Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Jul 4;82(10):1490–1497. doi: 10.1002/jclp.70176

Personality Traits in Borderline Personality Disorder: A Cluster Analysis Based on the Millon Test Scores

Jose Manuel Lopez‐Villatoro 1,2, Marina Diaz‐Marsa 1,2,3, Ivan Perez‐Diez 3,4, Alejandra Galvez‐Merlin 1,2,✉, Clara Prittwitz 1, Jose Luis Carrasco 1,2,3
PMCID: PMC13544488  PMID: 42400289

ABSTRACT

Introduction

The complexity of the treatment of borderline personality disorder (BPD) is largely due to its heterogeneity. The aim of this study is to identify empirically derived patterns of co‐occurring personality traits in individuals with primary BPD and to examine their clinical correlates.

Methods

Millon's personality traits (MCMI‐IV) of 97 BPD patients were assessed and a cluster analysis of these traits was performed. Patients were evaluated on different clinical dimensions of BPD and compared between the different clusters found.

Results

Three trait‐pattern clusters were obtained. Pattern 1 showed higher scores for borderline, histrionic, narcissistic and antisocial traits. Pattern 2 had higher scores for borderline, dependent and avoidant traits. Pattern 3 had lower levels of all the traits. Patients in patterns 1 and 2 showed greater scores on anxious‐depressive scales. Pattern 1 also showed greater scores for anger expression. Pattern 3 patients were characterized for receiving more pharmacological treatment.

Conclusions

Findings provide evidence for three patterns of co‐occurring traits within BPD, consistent with dimensional conceptualizations and longstanding categorical overlap. These patterns align with externalizing‐leaning, internalizing‐leaning, and attenuated configurations and may help generate hypotheses for early characterization and treatment personalization.

Keywords: assesment, borderline personality disorder, heterogeneity, personality traits, treatment

1. Introduction

Borderline personality disorder (BPD) is a severe and persistent mental disorder characterized by affective instability and impulsive behaviors, affecting self‐image, interpersonal relationships, affect, and behavior (American Psychiatric Association 2013).

Current studies show that BPD is a treatable disorder, and multiple psychotherapies have been shown to be effective (Stoffers‐Winterling et al. 2022; Zanarini et al. 2003). However, a significant subgroup shows a partial or poor response (Woodbridge et al. 2022). In this context, the marked heterogeneity of BPD, both in symptom presentation and concurrent personality traits, remains a major obstacle to clinical decision‐making and treatment personalization (Videler et al. 2019; Winsper et al. 2015). This heterogeneity is mainly due to the interaction of genetic, environmental, and neurobiological factors (Paris 2018).

A major source of this heterogeneity is the overlap among categorical personality disorder (PD) diagnoses. Categorical nosologies (e.g., DSM‐IV/DSM‐5 Section II clusters) allow patients to meet criteria for multiple PDs, generating extensive comorbidity and within‐diagnosis variability. Decades of research on clinical practice have documented these limitations and advocated for dimensional trait approaches that conceptualize personality pathology in terms of continuous trait domains rather than mutually exclusive categories (Clark 2007; First et al. 2002; Livesley 2003; Trull and Durrett 2005; Widiger and Trull 2007). In line with this shift, both the DSM‐5 Alternative Model for Personality Disorders (AMPD) and the ICD‐11 framework describe PDs through a combination of severity and maladaptive trait domains (e.g., Negative Affectivity, Detachment, Antagonism/Dissociality, Disinhibition, Anankastia, Psychoticism), with ICD‐11 including an optional borderline pattern specifier (American Psychiatric Association 2013; World Health Organization 2019).

Within these dimensional perspectives, one promising strategy to parse BPD heterogeneity is to examine empirically derived patterns of co‐occurring traits measured with widely used instruments. Millon's model (Millon et al. 2018), operationalized by the Millon Multiphasic Personality Inventory (MCMI‐IV), assesses broad personality trait constellations spanning expressive/interpersonal styles, cognitive‐affective functioning, and personality organization. Although historically framed in categorical terms, the MCMI‐IV yields continuous base‐rate scores that can be analyzed dimensionally to uncover naturally occurring configurations of traits in clinical samples. Mapping such configurations in patients with a primary BPD diagnosis can illuminate whether commonly observed clinical “types” reflect trait mixtures consistent with dimensional models rather than discrete categorical subtypes.

In this line, the aim of this work was to assess Millon's personality traits (using the MCMI‐IV) in a clinical cohort of individuals with primary BPD and to apply cluster analysis to identify patterns of co‐occurring traits. Furthermore, to assess their clinical relevance, we compared the resulting groups on variables that are fundamental to the affective and behavioral profile of BPD: affective instability (anxious‐depressive symptoms), feelings of emptiness, aggression, impulsivity, clinical severity, overall functioning, duration of the disorder, self‐harm and suicide attempts, and psychiatric medication use. This design allows us to examine whether trait configurations are associated with significant differences in the clinical expression of BPD.

2. Materials and Methods

2.1. Participants

The sample consisted of 97 participants recruited from a specialized Personality Disorders Day Hospital unit of the Hospital Clínico San Carlos of Madrid (Spain). They were diagnosed with BPD according to DSM‐5 criteria (American Psychiatric Association 2013) through an initial interview with the head of the personality disorders unit (criterion for admission to the day hospital). In line with our center's clinical practice, Axis‐II diagnoses are documented using the SCID‐II (as explained in the section on instruments), but the primary diagnosis of BPD is confirmed when borderline symptoms are predominant and represent the main source of distress and functional impairment in patients. They presented moderate‐severe severity (CGI (clinical global impression) > 4) and moderate dysfunctionality (GAF (Global Assessment of Functioning Scale) < 65).

Patients with the following criteria were excluded from the study: (1) having a neurological or medical illness that could affect brain functions; (2) having an IQ < 85; (3) having had schizophrenia, schizophreniform disorder or bipolar disorder in their lifetime; (4) having a major depressive episode or a substance use disorder at the time of the study that could affect neuropsychological performance.

All patients received detailed information about the study and signed written informed consent prior to their participation in the research. The clinical research study was approved by the Clinical Research Ethics Committee of the Hospital Clínico San Carlos of Madrid (Spain).

2.2. Instruments

The collection of clinical variables was performed by experienced psychiatrists and psychologists at the beginning of the study. All patients were interviewed using the Structural Interview for Personality Disorders (SCID‐II; First et al. 1997). Severity was measured with the Clinical Global Scale (CGI; Perez et al. 2007) and chronicity was assessed with the Global Assessment of Functioning Scale (GAF; Hall 1995) and years of illness duration.

Personality traits were assessed through the Millon Multiphasic Personality Inventory (MCMI‐IV; Millon et al. 2018). The MCMI‐IV was scored using standard procedures as recommended by the test manual (Millon et al. 2018), yielding Base Rate (BR) scores for each personality scale. As reported in the MCMI‐IV manual, internal consistency (Cronbach's alpha) median values are α = 0.84 for the personality pattern scales and α = 0.83 for the clinical syndrome scales (Millon et al. 2018).

Other clinical variables, such as anxious and depressive symptoms, were assessed with the Hamilton Anxiety Scale (HARS; Hamilton 1959) and the Montgomery‐Asberg Depression Scale (MADRS; Montgomery and Åsberg 1979), respectively. Impulsivity was measured with the Barrat Impulsivity Scale (BIS‐11; Patton et al. 1995) and aggressive behavior was assessed through the State‐Trait Anger Expression Inventory (STAXI‐2, Spielberger 1999). The chronic feeling of emptiness was measured with the Feeling of Emptiness Questionnaire (CSV; Palomares et al. 2020), and the assessment of suicide risk was performed through an ad hoc interview, where the number of suicide attempts and the frequency of self‐injury without suicidal intent were recorded, with special emphasis on the difference between the two types of behavior. All attempts showed a desire for physical self‐elimination, as distinct from the impulse to self‐injure as a way to reduce anxiety, anger, or pain.

2.3. Statistical Analysis

Sociodemographic, clinical, and functional data of the total BDP patients were described using means and standard deviations for quantitative scores, and proportions of cases for qualitative variables. Secondly, a k‐means cluster analysis of the following MCMI‐IV scales was performed: Aggressive/Sadistic, Antisocial, Avoidant, Borderline, Compulsive, Dependent, Histrionic, Paranoid, Schizoid, Schizotypal, and Self‐defeating. The k‐means analysis allows individuals to be grouped into clusters according to the given variables. To select the number of clusters, we use the NbClust package, which uses 30 indexes to carry out the selection of the optimal number of clusters (Charrad et al. 2014). Next, the quantitative clinical and socio‐demographic variables of age, years of duration of the disorder, anxiety, depression, feelings of emptiness, aggression, impulsivity, number of suicide attempts, severity, and functioning were then compared between the three cluster groups through an analysis of variance (ANOVA). Finally, the qualitative variables of sex, frequency of self‐harm, psychiatric hospitalizations, and pharmacological treatment were compared between the three cluster groups using the Chi‐square test. Effect sizes were estimated using Eta‐squared for ANOVA and Cramer's V for Chi‐square. Results were considered significant at below the significance level of less than 0.05. The analyzes were performed using R Studio Software version 4.0.2.

3. Results

Table 1 shows the sociodemographic, clinical, and functional variables.

Table 1.

Sociodemographic, clinical, and functional data of the total BDP patients.

BPD Patients
Sex (% ref. women) 83.50
Age (in years) (M ± SD) (30.88 ± 10.04)
Qualification level (% higher than secondary education) 53.97
Marital status (% Married) 32.70
(% Single/separated) 67.30
Working status
Unemployed (%) 74.70
Student (%) 15.30
Employed (%) 10.20
Medication (%yes) 92.9
Antidepressant prescription (% yes) 79.20
Antipsychotic prescription (% yes) 59.40
Antiepileptic prescription (% yes) 57.30
Benzodiazepine prescription (% yes) 62.40
CGI (M ± SD) 5.20 (0.80)
GAF (M ± SD) 60.23 (6.51)
Open in a new tab

Abbreviations: DS, standard= standard deviation; M, mean; N, sample size.

The mean scores obtained in the different personality traits of all BPD patients are shown in Table 2.

Table 2.

Mean and standard deviation in Million personalities scales for the whole sample.

M DT
Aggressive/Sadistic 70.81 26.14
Antisocial 72.15 29.71
Avoidant 81,94 21,381
Borderline 88,321 17,8485
Compulsive 54,98 66,55
Dependent 69,71 81,98
Histrionic 45,63 59,41
Paranoid 75,29 26,565
Schizoid 77,57 25,903
Schizotypal 89,026 26,3547
Self‐Defeatting 81,90 17,380
Open in a new tab

Abbreviations: Agressive/Sadistic = aggressive‐sadistic personality traits; antisocial = antisocial personality traits; avoidant = avoidant personality traits; borderline = borderline personality traits; compulsive = obsessive‐compulsive personality traits; dependent = dependent personality traits; DT = standard deviation; histrionic = histrionic personality traits; M = mean; paranoid = paranoid personality traits; schizoid = schizoid personality traits; schizotypal = schizotypal personality traits; self‐defeatting = self‐defeating personality traits.

To improve transparency regarding sample composition, Table 3 presents the Axis II comorbidities identified through the SCID‐II interview. As is common in personality disorders, more than two‐thirds of participants presented significant traits of at least one additional PD.

Table 3.

Axis II personality disorder comorbidities in the total BPD sample using MCMI‐IV.

Personality Disorder (SCID‐II) % (n)
Avoidant 72.2 (70)
Obsessive‐compulsive 53.7 (52)
Depressive 53.7 (52)
Dependent 44.4 (43)
Paranoid 38.9 (38)
Schizotypal 35.2 (34)
Narcissistic 34.0 (33)
Passive‐aggressive/Negativistic 27.8 (27)
Schizoid 27.8 (27)
Antisocial 25.9 (25)
Histrionic 16.7 (16)
Open in a new tab

Note: Percentages reflect the proportion of BPD patients meeting criteria for at least significant traits of each additional PD according to SCID‐II. Multiple comorbid diagnoses per patient are possible.

A cluster analysis was performed based on the personality traits of the MCMI‐IV. The results showed three clusters explaining 38.6% of the variance of the factors. Cluster 1 consisted of 31 of the patients, cluster 2 was observed in 25 of the patients, and cluster 3 in 37 of the patients (Figure 1).

Figure 1.

Figure 1

Open in a new tab

Millon's personality traits in clusters of BPD patients. n: sample size; aggressive‐sadistic = aggressive‐sadistic personality traits; antisocial = antisocial personality traits; avoidant = avoidant personality traits; borderline = borderline personality traits; compulsive = obsessive‐compulsive personality traits; dependent = dependent personality traits; histrionic = histrionic personality traits; paranoid = paranoid personality traits; sadistic = sadistic personality traits; schizoid = schizoid personality traits; schizotypal = schizotypal personality traits; self‐defeating = self‐defeating personality traits.

Patients in cluster 1 were characterized by significantly higher levels of histrionic (p < 0.001), narcissistic (p < 0.001), antisocial (p < 0.001), aggressive‐sadistic (p < 0.001), and schizotypal (p < 0.001) personality traits. These patients also showed, compared with the other two groups, significantly lower values of paranoid (p < 0.001) and obsessive‐compulsive (p < 0.001) personality traits.

Patients in cluster 2 showed significantly higher levels, compared with patients in clusters 1 and 3, of dependent (p < 0.001), self‐defeating (p < 0.001), and avoidant (p < 0.001) personality traits.

Significantly lower levels of schizotypal (p < 0.001), schizoid (p < 0.001), paranoid (p < 0.001), self‐destructive (p < 0.001), and borderline (p < 0.001) personality traits were observed in patients in cluster 3 compared with patients in the other two clusters.

Next, differences in sociodemographic, clinical, and functionality variables were analyzed among the three cluster groups found (Table 4). Patients in clusters 1 and 2 showed significantly higher scores on the Hamilton Anxiety Scale (HARS) (F = 3.89, p < 0.024, η 2 = 0.08), on the Montgomery‐Asberg Depression Scale (MADRS) (F = 6.77 m, p < 0.002, η 2 = 0.13) and on the Feeling of Emptiness Questionnaire (CSV) (F = 8.79, p < 0.001, 0. 17), compared to patients in cluster 3. On the other hand, patients in cluster 1 had more years of illness duration (F = 3.82, p < 0.026, η 2 = 0.08) and higher scores on the State (F = 6.47, p < 0.002, η 2 = 0.13) and Trait subscales of the STAXI (F = 5.39, p < 0.006, η 2 = 0.11), relative to patients in clusters 2 and 3. Significant differences were also found between the 3 groups in the consumption of psychiatric medication, with patients in cluster 3 taking significantly more psychiatric medication (χ 2 = 8.75, p = 0.013, V = 0.3) than patients in the other two groups, although no differences were observed between the types of drugs evaluated. No significant differences were found in Barrat Impulsivity Scale (BIS‐11) scores, Clinical Global Scale (CGI), Global Functionality Scale (GAF), age, gender, number of self‐injuries, or number of suicide attempts.

Table 4.

Differences between clusters for sociodemographic, clinical, and functionality variables.

Variable Cluster 1 M (SD) Cluster 2 M (SD) Cluster 3 M (SD) F p‐value η2
Age 32.06 (8.78) 30.51 (9.96) 29.59 (12.64) 0.40 0.674 0.01
Years of disease duration 20.48 (12.64) 14.40 (8.38) 13.75 (11.05) 3.82 0.026 0.08
Anxiety 34.18 (12.61) 33.07 (13.74) 23.65 (14) 3.89 0.024 0.08
Depression 32 (12.21) 34.69 (11.51) 22.18 (12.78) 6.77 0.002 0.13
Feelings of emptiness 87.88 (25.52) 88.81 (20.57) 62.63 (19.66) 8.79 < 0.001 0.17
Aggressiveness‐state 31.65 (17.98) 21.86 (12.31) 17.63 (10.26) 6.47 0.002 0.13
Aggressiveness‐trait 30.39 (10.42) 24.14 (8.30) 23 (8.01) 5.39 0.006 0.11
Impulsivity 72.14 (16.67) 60.09 (15.93) 62.33 (20.25) 1.41 0.252 0.04
Number of suicide attempts 3 (3.1) 3.51 (2.85) 1.82 (1.59) 2.38 0.098 0.05
CGI 5.48 (0.85) 5 (0.78) 5.20 (0.68) 2.48 0.091 0.07
GAF 58.42 (8.51) 60.33 (5.40) 62.33 (5.3) 1.55 0.221 0.05
Cluster 1 (%) Cluster 2 (%) Cluster 3 (%) χ2 p‐value V
Sex Male 21.2 17 5.9 1.93 0.380 0.14
Female 78.8 83 94.1
Self‐harm No 44.1 37 62.5 4.38 0.625 0.15
Ocassional 17.6 21.7 12.5
Repeated 11.8 8.7 12.5
Very often 26.5 32.6 12.5
Pharmacological treatment Yes 82.4 97.9 100 8.75 0.013 0.3
Antidepressants 73.5 82.2 82.4 1.01 0.620 0.10
Antipsychotics 55.9 62.2 58.8 0.33 0.850 0.06
Antiepileptics 50 64.4 52.9 1.81 0.404 0.14
Benzodiazepines 60.6 59.9 56.3 0.53 0.766 0.08
Open in a new tab

Abbreviations: Aggressiveness‐state = STAXI state subscale score; Aggressiveness‐state = STAXI trait subscale score; Anxiety = Hamilton Anxiety Scale score; CGI = Clinical Global Impression; Depression = Montgomery‐Asberg Depression Scale score; F = coefficient of variances; Feeling of emptiness = Feeling of Emptiness Questionnaire scores; GAF = Global Assessment Functionally; Impulsivity = Barrat Impulsivity Scale score; M = mean; SD = standard deviation; V = Cramér's V (Chi‐square effect size); χ2 = Chi‐square; η2 = Eta square (ANOVA effect size).

4. Discussion

This study identified three empirically derived trait‐patterns of co‐occurring personality features among individuals with a primary diagnosis of BPD, based on MCMI‐IV scores obtained via k‐means clustering.

Specifically, trait‐pattern 1 showed significantly higher histrionic, narcissistic, antisocial, aggressive–sadistic, and schizotypal traits together with significantly lower paranoid and obsessive–compulsive traits. Likewise, trait‐pattern 2 showed significantly higher dependent, avoidant, and self‐defeating traits. Finally, trait‐pattern 3 displayed globally lower scores ‐including significantly lower schizotypal, schizoid, paranoid, self‐defeating, and borderline traits.

Our results are best interpreted within contemporary dimensional frameworks of personality pathology. In the DSM‐5 Alternative Model for Personality Disorders (AMPD), personality dysfunction (Criterion A) is qualified by maladaptive trait domains (Criterion B: Negative Affectivity, Detachment, Antagonism, Disinhibition, Psychoticism). Similarly, ICD‐11 conceptualizes personality disorder severity and specifies trait domains (Negative Affectivity, Detachment, Dissociality, Disinhibition, Anankastia), with an optional Borderline pattern specifier.

Against this backdrop, the pattern we labeled pattern 1 ‐marked by elevations in narcissistic, antisocial, histrionic, and aggressive‐sadistic traits‐ aligns with higher Antagonism/Dissociality and Disinhibition, consistent with externalizing liability and greater anger expression. Pattern 2 ‐characterized by avoidant, dependent, and self‐defeating features‐ maps into Negative Affectivity and Detachment (with potential Anankastia contributions), consistent with an internalizing/anxious‐withdrawn profile and our observed elevations in anxiety, depression, and emptiness. Pattern 3 showed globally lower trait elevations and comparatively reduced affective burden; it would be a group with a borderline structure closer to neuroticism, showing more emotionally stable and with less severity of pathology (Esguevillas et al. 2018).

Taken together and consistent with longstanding critiques of categorical PD models and diagnostic overlap (e.g., Clark 2007; Livesley 2003; Trull and Durrett 2005; Widiger and Trull 2007), these data provide evidence for three empirically derived patterns of co‐occurring traits among individuals with BPD, rather than discrete categorical subtypes. Future confirmatory studies should deploy direct AMPD/ICD‐11 trait assessments, preregistered mappings, and longitudinal designs to determine whether these patterns show differential prognostic or treatment relevance.

In order to characterize the clinical expression more concretely, we compared BPD‐relevant clinical and functional variables across the three trait‐patterns. Trait patterns 1 and 2 showed significantly higher levels of depression, anxiety, and feelings of emptiness compared to trait pattern 3. In addition, patients in group 1 exhibited a longer duration of illness, suggesting a more chronic course. This group also showed higher levels of anger, both state and trait, which may be associated with a greater tendency toward impulsive–aggressive behaviors. Finally, group 3 had a higher frequency of pharmacological treatment prescriptions, with no significant differences in the type of psychiatric medication used.

Identifying empirically derived trait‐patterns within BPD refines the phenotyping of a highly heterogeneous condition (Videler et al. 2019; Winsper et al. 2015). In the context of rising service demand (Chanen et al. 2017), early recognition of these trait configurations can support more efficient allocation of clinical resources and inform personalized, timely interventions. Rather than defining categorical subtypes, this approach characterizes co‐occurring trait constellations that may guide assessment focus and treatment planning for distinct clinical needs. The clinical implications of this trait‐based approach extend beyond phenotyping and align with a broader literature on the use of dimensional personality science in treatment planning. Harkness and Lilienfeld (1997) articulated four key roles that personality assessment can play in clinical care: informing where to focus change efforts, fostering realistic expectations of therapeutic gains, facilitating effective treatment matching, and enhancing self‐development. Bagby et al. (2016) expanded on this framework by reviewing evidence that dimensional trait profiles can guide both the selection of therapeutic targets and the anticipation of treatment barriers. Lengel et al. (2016) further argued that translational applications of personality science allow clinicians to move beyond symptomatic diagnosis toward interventions tailored to the underlying trait architecture of each patient and noted specifically that traits such as disinhibition and antagonism may predict treatment‐interfering behaviors such as poor homework compliance, missed appointments, and impulsive disengagement. These theoretical frameworks map well onto the patterns identified in the present study. Pattern 1, with elevated antagonism and disinhibition, corresponds most closely to the profile that Lengel et al. (2016) and Bagby et al. (2016) associate with the greatest treatment‐planning challenge: the clinician may need to address treatment motivation proactively, adjust expectations around engagement, and prioritize behavioral dyscontrol as a primary treatment target before working on deeper emotional processing. Pattern 2, with its high negative affectivity and detachment, aligns with a profile for whom the treatment focus may be fruitfully directed toward interpersonal sensitivity, fear of abandonment, and maladaptive self‐defeating schemas, consistent with approaches such as Schema Therapy or Mentalization‐Based Treatment. Pattern 3, with globally lower trait severity, may reflect a subgroup for whom standard evidence‐based protocols are more readily applicable with relatively fewer trait‐related treatment barriers. These distinctions are particularly relevant given that, despite the general treatability of BPD, a meaningful subset of patients shows partial or poor treatment response (Woodbridge et al. 2022), and the AMPD framework — whose clinical utility has been described in detail by Waugh (2019) and Waugh et al. (2019) — offers a structured approach to translating trait profiles into individualized case formulations and treatment goals. We acknowledge that these are hypothesis‐generating propositions: our cross‐sectional design does not permit causal inferences about treatment outcome, and prospective studies are needed to test whether the trait‐patterns identified here predict differential response to specific interventions.

Finally, this study has limitations. First, the modest sample size due to the complexity of the personality assessment constrains statistical power, limiting inferences about associations between trait‐patterns and specific clinical characteristics.

Following the sample limitations, patients were undergoing drug treatment during the study, which could influence the results. However, the withdrawal of these drugs has serious clinical consequences. To improve this limitation, medication in the patients studied was limited to that necessary to obtain a moderate behavioral stability that would allow their psychological treatment in the Day Hospital, but in all cases, we avoided studying patients who showed signs of sedation or slowing.

In addition, as already discussed, the sample was composed of patients with moderate‐severe severity (CGI (clinical global impression) > 4) and high dysfunctionality (GAF (Global Assessment of Functioning Scale) < 65). Therefore, the findings of our study cannot be generalized to all BPD patients, as our patients were shown to be severely affected by the disorder and thus do not represent the average BPD population.

Regarding the methodology, the Barratt Impulsivity Scale (BIS‐11), used to measure impulsivity, although a classic research instrument, may not comprehensively assess all dimensions of impulsivity. Future research should incorporate more recent multidimensional measures such as the UPPS Impulsive Behavior Scale (Lynam et al. 2006).

And finally, no formal correction for multiple comparisons was performed in the data analysis, which increases the risk of cumulative type I error; therefore, our findings should be interpreted with caution and considered provisional until replicated in larger samples.

5. Conclusions

This study provides evidence for three empirically derived patterns of co‐occurring personality traits among individuals with a primary diagnosis of BPD, based on MCMI‐IV scores, rather than for discrete categorical subtypes. Specifically, Pattern 1 showed higher Cluster B traits of the DSM‐5, with greater anger expression and longer illness duration; Pattern 2 was characterized by borderline traits along with others from Cluster C of DSM‐5 and greater negative affect (anxiety, depression, emptiness); and Pattern 3 displayed globally lower trait elevations and comparatively lower affective burden, alongside a higher proportion of patients receiving pharmacotherapy. Framed within contemporary dimensional models of personality pathology, these findings suggest that common clinical “types” in BPD may reflect trait constellations with distinct affective/behavioral profiles. Clinically, identifying such trait patterns may aid early clinical characterization and inform hypothesis‐generating directions for treatment personalization (e.g., targeting anger/behavioral dyscontrol in externalizing‐leaning configurations; addressing interpersonal sensitivity and negative affect in internalizing‐leaning configurations). These results suggest that studying personality traits in BPD patients could facilitate a more personalized treatment and be helpful for health and social services involved in the psychosocial approaches.

Ethics Statement

The clinical research study was approved by the Clinical Research Ethics Committee of the Hospital Clínico San Carlos, and it was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki).

Consent

All patients received detailed information about the study and signed written informed consent prior to their participation in the research.

Conflicts of Interest

The authors declare no conflicts of interest.

Acknowledgments

This work was supported by the PI20/01471 project, integrated in the Plan National de I + D + I, AES 2020–2023; funded by the ISCIII and co‐funded by the European Regional Development Fund (ERDF) “A way to make Europe”, a Complutense University of Madrid predoctoral grant (ref. FPU21/04434) and an Autónoma University of Madrid predoctoral grant (ref. FPU21/06359). The funding sources were involved neither in the design of the study, nor in the collection, analysis, and interpretation of the data, nor in the writing of the report, nor in the decision to submit the article for publication.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

References

  1. American Psychiatric Association . 2013. Diagnostic and Statistical Manual of Mental Disorders (5th Ed.). American Psychiatric Association. [Google Scholar]
  2. Bagby, R. M. , Gralnick T. M., Al‑Dajani N., and Uliaszek A. A.. 2016. “The Role of the Five‑Factor Model in Personality Assessment and Treatment Planning.” Clinical Psychology: Science and Practice 23, no. 4: 365–378. 10.1111/cpsp.12170. [DOI] [Google Scholar]
  3. Chanen, A. , Sharp C., and Hoffman P.. 2017. “Prevention and Early Intervention for Borderline Personality Disorder: A Novel Public Health Priority.” World Psychiatry: Official Journal of the World Psychiatric Association (WPA) 16, no. 2: 215–216. 10.1002/wps.20429. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Charrad, M. , Ghazzali N., Boiteau V., and Niknafs A.. 2014. “NbClust: An R Package for Determining the Relevant Number of Clusters in a Data Set.” Journal of Statistical Software 61: 1–36. [Google Scholar]
  5. Clark, L. A. 2007. “Assessment and Diagnosis of Personality Disorder: Perennial Issues and an Emerging Reconceptualization.” Annual Review of Psychology 58: 227–257. 10.1146/annurev.psych.57.102904.190200. [DOI] [PubMed] [Google Scholar]
  6. Esguevillas, Á. , Díaz‐Caneja C. M., Arango C., et al. 2018. “Personality Organization and Its Association With Clinical and Functional Features in Borderline Personality Disorder.” Psychiatry Research 262: 393–399. 10.1016/j.psychres.2017.09.013. [DOI] [PubMed] [Google Scholar]
  7. First, M. B. , Bell C. C., Cuthbert B., et al. 2002. “Personality Disorders and Relational Disorders: A Research Agenda for Addressing Crucial Gaps in DSM.” In A Research Agenda for DSM—V, edited by Kupfer D. J., First M. B., Regier D. A., 123–199. American Psychiatric Association. [Google Scholar]
  8. First, M. B. , Gibbon M., Spitzer R. L., Williams J. B. W., and Benjamin L. S.. 1997. Structured Clinical Interview for DSM‐IV Axis II Personality Disorders (SCID‐II. American Psychiatric Press, Inc. [Google Scholar]
  9. Hall, R. C. W. 1995. “Global Assessment of Functioning.” Psychosomatics 36, no. 3: 267–275. 10.1016/S0033-3182(95)71666-8. [DOI] [PubMed] [Google Scholar]
  10. Hamilton, M. 1959. “The Assessment of Anxiety States by Rating.” British Journal of Medical Psychology 32: 50–55. 10.1111/j.2044-8341.1959.tb00467.x. [DOI] [PubMed] [Google Scholar]
  11. Harkness, A. R. , and Lilienfeld S. O.. 1997. “Individual Differences Science for Treatment Planning: Personality Traits.” Psychological Assessment 9, no. 4: 349–360. 10.1037/1040-3590.9.4.349. [DOI] [Google Scholar]
  12. Lengel, G. J. , Helle A. C., DeShong H. L., Meyer N. A., and Mullins‐Sweatt S. N.. 2016. “Translational Applications of Personality Science for the Conceptualization and Treatment of Psychopathology.” Clinical Psychology: Science and Practice 23, no. 3: 288–304. 10.1111/cpsp.12163. [DOI] [Google Scholar]
  13. Livesley, W. J. 2003. “Diagnostic Dilemmas in the Classification of Personality Disorder.” In Advancing DSM: Dilemmas in Psychiatric Diagnosis, edited by Phillips K., First M., and Pincus H. A., 153–189. American Psychiatric Association Press. [Google Scholar]
  14. Lynam, D. R. , Smith G. T., Whiteside S. P., and Cyders M. A.. 2006. The UPPS‐P: Assessing Five Personality Pathways To Impulsive Behavior (Technical Report). Purdue University. [Google Scholar]
  15. Millon, T. , Grossman S., and Millon C., 2018. MCMI‐IV. Inventario Clínico Multiaxial de Millon‐IV. Manual. Madrid: Pearson Clinical.
  16. Montgomery, S. A. , and Åsberg M.. 1979. “A New Depression Scale Designed to Be Sensitive to Change.” British Journal of Psychiatry 134: 382–389. 10.1192/bjp.134.4.382. [DOI] [PubMed] [Google Scholar]
  17. Palomares, N. , Arévalo J., Soto A., et al. 2020. “The Feeling of Emptiness Questionnaire (CSV). Elaboration and Validation.” Actas Espanolas de Psiquiatria 48, no. 5: 220–228. [PubMed] [Google Scholar]
  18. Paris, J. 2018. “The Nature of Borderline Personality Disorder: Multiple Dimensions, Multiple Symptoms, but One Category.” Journal of Personality Disorders 32, no. 1: 23–40. 10.1521/pedi.2007.21.5.457. [DOI] [PubMed] [Google Scholar]
  19. Patton, J. H. , Stanford M. S., and Barratt E. S.. 1995. “Factor Structure of the Barratt Impulsiveness Scale.” Journal of Clinical Psychology 51: 768–774. 10.1002/1097-4679(199511)51:6<768::AID-JCLP2270510607>3.0.CO;2-1. [DOI] [PubMed] [Google Scholar]
  20. Perez, V. , Barrachina J., Soler J., et al. 2007. “The Clinical Global Impression Scale for Borderline Personality Disorder Patients (CGI‐BPD): a Scale Sensible to Detect Changes.” Actas Espanolas de Psiquiatria 35, no. 4: 229–235. [PubMed] [Google Scholar]
  21. Spielberger, C. D. 1999. Professional manual for the State‐Trait Anger Expression Inventory‐2 (STAXI‐2). Psychological Assessment Resources. [Google Scholar]
  22. Stoffers‐Winterling, J. M. , Storebø O. J., and Simonsen E.. 2022. “Perspectives on Dialectical Behavior Therapy and Mentalization‐Based Therapy for Borderline Personality Disorder: Same, Different, Complementary?.” Psychology research and behavior management 15: 3179–3189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Trull, T. J. , and Durrett C. A.. 2005. “Categorical and Dimensional Models of Personality Disorder.” Annual Review of Clinical Psychology 1: 355–380. 10.1146/annurev.clinpsy.1.102803.144009. [DOI] [PubMed] [Google Scholar]
  24. Videler, A. C. , Hutsebaut J., Schulkens J. E. M., Sobczak S., and van Alphen S. P. J.. 2019. “A Life Span Perspective on Borderline Personality Disorder.” Current Psychiatry Reports 21, no. 7: 51. 10.1007/s11920-019-1040-1. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Waugh, M. H. , Bishop J. L., and Schmidt M. R.. 2019. “The AMPD in assessment, treatment planning, and clinical supervision.” In DSM‐5 Alternative Model for Personality Disorders, edited by Hopwood In. C. J., Mulay A. L., and Waugh M. H., 184–208. Routledge. [Google Scholar]
  26. Waugh, M. H. 2019. “Clinical utility and application of the AMPD.” In DSM‐5 Alternative Model for Personality Disorders, edited by Hopwood In. C. J., Mulay A. L., and Waugh M. H., 96–140. Routledge. 10.4324/9781315205076-5. [DOI] [Google Scholar]
  27. Widiger, T. A. , and Trull T. J.. 2007. “Plate Tectonics in the Classification of Personality Disorder: Shifting to a Dimensional Model.” American Psychologist 62, no. 2: 71–83. 10.1037/0003-066X.62.2.71. [DOI] [PubMed] [Google Scholar]
  28. Winsper, C. , Marwaha S., Lereya S. T., Thompson A., Eyden J., and Singh S. P.. 2015. “Clinical and Psychosocial Outcomes of Borderline Personality Disorder in Childhood and Adolescence: A Systematic Review.” Psychological Medicine 45, no. 11: 2237–2251. 10.1017/S0033291715000318. [DOI] [PubMed] [Google Scholar]
  29. World Health Organization . 2019. International Classification of Diseases 11th Revision (ICD‐11). World Health Organization (WHO). [Google Scholar]
  30. Woodbridge, J. , Townsend M., Reis S., Singh S., and Grenyer B. F.. 2022. “Non‐response to Psychotherapy for Borderline Personality Disorder: A Systematic Review.” Australian and New Zealand Journal of Psychiatry 56, no. 7: 771–787. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Zanarini, M. C. , Frankenburg F. R., Hennen J., and Silk K. R.. 2003. “The Longitudinal Course of Borderline Psychopathology: 6‐year Prospective Follow‐up of the Phenomenology of Borderline Personality Disorder.” American Journal of Psychiatry 160, no. 2: 274–283. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Articles from Journal of Clinical Psychology are provided here courtesy of Wiley

RESOURCES