A pilot randomized controlled trial of mentalization-based treatment and structured clinical management for outpatients with borderline personality disorder in South Korea

Abstract

Background

To evaluate the effectiveness of mentalization-based treatment (MBT) for borderline personality disorder (BPD) compared to structured clinical management (SCM) in a South Korean clinical setting.

Methods

This single-blind pilot randomized controlled trial included 64 individuals with a primary diagnosis of BPD who underwent either MBT (n = 37) or SCM (n = 27) in an outpatient setting over 12 months. Both interventions included weekly individual and group sessions. The primary outcome measure was the number of suicide attempts. Secondary outcomes included depressive symptoms, suicide risk, hospitalization frequency and duration, mentalization difficulties, adult attachment styles, BPD symptom severity, and resilience. Outcomes were assessed at baseline and 6 and 12 months. Analyses were conducted using generalized estimating equations, following the intention-to-treat principle.

Results

Both treatment groups demonstrated a significant reduction in suicide attempts over 12 months, with no significant between-group differences in the primary outcome. However, participants receiving MBT showed consistent and sustained improvements in depressive symptoms, mentalization difficulties, and resilience. MBT was also associated with a significant reduction in attachment anxiety at 12 months, whereas the SCM produced earlier reductions in suicide risk and improvements in psychic equivalence. No significant changes were observed in attachment avoidance or in the hasty and incomplete mentalizing domains in either group.

Conclusions

This pilot trial provides preliminary evidence that both MBT and SCM were effective in reducing suicidal behaviors and enhancing psychological functioning among Koreans with BPD. MBT demonstrated greater long-term benefits in emotion regulation and adaptability, suggesting that it may serve as a culturally scalable intervention in non-Western clinical settings.

Background

Borderline personality disorder (BPD) is a complex and debilitating mental health condition characterized by emotional dysregulation, unstable interpersonal relationships, impulsivity, and a heightened risk of self-harm and suicide [1, 2]. Individuals diagnosed with BPD frequently experience significant distress, functional impairment, and high utilization of mental health services, imposing a substantial socioeconomic burden [3, 4]. BPD has long been stigmatized as treatment-resistant, which has frequently limited access to proper diagnosis and care [5]. However, the development of structured, manualized psychotherapies—supported by robust evidence from Western contexts—has now established BPD as a treatable condition [6–8]. Nevertheless, in many Asian cultures, barriers to an accurate diagnosis likely contribute to the underestimation of the prevalence [9], and the application of evidence-based treatments remains limited. Mentalization-based treatment (MBT), which is grounded in attachment theory [10, 11] and neuroscience [12], has consistently demonstrated efficacy and is now recognized as a core evidence-based treatment for BPD. By strengthening individuals’ capacity to mentalize—that is, to understand and reflect on their own and others’ mental states within interpersonal contexts—MBT directly targets emotional dysregulation and relational instability central to BPD [10]. The standard MBT program integrates weekly individual and group therapy sessions, each serving distinct yet complementary therapeutic functions that work synergistically to enhance mentalizing capacity across different relational contexts [10, 12, 13].

The efficacy of MBT in patients with BPD has been well documented in multiple randomized controlled trials (RCTs). Studies comparing MBT with specialist treatment-as-usual (S-TAU) have shown significant improvements in symptoms in both conditions [14, 15]. Other RCTs employed structured clinical management (SCM), a well-defined, supportive treatment protocol designed to control for the effects of structured care, and found that MBT led to greater and more enduring improvements, including reduced self-harm, hospitalization, and functional impairment [16]. Further evidence suggests that MBT may be particularly effective in patients with BPD and comorbid personality disorders [17]. However, as these findings are largely derived from Western populations, it remains critical to investigate whether the effectiveness of MBT extends to non-Western contexts.

In South Korea, cultural norms are characterized by collectivistic values that emphasize social harmony, emotional restraint, and hierarchical social relationships, which differ markedly from the individualistic frameworks prevalent in Western societies [18]. Within such contexts, emotional expression is often suppressed to preserve interpersonal harmony, and psychological distress may be communicated indirectly, for example through somatic complaints rather than explicit emotional disclosure [18, 19]. These norms may shape the manifestation of BPD, potentially contributing to heightened sensitivity to perceived rejection within hierarchical relationships and difficulties in articulating internal emotional states [20, 21].

Importantly, mentalization itself may be culturally embedded. While Western models of mentalization emphasize explicit, verbal reflection on one’s own and others’ mental states, Korean cultural contexts may privilege more implicit and context-dependent interpretations of intentions within established social roles [20]. The interdependent self-construal common in Korean society, in which identity is defined through relational obligations, may therefore interact with attachment-based interventions such as MBT in ways that differ from Western settings [19, 20]. Moreover, persistent stigma surrounding mental health treatment in Korea may influence help-seeking behaviors, therapeutic engagement, and willingness to disclose vulnerable emotional experiences in both individual and group therapy [9, 19].

Together, these cultural factors raise critical questions regarding whether the mechanisms of change proposed by MBT operate similarly across cultural contexts or whether cultural adaptation may be necessary. Empirical investigation in non-Western populations is thus essential to determine the cross-cultural applicability and effectiveness of MBT [19–21]. Despite growing recognition of these issues, empirical research on mentalization in individuals with BPD in South Korea remains limited. Prior studies have examined mentalization capacity and its associations with clinical characteristics, attachment patterns, and resilience in Korean samples, and preliminary reports have suggested that changes in mentalization may be related to symptomatic improvement [21, 22]. However, these studies have primarily relied on cross-sectional designs or uncontrolled, preliminary observations of treatment effects. To date, no RCT has systematically evaluated the effectiveness of MBT compared with an active control condition in a Korean clinical setting.

Against this cultural and clinical background, this study aimed to address this gap by evaluating the effectiveness of MBT in a South Korean clinical setting. Based on prior research, this study was designed as a 12-month RCT comparing MBT with SCM [16]. SCM offers structured and supportive therapy combined with case management aimed at enhancing general coping skills without a specific focus on improving mentalization [23]. Psychological and behavioral outcomes were assessed at baseline, and at 6, 12 months after treatment initiation, including the number of suicide attempts, hospitalization frequency and duration, BPD symptom severity, depressive symptoms, suicide risk, attachment characteristics, adverse childhood experiences, mentalization difficulties, and resilience.

This pilot study contributes to the growing body of literature by providing preliminary empirical evidence for the comparative effectiveness of these treatments in a non-Western setting. These findings are expected to inform the future development of standardized, evidence-based treatment protocols tailored to the unique needs of Korean patients with BPD.

Methods

Protocol and design

This study was designed and conducted as a pilot and feasibility RCT to evaluate the initial acceptability and effectiveness of MBT in a Korea clinical setting. Patients referred for specialized outpatient treatment of personality disorders at Gangnam Severance Hospital and Yeungnam University Hospital were randomly assigned to one of the two active treatment conditions. The study was approved by the institutional review boards of both hospitals (Gangnam Severance Hospital IRB No. 3–2021–0095; Yeungnam University Hospital IRB No. 2021–02-046), and was conducted in accordance with the Declaration of Helsinki. Assessments were conducted at baseline, and at 6 and 12 months following treatment initiation, with the 12-month assessment corresponding to the end of treatment. The trial was preregistered in the Clinical Research Information Service (CRIS; KCT0008228; registration date: 23 February 2023). Patients were provided with written information about the study and provided informed consent after receiving a full explanation. The treatments were funded by the Ministry of Health and Welfare, Republic of Korea. The participants did not receive any financial compensation, and no costs for treatment were charged.

Participants

Outpatient participants were recruited between 2021 and 2023. For eligibility, BPD had to be the primary diagnosis, as confirmed by the principal investigator of this research, a board-certified psychiatrist. The diagnosis was verified based on the Diagnostic and Statistical Manual of Mental Disorders, 5^th edition (DSM-5) criteria, and assessed using the Structured Clinical Interview for DSM-5 Personality Disorders (SCID-5-PD), administered by trained clinical psychologists or psychiatry residents. Information on psychotropic medication use at baseline was obtained through review of electronic medical records and prescription information provided by participants. The inclusion criteria were as follows: (1) diagnosis of BPD and (2) age 18–50 years. The exclusion criteria were as follows: (1) intellectual disability (full-scale IQ < 70) or organic brain disorder; (2) diagnosis of a psychotic disorder or bipolar I disorder according to the DSM-5; and (3) medical conditions likely to significantly affect mental status (e.g., cancer or other chronic illnesses). One hundred patients were assessed for eligibility. Of these, 13 patients did not meet the criteria for BPD and were excluded from the study. Among the 87 patients initially enrolled, 19 did not attend the diagnostic interview, three withdrew consent before randomization, and one could not be contacted. As a result, 64 patients were randomly assigned to one of the two treatment programs (MBT = 37, SCM = 27).

Randomization

Following consent, enrollment, and baseline assessment, the participants were randomized by a data management researcher at Gangnam Severance Hospital using a computer-generated random number sequence with a 2:1 allocation ratio (MBT:SCM). This ratio was determined a priori based on sample size estimation to ensure adequate statistical power for suicide-related outcomes, a clinically central endpoint in BPD, and is consistent with prior randomized controlled trials comparing MBT and SCM [16].

Interventions

MBT was implemented in accordance with the treatment guidelines described by Bateman and Fonagy (2006) [13]. Although standard outpatient MBT is typically delivered over 18 months [13, 16], the treatment duration in the present study was modified to 12 months to enhance feasibility within the Korean clinical setting, where access to long-term intensive psychotherapy remains limited due to a shortage of clinicians with specialized training in personality disorder treatment. This decision was informed by prior RCTs demonstrating meaningful clinical improvements with 12-month MBT protocols [16, 24], as well as by pragmatic considerations related to service delivery, patient retention, and the need to establish initial evidence for MBT effectiveness in Korean populations before implementing longer-term protocols. The modified 12-month program consisted of weekly individual therapy sessions (30 minutes) and weekly group therapy sessions (75 minutes) provided by psychiatrists or clinical psychologists who had completed basic or practitioner level MBT workshops. These workshops provided structured training in the theoretical framework, therapeutic stance, and core intervention strategies of MBT. Formal, ongoing MBT-specific supervision was not provided during the study period.

Individual MBT sessions focused on stabilizing affect, exploring attachment-related distress, and fostering moment-to-moment reflection on the patient’s internal emotional experiences within the therapeutic relationship [13]. Through close attention to both the patient’s and therapist’s mental states, individual sessions aimed to enhance the patient’s capacity to pause, reflect, and regulate emotional responses under stress, rather than responding impulsively [10, 12]. Group MBT sessions provided a structured interpersonal context in which patients could practice mentalizing in vivo [10, 13]. The group functioned as a “social laboratory,” allowing participants to observe multiple perspectives, test interpersonal assumptions, and receive real-time feedback on how their behaviors affected others. Interpersonal misunderstandings, emotional arousal, and relational ruptures that naturally emerged within the group were actively used as opportunities to promote mentalizing and reflective functioning [12, 13]. The integration of individual and group MBT was considered essential, as insights gained in individual sessions could be tested and consolidated in the group setting, while interpersonal challenges arising in the group could be explored in greater depth during individual therapy, thereby facilitating the generalization of mentalizing skills to patients’ everyday relationships [10, 12, 13].

SCM was performed using the same treatment duration and structure as MBT; however, there were differences in the therapeutic approaches. SCM was delivered in accordance with the treatment guidelines described by Mitchell et al. (2022) [23]. It provides a consistent and supportive therapeutic framework that emphasizes stabilization, problem-solving, and symptom management without a specific focus on enhancing mentalization capacity [16, 23]. Individual SCM sessions followed a structured format that included review of the patient’s week, identification of current problems, collaborative problem-solving using practical strategies, psychoeducation about BPD symptoms and triggers, and crisis planning when needed [23]. Therapists adopted a more directive, problem-focused stance compared to MBT’s exploratory approach, offering concrete advice and guidance to help patients manage immediate difficulties [23]. Group SCM sessions relatively focused on psychoeducation about BPD, skill-building (e.g., distress tolerance, emotion regulation), and structured problem-solving discussions, but did not emphasize exploration of mental states or interpersonal processes within the group as in MBT [23]. SCM was delivered by psychiatrists or clinical psychologists. Both individual and group SCM therapists did not participate in formal SCM workshops; instead, they prepared for treatment delivery through structured team-based study of the SCM manual [23] prior to study initiation.

However, despite the use of standardized treatment manuals and therapist training, formal assessments of treatment fidelity or therapist adherence were not conducted.

Measures

Suicidal attempt

Suicide attempts were assessed using the Columbia-Suicide Severity Rating Scale (C-SSRS), a widely used clinician-administered instrument for evaluating suicidal ideation and behaviors [25]. The C-SSRS assesses both the severity of suicidal ideation and the occurrence of suicidal behaviors during the preceding six-month period at each assessment timepoint (baseline, 6 months, and 12 months after treatment initiation). Suicidal ideation was rated on a five-point scale (1 = wish to be dead to 5 = active suicidal ideation with a specific plan and intent), and suicidal behavior was evaluated based on the presence of actual attempts, interrupted or aborted attempts, and preparatory actions. In this study, a suicide attempt was defined as self-injurious behavior accompanied by at least some degree of intent to die. This act may lead to physical injury, or even if no injury occurs, it may involve a risk of harm.

Suicidal risk

Suicidal risk was evaluated using the suicidal risk section of the Protective and Vulnerable Factors Battery Test (PROVE-SR) developed and validated in a previous study [26]. The PROVE-SR consists of six items, and a weighted scoring system is used to calculate the total score. The total score ranges from 0 to 20, with higher scores indicating greater suicidal risk.

Hospitalization

The frequency and duration of psychiatric hospitalizations during the preceding six-month period were assessed through clinician-administered interviews at baseline, 6 months after treatment initiation, and 12 months after treatment initiation(endpoint).

Depressive symptoms

Depressive symptoms were assessed using the depressive symptom section of the PROVE battery (PROVE-DS), which was developed and validated in a previous study [26]. The PROVE-DS evaluates depressive symptoms experienced over the past two weeks, with higher scores indicating greater symptom severity. The scale consists of 15 items, each rated on a 5-point Likert scale ranging from 0 to 4, with total scores ranging from 0 to 48.

Mentalization difficulties

Mentalization difficulties were measured using the mentalization capacity section of the PROVE battery (PROVE-MC), which was developed and validated in a previous study [26]. The scale consists of 16 items, each rated on a 5-point Likert scale, ranging from 0 to 4. PROVE-MC evaluates five subdomains related to mentalization difficulties: lack of emotional awareness, lack of emotional expression, psychic equivalence, hasty and incomplete mentalization, and lack of mentalizing others. Higher total scores indicate greater mentalization impairment.

Attachment characteristics

Attachment characteristics were evaluated using the Adult Attachment Type (AAT) section of the PROVE battery (PROVE-AAT) developed and validated in a previous study [26]. This measure assesses attachment anxiety and avoidance and reflects the overall characteristics of an individual’s current relationships. The scale comprises nine items, with a total score ranging from 9 to 63.

Symptom severity of BPD

Self-reported BPD symptoms were evaluated using the Korean validated version of Borderline Scale of the Personality Assessment Inventory (PAI-BOR) [27]. Twenty-three items were answered on a 4-point Likert scale ranging from 0 to 3. The total score ranged from 0 to 69, with higher scores indicating greater severity of BPD symptoms.

Resilience

Resilience was assessed using the Korean Resilience Questionnaire subscale of the PROVE battery (PROVE-KRQ), developed and validated in a prior study [26]. The scale comprises 53 items rated on a 5-point Likert scale ranging from 0 to 4, with total scores ranging from 0 to 212. It has three subdomains: self-regulation, interpersonal relationships, and psychological optimism. Higher scores indicated greater resilience.

Statistical analyses

Statistical analyses were performed using SPSS version 27 (IBM, Armonk, NY) with default settings unless otherwise specified. All analyses adhered to standard procedures for handling repeated measures and missing data. Data analyses followed the intention-to-treat (ITT) principle [28] and incorporated all available data from follow-up assessments at 6 and 12 months. Patients were included in the ITT sample if they were randomized and attended the baseline assessment session where they were informed of the treatment conditions.

Baseline sociodemographic and clinical characteristics were compared using the chi-squared test for categorical variables and the independent-sample t-test for continuous variables. Given that these comparisons were exploratory and covariates were accounted for in the statistical model, no multiple comparison corrections were applied.

Outcome measures were analyzed using generalized estimating equations (GEEs) to account for within-subject correlations across repeated measures. Age and sex were included as covariates to control for potential confounding effects, with sex recorded based on medical records rather than self-identified gender. Psychotropic medication use was specifically adjusted because of the observed systematic baseline differences between MBT and SCM in the chi-squared test analysis. The covariance structure for repeated measures over time was determined based on model-fit criteria to ensure optimal estimation. The GEE was selected over generalized mixed-effects models (GLMMs) because of its robustness in small sample sizes and its ability to provide population-averaged estimates without requiring distributional assumptions for random effects. Given the limited sample size, GLMMs present estimation challenges, possibly owing to model complexity and parameter sensitivity. To ensure robustness and avoid convergence issues, GEEs were deemed a more reliable approach for analyzing longitudinal data in this study.

As Little’s MCAR test (p > 0.05) indicated that the missing data were missing completely at random (MCAR), no imputation was performed. The GEE was employed because it provides valid estimates under the assumption of missing data at random, which is a less restrictive assumption than MCAR.

Given the skewed distributions, most continuous outcome variables were analyzed using GEE gamma regression with a log link, incorporating first-order autoregression to account for within-subject correlations. Because gamma regression cannot handle zeros, a small value (0.01) was added to all scores if zeros existed. Zero-inflated count variables, including the number of suicide attempts, hospitalization days, and number of hospital visits, were analyzed using GEE with a Tweedie distribution (because of many zero counts) with a log link and first-order autoregression for the repeated part.

The proportion of missing data increased with each follow-up assessment. Overall, 51.56% of patients (33 of 64) completed all three assessments (baseline, 6 months, and 12 months) during the treatment period. There was no significant difference between MBT and SCM in terms of the number of patients who completed all three follow-up assessments: n = 18 (48.6%) for MBT and n=15 (55.6%) for SCM. The distribution of the complete categories did not differ between groups ( = 1.321, df = 2, p = 0.516). Figure 1 illustrates the progression of the participants throughout the trial.

Fig. 1.

CONSORT flow diagram. Note. MBT, mentalization-based treatment; SCM, structured clinical management

Results

Baseline characteristics

The baseline sociodemographic and clinical characteristics of the participants are summarized in Tables 1 and 2, with no significant differences between the MBT (n = 37) and SCM (n = 27) groups regarding age (MBT: 27.92 ± 7.213 years, SCM: 27.00 ± 7.109 years; p = 0.614), IQ (MBT: 101.71 ± 11.58, SCM: 101.65 ± 10.24; p = 0.983), or sex composition (MBT: 86.5% female, SCM: 85.2% female; p = 0.882). A significant difference was observed in medication usage, with a higher proportion of participants in the SCM group (88.9%) using psychotropic medication compared to the MBT group (70.3%) (χ² = 4.26, p = 0.039). Given this baseline imbalance, psychotropic medication use was included as a covariate in subsequent longitudinal analyses.

Table 1.

Baseline sociodemographic characteristics of the MBT and SCM groups

	MBT Group (n = 37)	SCM Group (n = 27)	p1
Age, mean (SD)	27.92 (7.213)	27 (7.109)	0.614
IQ, mean (SD)	101.71 (11.58)	101.65 (10.24)	0.983
Sex: female, n (%)	32 (86.5%)	23 (85.2%)	0.882
Education year, mean (SD)	14.84 (2.035)	14.33 (2.287)	0.356
Marital status, n (%)			0.947
Single	28 (75.7%)	22 (81.5%)
Married	4 (10.8%)	2 (7.4%)
Cohabiting	3 (8.1%)	2 (7.4%)
Divorced	2 (5.4%)	1 (3.7%)
Occupation, n (%)			0.430
Full-time job	9 (24.3%)	3 (11.1%)
Temporary job	5 (13.5%)	4 (14.8%)
Student	12 (32.4%)	11 (40.7%)
Unemployed	5 (13.5%)	1 (3.7%)
Others	4 (10.8%)	2 (7.4%)
Unknown	2 (5.4%)	6 (22.2%)
Death, n (%)			0.220
Yes	2 (5.4%)	0 (0%)
No	35 (94.6%)	27 (100%)
Household income, n (%)			0.366
High	7 (18.9%)	7 (25.9%)
Medium	17 (45.9%)	10 (37.0%)
Low	8 (21.6%)	4 (14.8%)
Unknown	5 (13.5%)	6 (22.2%)
Hospitalization, n (%)			0.721
Yes	15 (40.5%)	11 (40.7%)
No	18 (48.6%)	14 (51.9%)
Unknown	4 (10.8%)	2 (7.4%)
Medication, n (%)			0.039
Yes	26 (70.3%)	24 (88.9%)
No	8 (21.6%)	1 (3.7%)

MBT, mentalization-based treatment; SCM, structured clinical management; SD, standard deviation

¹ Significance level for group differences; t-test was used for continuous measures, chi-squared test was used for frequencies/proportions

Table 2.

Baseline clinical characteristics of the MBT and SCM groups

	MBT Group (n = 37)	SCM Group (n = 27)	p1
Suicide attempt past 6 months, mean (SD)	2.69 (3.022)	1.84 (2.495)	0.249
Number of hospital visits past 6 months, mean (SD)	1.49 (3.453)	0.44 (0.698)	0.082
Hospital duration past 6 months, mean (SD)	9.76 (15.014)	5.85 (9.482)	0.239
Depressive symptoms	33.00 (10.821)	31.65 (11.440)	0.639
Suicidal risk	10.50 (5.998)	8.88 (5.559)	0.285
Mentalization difficulties
Lack of emotional awareness	11.47 (3.989)	10.35 (3.286)	0.243
Lack of expression and interaction	8.67 (4.666)	7.19 (3.533)	0.162
Psychic equivalence mode	5.94 (2.028)	5.81 (1.650)	0.778
Hasty and incomplete mentalizing	6.19 (2.459)	4.69 (3.553)	0.070
Lack of mentalizing other	4.69 (3.553)	5.35 (2.813)	0.358
Lack of mentalizing situation	3.12 (1.883)	3.26 (1.657)	0.776
Total score, mean (SD)	36.94 (9.055)	32.69 (9.756)	0.082
Attachment characteristics (n, %)			0.069
Preoccupied	11 (29.7%)	14 (51.9%)
Dismissing	4 (10.8%)	0 (0.0%)
Confused	21 (56.8%)	12 (44.4%)
Secure	0 (0.0%)	0 (0.0%)
Unknown	1 (2.7%)	1 (3.7%)
Adverse childhood experiences
Emotional abuse	9.94 (3.726)	9.85 (3.414)	0.916
Physical abuse	13.39 (7.817)	10.96 (7.252)	0.219
Sexual abuse	4.97 (6.934)	4.85 (6.188)	0.941
Neglect	5.86 (7.068)	4.73 (6.515)	0.523
Domestic violence	10.08 (8.627)	9.81 (9.944)	0.908
Bullying	9.78 (6.821)	10.46 (6.153)	0.687
BPD symptom severity, mean (SD)	46.58 (13.708)	48.23 (11.497)	0.620
Resilience, mean (SD)	140.70 (31.787)	137.35 (22.496)	0.666

MBT, mentalization-based treatment; SCM, structured clinical management; BPD, borderline personality disorder; SD, standard deviation

¹ Significance level for group differences; t-test was used for continuous measures, chi-squared test was used for frequencies/proportions

Participants flow and attrition

In the MBT group, three participants were lost to follow-up due to loss of contact before the 12-month assessment, and 16 participants withdrew consent during the study period. In the SCM group, one participant was lost to follow-up due to loss of contact before the 6-month assessment, and 11 participants withdrew consent during the study period.

During the study period, two deaths occurred in the MBT group, both before the 6-month assessment. No deaths were observed in the SCM group. Participants who died were included in the intention-to-treat analyses up to the point of their last available assessment. Details of participant attrition and deaths are presented in the participant flow diagram (Fig. 1).

Primary outcome

There was a significant reduction in suicide attempts over time across both treatment conditions. The Wald χ² test for time showed a highly significant effect with MBT (χ² = 30.065, p < 0.001) and SCM (χ² = 9.409, p = 0.009) (Table 3). At the 6-month follow-up, the incidence rate ratios (IRRs) for suicide attempts in the MBT and SCM groups were 0.275 (95% confidence interval [CI]: 0.114–0.667, p = 0.004) and 0.084 (95% CI: 0.011–0.648, p = 0.017), respectively (Table 3). At the 12-month follow-up, the IRR further decreased in both groups (MBT, 0.041; 95% CI: 0.013–0.129; p < 0.001; SCM, 0.120; 95% CI: 0.017–0.865; p = 0.035), indicating a sustained treatment effect over time (Table 3). Figure 2Aalso displays the primary outcome variable estimates derived from the GEE used to assess the treatment effect and illustrates the rate of reduction in suicidal behaviors in the two groups. However, the Wald χ² test for group effect showed no significant difference between MBT and SCM (χ² = 1.924, p = 0.165), suggesting that both treatment modalities contributed similarly to the reduction in suicide attempts (Table 4).

Table 3.

Comparison of time effects for primary and secondary outcomes between the MBT and SCM groups

	MBT (n = 37)					SCM (n = 27)
Outcome and Assessment	Wald χ2	6-month IRR	6-month 95% CI	1-year IRR	1-year 95% CI		Wald χ2	6-month IRR	6-month 95% CI	1-year IRR	1-year 95% CI
Suicide attempt	30.065***d	0.275**	0.114/0.667	0.041***	0.013/0.129		9.409**	0.084*	0.011/0.648	0.120*	0.017/0.865
Number of hospital visits	14.911**	0.201**	0.067/0.602	0.106***	0.033/0.343		6.500*	0.294*	0.114/0.755	0.405	0.142/1.161
Hospital duration	8.658*	0.244*	0.063/0.944	0.168**	0.050/0.565		9.183*	0.214*	0.059/0.769	0.136*	0.027/0.696
Depressive symptoms	8.185*	0.840*	0.720/0.980	0.710**	0.557/0.905		5.243	0.911	0.744/1.114	0.767*	0.611/0.963
Suicidal risk	4.155	0.858	0.665/1.108	0.767 *	0.591/0.995		10.153**	0.597**	0.427/0.835	0.646**	0.490/0.851
Lack of emotional awareness	14.586**	0.886**	0.814/0.965	0.730**	0.608/0.877		4.253	0.888	0.427/0.835	0.844	0.706/1.009
Lack of expression and interaction	5.679	0.857	0.721/1.019	0.804*	0.668/0.969		2.121	0.859	0.700/1.054	0.904	0.690/1.184
Psychic equivalence mode	2.986	0.883	0.759/1.027	0.883	0.740/1.054		12.048**	0.841*	0.734/0.963	0.822**	0.710/0.953
Hasty and incomplete mentalizing	0.149	0.997	0.849/1.171	0.971	0.783/1.204		0.434	1.007	0.830/1.222	1.070	0.837/1.368
Lack of mentalizing other	5.250	0.865	0.664/1.129	0.671*	0.476/0.944		15.185**	0.763	0.557/1.047	0.524***	0.375/0.731
Mentalization difficulties, total	19.53***	0.891***	0.846/0.939	0.809**	0.716/0.914		5.337	0.907	0.810/1.017	0.859*	0.750/0.982
Attachment avoidance	2.272	0.999	0.926/1.077	1.041	0.977/1.109		4.342	0.988	0.909/1.073	0.928	0.842/1.024
Attachment anxiety	4.804	0.961	0.918/1.007	0.929*	0.864/0.998		5.744	0.923*	0.864/0.986	0.908	0.815/1.010
Resilience	17.771***	1.091***	1.042/1.141	1.132***	1.064/1.205		8.360*	1.096**	1.029/1.168	1.084	0.994/1.182
BPD symptom severity	14.761**	0.873***	0.810/0.940	0.87**	0.795/0.953		4.997	0.930*	0.873/0.991	0.886	0.779/1.009

MBT, mentalization-based treatment; SCM, structured clinical management; MBT, mentalization-based treatment; SCM, structured clinical management; BPD, borderline personality disorder

p < 0.05 ^**p < 0.01 ^***p < 0.001

a. Clinical countable outcomes are presented as incidence rate ratios (IRRs) with 95% confidence intervals (CIs), derived from exponentiated generalized estimating equation (GEE) model coefficients (Exp(β)); Psychological assessment scale outcomes are presented as relative risks (RRs), with 95% CI, derived from exponentiated GEE model coefficients (Exp(β))

Fig. 2.

Suicide attempts, depressive symptoms, and resilience scores over 12 months among patients randomly assigned to receive MBT or SCM for patients with borderline personality disorder. Estimates were derived from generalized estimating equation models used to assess the treatment effects. MBT, mentalization-based treatment; SCM, structured clinical management; CI, confidence interval

Table 4.

Time and group effects on primary and secondary outcomes using generalized estimating equations

	Time effect					Group effect
Outcome and assessment	Wald χ2	6-month (IRR/RR)	6-month 95% CI	1-year (IRR/RR)	1-year 95% CI	Wald χ2	Relative change (IRR/RR)	95% CI
Suicide attempt	23.817***	0.164***	0.062/0.434	0.071***	0.018/0.282	1.924	1.574	0.829/2.986
Number of hospital visits	17.849***	0.229***	0.108/0.484	0.190***	0.079/0.461	3.103	2.478	0.903/6.804
Hospital duration	16.161***	0.231**	0.088/0.607	0.155***	0.058/0.413	1.595	1.691	0.748/3.821
Depressive symptoms	13.852**	0.868*	0.765/0.985	0.727***	0.614/0.861	0.123	1.031	0.867/1.227
Suicidal risk	11.986**	0.724**	0.583/0.900	0.709**	0.583/0.862	2.108	1.242	0.927/1.663
Lack of emotional awareness	16.889***	0.886**	0.817/0.962	0.779***	0.684/0.887	0.769	1.084	0.905/1.298
Lack of expression and interaction	6.195*	0.858*	0.751/0.979	0.850*	0.723/0.998	1.636	1.195	0.909/1.572
Psychic equivalence mode	9.916**	0.868**	0.782/0.963	0.858**	0.762/0.965	0.373	1.049	0.899/1.225
Hasty and incomplete mentalizing	0.058	1.003	0.886/1.136	1.017	0.865/1.194	3.942 *	1.385 *	1.004/1.911
Lack of mentalizing others	16.901***	0.820	0.668/1.008	0.590***	0.457/0.760	0.083	0.963	0.746/1.243
Mentalization difficulties, total	17.778***	0.901***	0.848/0.958	0.834***	0.762/0.913	2.871	1.136	0.980/1.318
Attachment avoidance	0.129	0.995	0.941/1.052	0.990	0.936/1.047	4.861*	1.117*	1.012/1.232
Attachment anxiety	9.669**	0.943**	0.907/0.980	0.920**	0.864/0.979	0.759	1.035	0.958/1.118
Resilience	22.779***	1.094***	1.053/1.136	1.112***	1.056/1.170	0.671	1.033	0.955/1.118
BPD symptom severity	18.653***	0.898***	0.854/0.944	0.878**	0.813/0.948	0.008	0.995	0.881/1.123

BPD, borderline personality disorder

p < 0.05 ^**p < 0.01 ^***p < 0.001

^aClinical countable outcomes are presented as incidence rate ratios (IRRs) with 95% confidence intervals (CIs), derived from exponentiated generalized estimating equation model coefficients (Exp(β)); Psychological assessment scale outcomes are presented as relative risks (RRs), with 95% CI, derived from exponentiated GEE model coefficients (Exp(β))

^bGroup effect represents the incidence rate ratio (IRR) or relative ratio (RR) of the MBT group compared to the SCM group, where values greater than 1 indicate a higher incidence in the MBT group and values less than 1 indicate a lower incidence relative to the SCM group

Secondary outcomes

Hospitalization

The reduction in hospital visits and duration was statistically significant across both the MBT and SCM groups, as shown in Tables 3 and 4. The MBT group experienced a marked decrease in hospital visits at 12 months (IRR = 0.106, 95% CI: 0.033–0.343, p < 0.001) compared with the SCM group (IRR = 0.405, 95% CI: 0.142–1.161, p = 0.092) as presented in Table 3. Hospital duration followed a similar trend, with MBT showing a greater reduction at 12 months (IRR = 0.168, 95% CI: 0.050–0.565, p = 0.004) compared to SCM (IRR = 0.136, 95% CI: 0.027–0.696, p = 0.017) (Table 3). Although both groups showed similar downward trends in hospital utilization, Table 4 highlights that no significant group effect was found for hospital visits (χ² = 3.103, p = 0.078) or hospital duration (χ² = 1.595, p = 0.207). This suggests that, while both treatment modalities contributed to reduced hospital utilization over time, their relative effectiveness did not differ significantly.

Depressive symptoms

Figure 2B illustrates that both the MBT and SCM groups demonstrated an overall reduction in depressive symptoms over time, indicating the effectiveness of both treatments. MBT showed a more pronounced decline, with significant reductions from baseline to 6 months (relative risk [RR] = 0.840, 95% CI: 0.720–0.980, p = 0.027) and from baseline to 12 months (RR = 0.710, 95% CI: 0.557–0.905, p = 0.006), as presented in Table 3. Compared to SCM, which showed a reduction at 6 months (RR = 0.911, 95% CI: 0.744–1.116, p = 0.364) but achieved statistical significance at 12 months (RR = 0.767, 95% CI: 0.611–0.963, p = 0.022), MBT demonstrated more rapid and consistent improvement throughout the 12-month period.

Suicidal risk

Suicidal risk significantly decreased in the MBT group at 12 months (RR = 0.767, 95% CI: 0.591–0.995, p = 0.046), whereas SCM showed a reduction from 6 months (RR = 0.597, 95% CI: 0.427–0.835, p = 0.003), which remained significant at 12 months (RR = 0.646, 95% CI: 0.490–0.851, p = 0.002) (Table 3). Comparison of group effects, as shown in Table 4, indicated no significant difference (χ² = 2.108, p = 0.147), suggesting that both MBT and SCM effectively reduced suicidal risk without a distinct advantage for either treatment.

Mentalization difficulties

Marked improvements in mentalization difficulties were observed over time (χ² = 17.778, p < 0.001) (Table 4). The MBT group demonstrated a robust reduction (RR = 0.809, 95% CI: 0.716–0.914, p = 0.001), while the SCM group exhibited a comparable trend (RR = 0.859, 95% CI: 0.750–0.982, p = 0.026), as presented in Table 3.

Examining specific aspects of mentalization difficulties, significant time effects were found for lack of emotional awareness (χ² = 16.889, p < 0.001), lack of emotional expression (χ² = 6.195, p = 0.045), psychic equivalence (χ² = 9.916, p = 0.007), lack of mentalizing others (χ² = 16.901, p < 0.001), and lack of mentalizing situations (χ² = 8.073, p = 0.018) (Table 4). Across all participants, these aspects showed consistent reductions over time, with the most pronounced effects observed at 12 months.

Conversely, no significant time effect was observed for hasty and incomplete mentalizing (χ² = 0.058, p = 0.972) (Table 4). However, a significant group effect was detected (RR = 1.385, 95% CI: 1.004–1.911, p = 0.047), indicating a difference between MBT and SCM in the trajectory of this aspect, with the MBT group exhibiting a higher tendency for hasty and incomplete mentalizing (Table 4).

At the 12-month follow-up, MBT participants exhibited significant improvements in lack of emotional awareness (RR = 0.730, 95% CI: 0.608–0.877, p = 0.001), lack of emotional expression (RR = 0.804, 95% CI: 0.668–0.969, p = 0.022), lack of mentalizing others (RR = 0.671, 95% CI: 0.476–0.944, p = 0.022), and total score (R = 0.809, 95% CI: 0.716–0.914, p = 0.001) (Table 3). SCM participants showed notable improvements at 12 months, particularly in terms of psychic equivalence (RR = 0.822; 95% CI: 0.710–0.953; p = 0.009), lack of mentalizing others (RR = 0.524, 95% CI: 0.375–0.731, p < 0.001), and lack of mentalizing situations (RR = 0.569, 95% CI: 0.362–0.895, p = 0.015) (Table 3).

Attachment characteristics

Table 4 showed a significant time effect for attachment anxiety scores (χ² = 5.744, p = 0.017), indicating a reduction over time across both groups. However, Table 3 shows the distinct within-group patterns. The MBT group exhibited a significant decrease in attachment anxiety scores at 12 months (RR = 0.929, 95% CI: 0.864–0.998, p = 0.045), indicating a sustained, long-term improvement. In contrast, the SCM group showed a significant reduction only at 6 months (RR = 0.923; 95% CI: 0.864–0.986; p = 0.021), with no sustained effect at 12 months. These results highlight that while both treatments reduced attachment anxiety over time, only the MBT group demonstrated a long-term downward trend in attachment anxiety.

For attachment avoidance scores, both Tables 3 and 4 indicated no significant time effect (χ² = 2.272, p = 0.132), suggesting that neither MBT nor SCM produced meaningful changes in attachment avoidance throughout the 12-month follow-up period.

BPD symptom severity

Both MBT and SCM groups exhibited notable reductions in overall BPD symptom severity (χ² = 18.653, p < 0.001) (Table 4). The MBT group demonstrated a persistent decline, with significant reductions observed at 6 months (RR = 0.873; 95% CI: 0.810–0.940; p < 0.001) that were sustained through 12 months (RR = 0.870; 95% CI: 0.795–0.953; p = 0.003). In contrast, the SCM group achieved a significant reduction only at 6 months (RR = 0.930, 95% CI: 0.873–0.991, p = 0.031), but this improvement was not significant by 12 months (RR = 0.886, 95% CI: 0.779–1.009, p = 0.067). This pattern underscores the more consistent and durable symptom reduction achieved by MBT over the study period, which is consistent with the results presented in Tables 3 and 4.

Resilience

Table 4 presents a clear time effect for resilience scores, highlighting improvements in both the MBT and SCM groups, consistent with the trends depicted in Fig. 2C. The MBT group achieved significant gains at 6 months (RR = 1.091, 95% CI: 1.042–1.141, p < 0.001) and sustained improvements at 12 months (RR = 1.132, 95% CI: 1.064–1.205; p < 0.001) (Table 3). The SCM group showed significant improvement at 6 months (RR = 1.096, 95% CI: 1.029–1.168; p = 0.004), but the 12-month improvement (RR = 1.084, 95% CI: 0.994–1.182, p = 0.067) failed to reach statistical significance (Table 3). Despite no significant group effect (χ² = 0.671, p = 0.413) (Table 4), the MBT group exhibited a clearer upward trajectory, underscoring a more distinct and sustained long-term improvement.

Discussion

In the present study, we investigated the effectiveness of MBT in individuals with BPD over a 12-month RCT, using SCM as an active comparator. Although the present study did not find significant differences between the MBT and SCM groups, both treatments contributed to meaningful symptom reduction and functional improvement in Korean patients with BPD. Specifically, both MBT and SCM significantly reduced suicide attempts, alleviated depressive symptoms, and decreased mentalization difficulties. However, no statistically significant differences were observed between the two treatment modalities in terms of reducing suicide attempts or hospital dependency, suggesting that both interventions exerted comparable effects.

This finding is consistent with previous research demonstrating the efficacy of both MBT and SCM in reducing suicidality and reliance on inpatient care among individuals with BPD [16]. A plausible interpretation is that both interventions share several active ingredients, including regular and structured clinical contact, active monitoring of suicidality, and the availability of crisis-focused interventions when needed. In addition, both treatments were delivered in a relatively high-intensity outpatient format that combined weekly individual and group sessions, which may have further amplified the effects of these shared elements. Such a level of therapeutic intensity and continuity of care may be sufficient to produce meaningful improvements in acute outcomes (e.g., suicide attempts and hospitalization), even when the theoretical targets of treatment differ. In addition to these shared therapeutic factors, it should be noted that participants in the SCM group exhibited significantly higher rates of psychotropic medication use at baseline. Although medication use was statistically controlled for in all longitudinal analyses, the higher baseline medication rates in SCM may suggest that this group had more severe acute symptoms requiring pharmacological stabilization. This clinical profile difference, rather than medication effects per se, might partly explain the earlier symptom reduction trajectory in SCM. Accordingly, the interpretation of between-group differences should be made with caution.

Although two deaths occurred in the MBT group, both events appeared to be associated with circumstances that were difficult to anticipate or prevent within the context of the study. One participant died by suicide following acute external psychosocial stressors, including severe emotional and verbal conflict within the family environment shortly before the event. The second death occurred within one month after treatment consent and before sufficient therapeutic engagement or longitudinal monitoring could be established. Given the limited clinical exposure preceding both events, these cases should not be interpreted as evidence of treatment-related adverse effects. Nevertheless, they underscore the exceptionally high-risk nature of BPD populations and reinforce the need for intensive risk assessment and monitoring, particularly during periods of acute psychosocial stress and during the early phase of treatment engagement.

Despite these overall similarities in primary outcomes, important differential patterns emerged in secondary outcomes and their trajectories over time. In terms of depressive symptomatology, both MBT and SCM significantly reduced depressive symptoms over time, supporting prior research on the effectiveness of structured BPD treatment in alleviating mood symptoms [16]. However, the MBT group demonstrated a more consistent and sustained improvement in depressive symptoms, whereas the SCM group only exhibited a significant reduction at 12 months but not at 6 months. This pattern suggests that SCM’s early focus on crisis stabilization and problem-solving may have initially prioritized the management of acute stressors rather than targeting deeper emotional regulation deficits, which could explain the delayed reduction in depressive symptoms. Over time, as the patients in the SCM group experienced structured therapeutic support and problem-focused strategies, a more gradual improvement in mood symptoms emerged, which became significant after 12 months. In contrast, MBT, which targets mentalization and emotional regulation from an early stage, likely facilitated a steadier trajectory of depressive symptom improvement throughout the intervention [16, 24]. Similarly, suicide risk decreased in both groups, with early reductions in SCM at 6 months and gradual yet sustained reductions in MBT at 12 months. In a Korean study predicting self-harm behavior in a clinical population, rumination was the best predictor of lifetime non-suicidal self-injury [29]. These findings suggest that while both treatments effectively reduce self-harm behaviors and depressive symptoms in the Korean clinical population, their mechanisms may differ, warranting further research on the long-term impact of each approach.

Regarding mentalization difficulties, both MBT and SCM led to overall improvement. A more detailed examination of the mentalization subdomains reveals that MBT demonstrated stronger effects in improving emotional awareness, emotional expression, and mentalizing others, consistent with its theoretical framework emphasizing the role of mentalization in self-regulation and interpersonal understanding [12, 16, 30]. Our findings have several important cultural implications. In Korean society, where emotional restraint and indirect communication are often emphasized, difficulties in emotional awareness and expression may be more pronounced. Therefore, therapies explicitly targeting emotional awareness, such as MBT, may be particularly beneficial for patients struggling with internalized emotional difficulties.

In contrast, SCM showed greater improvement in psychic equivalence. Psychic equivalence refers to a rigid cognitive state in which thoughts and emotions are perceived as absolute reality, limiting mental flexibility and adaptive perspective taking. Unlike MBT, which focuses on the reflective exploration of mental states [12], SCM may encourage patients to actively restructure their thoughts through structured interventions and practical guidance from therapists. Furthermore, in Korean culture, where social conformity and external validation play significant roles in shaping individual experiences, structured guidance from therapists and a clear problem-solving framework may be more effective in modifying rigid thought patterns than an exploratory approach like MBT.

Interestingly, a significant group difference was observed in hasty and incomplete mentalization, with the MBT group exhibiting relatively higher levels of this tendency over the treatment period compared to the SCM group. One possible explanation for this pattern is that MBT actively encourages patients to explore their own mental states, interpersonal meanings, and subjective experiences [10, 12, 13]. This therapeutic emphasis may temporarily heighten patients’ awareness of rapid, superficial, or impulsive mentalizing tendencies, or even increase their explicit expression on self-report measures, before more stable reflective capacity is consolidated. In contrast, SCM places greater emphasis on pragmatic problem-solving, psychoeducation, and behavioral stabilization [16, 23], which may reduce opportunities for explicit exploration of mentalizing processes and thereby result in lower observed levels of hasty and incomplete mentalization.

Despite this between-group difference, no significant time effect was observed for hasty or incomplete mentalization in either treatment condition. Hasty and incomplete mentalization refers to impulsive, premature, and superficial interpretations of mental states, leading to misunderstandings in interpersonal interactions. Unlike other mentalization deficits that may be more directly influenced by structured reflection or guided problem-solving, hasty and incomplete mentalizing may be a more deeply ingrained cognitive habit, making it more resistant to short-term intervention. This resistance to change may be particularly pronounced in Korean cultural contexts, where a specific social cognitive strategy known as nunchi is deeply embedded in everyday interpersonal functioning. Consistent with this interpretation, preliminary evidence from Korean samples suggests that hasty and incomplete mentalization may not be specific to BPD pathology but may instead reflect a broader cultural cognitive pattern [21]. Nunchi refers to the ability to quickly perceive and accurately read others’ mental states, emotions, and social cues, and to act appropriately in response [31–33]. It comprises two interrelated components: rapid perception of others’ internal states and context-appropriate behavioral responses [31]. In Korean society, where high-context communication and collectivist values predominate, nunchi functions as a culturally adaptive and highly valued cognitive skill that enables individuals to navigate complex social hierarchies, maintain group harmony, and respond appropriately to implicit social expectations [31, 32]. Empirical research demonstrates that nunchi is positively associated with psychological well-being, self-esteem, and relationship quality in Korean populations [32, 33], underscoring its adaptive function within this cultural context. However, the very cognitive processes that make nunchi culturally adaptive—rapid, automatic, context-dependent inference of mental states based on minimal cues—may overlap substantially with what Western mentalization frameworks characterize as “hasty” or “incomplete” mentalizing [32]. Where rapid decision-making, social conformity, and appropriate responsiveness to hierarchical social cues are highly valued [18], individuals develop habitual cognitive shortcuts to quickly assess social situations and respond appropriately. These automatic, culturally reinforced cognitive patterns may make it particularly challenging for individuals to slow down and engage in deliberate, reflective mentalization, even in therapeutic settings [19, 20]. While Western conceptualizations of mentalization emphasize explicit, verbalized reflection on internal mental states, Korean cultural contexts may favor more implicit, context-dependent understanding of others’ intentions within established social roles [20]. The cultural emphasis on emotional restraint and indirect communication further complicates this, as Korean patients with BPD may face additional barriers to identifying and articulating internal emotional states beyond the core deficits associated with the disorder itself [18–20]. Thus, the resistance to change in hasty and incomplete mentalization observed in this trial may reflect not only individual psychopathology but also the deep cultural embedding of nunchi-based cognitive strategies that are normative, adaptive, and socially reinforced in Korean society.

Given these culturally specific challenges, future MBT implementations in Korean settings should consider several targeted adaptations. First, therapists could explicitly acknowledge the tension between cultural expectations for rapid social responsiveness and therapeutic goals of reflective pause, framing deliberate mentalization as an adaptive skill for managing emotionally charged relationships rather than cultural deviance [14, 17]. Second, group MBT could place greater emphasis on elaborating and working in depth with culturally salient interpersonal agendas as they naturally emerge within the group—such as difficulties related to hierarchical workplace dynamics or family obligations embedded in collectivistic values—thereby providing repeated opportunities to practice mentalization [13, 14]. Third, extending treatment duration beyond 12 months or incorporating booster sessions may be necessary for this particularly resistant domain, given that hasty mentalization appears to be a deeply ingrained pattern reinforced by cultural norms [14]. Finally, integrating mindfulness-based techniques congruent with Korean Buddhist traditions may offer a culturally acceptable pathway to slowing down automatic cognitive processes [13]. These proposed adaptations maintain fidelity to MBT’s core principles while working within Korean cultural frameworks, and future research should systematically test whether they enhance treatment effectiveness for hasty and incomplete mentalization in Korean and other Asian populations [13, 14].

The results for adult attachment types showed significant time effects for anxiety reduction in both groups, although only MBT demonstrated a sustained downward trend at 12 months. The lack of change in avoidance scores aligns with prior research, suggesting that attachment avoidance in patients with BPD may be more resistant to change and may require longer treatment durations or more intensive relational interventions [11].

Both MBT and SCM reduced BPD symptom severity, as measured by the PAI-BOR, which is consistent with previous research indicating that structured interventions can mitigate the core psychopathology of BPD [16]. However, although MBT was associated with more consistent and sustained reductions, SCM facilitated significant symptom improvement. The more sustained improvement observed in the MBT may reflect its explicit focus on enhancing mentalization, thereby addressing the underlying mechanisms of affective instability, impulsivity, and interpersonal dysfunction that characterize BPD.

In addition to symptom reduction, both treatments were associated with a gain in psychological resilience. Resilience, which reflects an individual’s ability to adapt to stress and recover from adversity, is a crucial factor in the long-term clinical outcomes of individuals with BPD [34]. However, the trajectory of the resilience gain differed between the two treatments. The MBT participants demonstrated a more consistent and sustained increase in resilience, suggesting that enhanced mentalization and reflective functioning may contribute to long-term emotional adaptability. SCM participants, on the other hand, showed initial resilience improvements that plateaued over time, indicating that while structured support fosters early stability, it may not promote continued psychological growth without additional interventions.

In summary, this study was designed with the expectation that MBT would demonstrate advantages over SCM, particularly in domains related to mentalization and longer-term adaptation. Contrary to this expectation, both treatments showed comparable effects on primary outcomes such as suicide attempts and hospitalization. This finding may reflect the high-intensity, structured outpatient setting common to both interventions, shared therapeutic elements relevant to acute risk reduction, and the absence of formal treatment fidelity assessment. Future studies should incorporate fidelity monitoring and longer follow-up periods to clarify treatment differentiation and long-term effects. Given that RCTs of evidence-based treatments for BPD have not consistently identified a single superior modality [35], there is an increasing consensus that treatment selection should be based on patients’ core difficulties, treatment preferences, and the availability of appropriate treatment options [5].

The comparable reduction in suicide attempts suggests that SCM, despite its relatively straightforward structure, may be as effective as MBT in crisis stabilization and self-harm prevention, making it a viable treatment option in certain clinical settings. MBT presents distinct practical advantages, including less intensive training requirements for therapists [5, 16] and a clearly manualized framework that standardizes therapeutic stance and interventions, thereby enhancing consistency and treatment fidelity across providers. Notably, the present study found that MBT was associated with more sustained improvements in secondary outcomes such as resilience, mentalization, and depressive symptoms, suggesting that MBT may offer greater long-term benefits for patients’ personality functioning and adaptive capacity. In Korea, where clinicians with specialized training in personality disorder treatment remain scarce and access to evidence-based training programs is limited, these features of MBT are particularly advantageous. Although this study did not demonstrate significant superiority of MBT over SCM, its practical benefits underscore its potential as an effective and scalable intervention in the Korean mental healthcare system.

Limitations

This study should be interpreted as a pilot randomized controlled trial evaluating the preliminary effectiveness and feasibility of MBT and SCM in a South Korean outpatient setting. As such, it has several important limitations. First, the relatively small sample size (N = 64) and high dropout rate (48%) represent significant limitations that may have affected the statistical power to detect between-group differences and limited the generalizability of the findings. While elevated dropout is common in BPD populations, the rate observed in this study was notably higher than the 22–40% reported in previous research [36–38]. Although individual characteristics commonly associated with BPD—such as emotional instability and ambivalence toward treatment—may increase vulnerability to dropout, several treatment- and context-related factors may have contributed to this higher attrition. Both MBT and SCM were delivered in a relatively intensive outpatient format requiring weekly individual and group sessions over a 12-month period, which may have imposed substantial demands in terms of time commitment and emotional burden. In addition, because the interventions were provided as part of a research protocol at no financial cost to participants, the absence of direct financial investment may have inadvertently reduced perceived treatment commitment or accountability for some individuals, potentially increasing the likelihood of premature discontinuation. At the same time, although therapists received protocol-based training prior to the study, MBT was newly implemented in these clinical settings, and limited prior experience with intensive, manualized psychotherapies may have influenced early therapeutic engagement and continuity. Consequently, the high attrition rate not only reduced the effective sample size for longitudinal analyses but also raises questions about treatment acceptability and feasibility in Korean clinical settings. Large-scale studies with adequate sample sizes, strategies to improve retention, and culturally adapted engagement strategies are needed to confirm these results. Second, the 12-month follow-up period may not have been sufficient to capture long-term treatment effects, particularly for MBT, which is theorized to exert its strongest impact over extended durations. Future studies should incorporate 18-month or longer follow-up assessments to evaluate the durability of treatment outcomes. Third, gender identity was not systematically assessed, and only biological sex was recorded based on medical records. This reflects current clinical documentation practices in South Korea, where standardized assessment and routine recording of gender diversity in psychiatric settings remain limited. Nevertheless, emerging evidence suggests that gender diversity is clinically relevant in the context of BPD and suicidality, particularly given elevated rates of emotional dysregulation and self-harm reported in gender-diverse populations [39]. Future research conducted in Korean clinical settings should incorporate more inclusive and standardized measures of gender identity as these practices become more widely implemented. Finally, formal ongoing MBT-specific supervision was not available during the trial, and treatment adherence and therapist fidelity were not systematically assessed, which may have introduced potential variability in treatment delivery and implementation. Rigorous adherence monitoring and fidelity checks will be essential in subsequent research to examine their impact on clinical outcomes.

Conclusions

This pilot study provides preliminary evidence for the effectiveness of MBT and SCM in reducing suicide attempts, depressive symptoms, and hospital utilization in individuals with BPD. While no significant differences were found between the two treatments in terms of primary outcomes, MBT demonstrated a more consistent and sustained improvement in depressive symptoms, mentalization capacity, and psychological resilience. These findings suggest that MBT may offer long-term advantages in fostering emotional regulation and adaptability, whereas SCM provides a structured and effective approach to crisis management and symptom stabilization.

Additionally, the absence of significant differences in some outcomes may, in part, be explained by the potential cross-utilization of MBT strategies by therapists in SCM. Future studies should monitor treatment fidelity to distinguish between the distinct contributions of each modality.

Given the cultural considerations that affect mentalization and cognitive styles, future research should explore culturally tailored interventions to address specific mentalization deficits and enhance treatment efficacy. Additionally, long-term follow-up studies are needed to determine whether MBT’s sustained benefits translate into lower relapse rates and greater functional improvements over extended periods.

Abbreviations

BPD

Borderline personality disorder

RCT

Randomized controlled trials

MBT

Mentalization-based treatment

SCM

Structured clinical management

S-TAU

Specialist treatment-as-usual

DSM-5

The Diagnostic and Statistical Manual of Mental Disorders, 5th edition

PAI-BOR

Personality Assessment Inventory-Borderline Features Scale

SCID-5-PD

Structural Clinical Interview for the Diagnostic and Statistical Manual of Mental Disorders-Personality Disorders

C-SSRS

Columbia-Suicide Severity Rating Scale

IRB

Institutional Review Board

PROVE

Protective and Vulnerable Factors Battery Test

ITT

Intention-to-treat

GEE

Generalized estimating equation

GLMM

Generalized mixed-effects model

MCAR

Missing completely at random

CI

Confidence Interval

RR

Relative risk

IRR

Incidence rate ratio

Author contributions

JHS, BHK, and HKS conceived the project and main ideas, and prepared proof of the outline. HKS, SWC, HP, UJO, JYL, and BC examined the participants, and acquired and organised the data. HKS and WJC drafted the manuscript and prepared Figs. 1–2. WJC and HKS performed the statistical analyses and designed the figures. HKS, WJC, YSC, JHS, and BHK contributed to the interpretation of the results and worked on the manuscript. All authors discussed the results and reviewed the manuscript. JHS supervised the project.

Funding

This research was supported by a grant of Patient-Centered Clinical Research Coordinating Center (PACEN) funded by the Ministry of Health & Welfare, Republic of Korea (grant number: RS-2020-KH095597)

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethical approval and consent to participate

The study was approved by the Institutional Review Boards of Gangnam Severance Hospital (IRB No. 3–2021–0095) and Yeungnam University Hospital (IRB approval number: 2021–02-046). Informed consent was obtained from all participants.

Consent for publication

The participants provided written consent to have their data published.

Competing interests

The authors declare no competing interests.