Laura Alsing Juul, Rebecca Engel Thomas, Rebecca Kjær Andersen, and Olivia Frost); and the independent coders of MBT adherence (Nana Lund Nørgaard and Anne Blom Corlin).

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Regional Ethics Committee of Health Research of the Capital Region of Denmark (approval number H-18023136; August 27, 2018) and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. All participants provided written informed consent.

A.B. declares receiving book royalties from Oxford University Press and American Psychiatric Publishing Inc. for publishing books on mentalizing. A.B. declares payment for training courses in MBT from Anna Freud Centre, London. No other disclosures were reported.

The trial was funded by Trygfonden (S.J., Grant No. 123488) and by the Mental Health Services in the Capital Region of Denmark Research Fund (S.J. and S.S., Grant No. N/A). The funder had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author had access to all study data and had final responsibility for the decision to submit it for publication.

S.J. and S.S. conceptualized the trial, secured the funding, designed the short-term MBT intervention in collaboration with A.B. and clinicians at the outpatient clinic, and were the principal investigator (S.J.) and sponsor (S.S.) of the trial. PS provided organizational support for the trial. F.W.F. assisted in the clinical implementation of the trial. The trial was overseen by a steering committee consisting of S.J., S.P., P.S., J.C.J., S.L., and S.S. S.J. recruited participantsand performed baseline assessments assisted by E.H., C.K.J., and M.R. Blinded outcome assessments were performed by E.H., C.K.J., M.R., and several student assistants. A.B. and S.B. trained and supervised the staff in both short-term and long-term MBT. J.C.J. supervised the design of the trial protocol and statistical analysis plan. M.H.O. and J.C.J. independently performed blinded statistical analyses. S.J., J.C.J., and M.H.O. had full access to all the blinded trial data and took responsibility for the integrity of the data and the accuracy of the data analysis. S.B. assisted in the coding of adherence to MBT. S.J., J.C.J., and S.S. drafted the paper. All authors commented on and approved the final version of the paper.

Data from this trial are not publicly available due to ethical grounds. Access to deidentified data from this trial may be obtained by contacting the corresponding author. All requests for data access must be approved by the Regional Ethics Committee of Health Research of the Capital Region of Denmark and the institution with whom the primary investigator is affiliated. Data access will be governed by a data-sharing agreement executed between the Capital Region of Denmark and the institution with whom the primary investigator is affiliated.