Borderline personality disorder and substance use disorder: a clinical and psychopathological investigation

Abstract

Background

Borderline Personality Disorder (BPD) is a severe psychiatric condition frequently associated with Substance Use Disorder (SUD), with high rates of comorbidity and poor clinical outcomes. Impulsivity and emotional dysregulation are core features shared by both disorders, suggesting common underlying mechanisms. However, less is known about the role of additional factors—such as alexithymia, coping strategies, and quality of life in contributing to dual diagnosis and SUD severity in individuals with BPD. This study aims to: (1) examine the clinical and demographic variables in a sample of patients with BPD comorbid with SUD, in comparison with a group of patients with BPD alone, and (2) identify clinical and psychopathological factors related to the presence of SUD and severity of SUD symptoms in the whole sample (BPD with and without SUD).

Methods

Seventy-five outpatients diagnosed with BPD were recruited and divided into two groups: BPD with comorbid SUD (n = 43) and BPD without SUD (n = 32). Diagnoses were established using SCID-5-CV and SCID-5-PD. Participants were assessed using validated instruments measuring BPD severity (BPDSI-IV), impulsivity (BIS-11), alexithymia (TAS-20), coping strategies (CISS), addiction vulnerability and severity (ABQ), quality of life (SAT-P), and functioning (SOFAS). Group differences were analyzed using chi-square and ANCOVA. Hierarchical logistic and linear regressions were conducted to identify factors associated with the SUD presence and severity.

Results

Patients with BPD and SUD were more frequently male and showed earlier BPD onset, greater symptom severity, higher impulsivity, higher vulnerability to addiction, greater alexithymia, and more use of task-oriented coping strategies. Logistic regression showed that impulsivity and vulnerability to SUD were significantly associated with comorbid SUD, while alexithymia reduced the effect of gender. Motor impulsiveness independently contributed to the presence of dual diagnosis. Linear regression analyses indicated that alexithymia and impulsivity were variables most strictly associated with SUD severity, jointly explaining approximately 60% of the variance. Specifically, difficulty describing feelings and externally oriented thinking contributed to a greater SUD severity.

Conclusions

Impulsivity and alexithymia appear to be key transdiagnostic factors underlying both the presence and severity of SUD in individuals with BPD. These findings highlight the importance of targeting deficits in emotional processing and impulse control in clinical interventions to improve outcomes in this clinical population.

Background

Borderline Personality Disorder (BPD) is a severe personality disorder, with a prevalence of almost 1.6% in the general population, which can rise to 5.9% in adult primary care settings [14]. It is characterized by a pervasive pattern of instability in interpersonal relationships and self-image and typically develops in early adulthood. Additionally, patients with BPD show poor frustration tolerance and experience intense fears of abandonment, to which they may respond with several risky behaviors, including substance abuse. Among clinical features, impulsivity and emotional dysregulation are considered core dimensions of BPD and represent key mechanisms underlying high-risk behaviors, including substance use. In this regard, converging evidence suggests that these psychopathological dimensions are also central to Substance Use Disorders (SUD), supporting the hypothesis of shared underlying mechanisms between the two conditions [36].

Comorbidity between BPD and substance abuse is common and well-documented: about 75% of BPD patients have concurrent substance abuse, and about 20% of patients diagnosed with a SUD are also diagnosed with BPD [60]. This co-occurrence is associated with a more severe clinical course, including higher rates of mood disorders, particularly Major Depressive Disorder (MDD), and poorer overall prognosis [64]. The association between BPD and SUD can be considered within a framework of shared etiological vulnerability. The shared liability may be mediated by common temperamental traits such as impulsivity, emotional instability, and neuroticism [26]. In addition to this biological predisposition, there are environmental factors, including early traumatic experiences, that contribute to the development of both disorders, potentially leading individuals to turn to substance use as a maladaptive form of self-medication [53].

Available data suggest that once one of the two disorders occurs (BPD or SUD), it is likely to become chronic, with a direct and reciprocal effect in the genesis and chronicization of the other disorder [60]. Zanarini et al. highlighted that while short-term remissions from alcohol and drug use disorders are common among individuals with borderline personality disorder (BPD), long-term remissions are less frequent, and recurrences are particularly likely following brief remission periods [65].

Amounting evidence identifies impulsivity as a key factor in SUD, influencing both the onset and the course of addictive behaviors. Individuals with SUD show higher levels of impulsivity, which has also been found to predict poorer treatment outcomes [11, 17, 21, 22, 51]. Nevertheless, impulsivity alone does not fully explain the clinical complexity observed in BPD patients with comorbid SUD. In recent years, growing interest has shifted toward additional transdiagnostic factors that may contribute to this comorbidity or to the severity of addictive behaviors. Among these, alexithymia, defined as difficulty in identifying and describing feelings and a tendency toward externally oriented thinking, has emerged as a central construct. Both BPD and SUD subjects present elevated levels of alexithymia, which is associated with impaired emotional processing and maladaptive regulation strategies. Alexithymia may also account for gender-related differences in clinical presentation; for example, females with BPD show higher levels of alexithymia with a greater tendency toward despair [2]. Gender itself represents another relevant but underexplored variable. Emerging evidence suggested that gender differences play a significant role in the expression and maintenance of BPD [2, 12, 52]. In particular, males with BPD are more likely to present externalizing symptoms (e.g., aggressiveness) and comorbid disorders (e.g., substance use). At the same time, females with BPD are more likely to present internalizing symptoms (e.g., affective instability) and comorbid disorders (e.g., major depressive disorder) [52]. These differences may indicate distinct patterns that lead to SUD vulnerability.

Dysfunctional coping strategies also play a crucial role in the development, course, and treatment outcomes of severe mental disorders, including BPD and SUD [35, 40]. Several studies have observed that both BPD and SUD patients tend to use fewer adaptive coping strategies compared to the general population [16, 40, 50]. Adaptive coping strategies have been associated with lower severity of addictive behaviors, better treatment adherence, and lower relapse rates; however, evidence regarding the specific coping profiles associated with SUD in BPD remains limited [35, 40].

In the context of recovery-oriented treatment for mental illnesses, the subjective perception of quality of life represents a key marker of disease burden and an important aspect of clinical and psychosocial outcomes in both BPD [57] and SUD patients [38]. Some authors have highlighted that, among drug users, psychosis and borderline personality disorder have the greatest impact on quality of life and functioning [15]. Despite the growing body of literature, there is still a lack of studies examining the role of impulsivity, alexithymia, coping strategies, and gender (in addition to other more studied demographic and clinical variables) in detecting both the presence and severity of SUD in individuals with BPD.

To address this gap, the purpose of the present cross-sectional study is to achieve two primary objectives. It is aimed to:

1. examine the clinical and demographic variables in a sample of patients with BPD comorbid with SUD, in comparison with a group of patients with BPD without SUD.

2. identify clinical and psychopathological factors related to the presence of SUD and severity of SUD symptoms in the whole sample (BPD with and without SUD).

Methods

Participants

Subjects were evaluated at the outpatient service of the Center for Personality Disorders in the Department of Neuroscience of the University of Turin, Italy.

Patients were divided into two distinct groups: 1) Group 1: patients with BPD comorbid with SUD; 2) Group 2: patients diagnosed with BPD without comorbid SUD.

An expert clinician (P.B.) made the diagnosis, which was confirmed with the Structured Clinical Interview for DSM- 5 - Clinical Version and Personality Disorders (SCID-5-CV and SCID-5-PD) [27, 28].

Group 1

Inclusion criteria

Outpatients aged between 18 and 60 years with a diagnosis of Borderline Personality Disorder (BPD) and Substance Use Disorder (SUD), according to the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5) criteria [1], were recruited.

All subjects included in the study were clinically stable (with no hospitalizations, no significant changes in pharmacotherapy, no admissions to the emergency room) for at least three months.

Exclusion criteria

1) comorbidity with any psychiatric disorder other than SUD and major depressive disorder (MDD); 2) patients with neurological disorders; 2) a co-occurring major depressive episode; 3) patients with symptom severity and functional impairment deemed by the clinician to be incompatible with the administration of the assessment instruments; 4) patients with intoxication and/or withdrawal symptoms in the last two weeks.

Group 2

Inclusion criteria

Outpatients aged between 18 and 60 years with a diagnosis of BPD according to DSM-5 criteria.

All subjects included in the study were clinically stable (with no hospitalizations, no significant changes in pharmacotherapy, and no emergency room admissions) for at least three months.

Exclusion criteria

1) comorbidity with any psychiatric disorder other than MDD; 2) patients with neurological disorders; 3) a co-occurring major depressive episode; 4) patients with symptom severity and functional impairment deemed by the clinician to be incompatible with the administration of the assessment instruments; 5) past and current alcohol use disorder or substance use disorder.

The trial was carried out in accordance with the recommendations of the Ethics Committee: Comitato Etico Interaziendale A.O.U. Città della Salute e della Scienza di Torino - A.O. Ordine Mauriziano - A.S.L. Città di Torino (approval code: 00483/2022). The principles of the Declaration of Helsinki were followed. Written informed consent was obtained from all subjects. Clinical trial number: not applicable.

Patients included in the study were evaluated using a semi-structured interview to assess demographic and clinical features. Data collected included sociodemographic variables (age, gender, level of education); psychiatric history (age of BPD onset; comorbidity with major depressive disorder (MDD)); current psychopharmacological medications; SUD type (alcohol, cocaine, MDMA, LSD, cannabis, opioids); SUD treatment (disulfiram, acamprosate, methadone, buprenorphine); BPD and SUD psychopathology (BPD specific symptoms; vulnerability for SUD; severity of addictive behaviors); traumatic experiences; alexithymia; impulsivity; coping strategies; functional milestones (employment status, stable romantic relationship, and independent living); global functioning, and subjective quality of life.

Measures

All patients were assessed with the following assessment instruments:

• Borderline Personality Disorder Severity Index - IV (BPDSI-IV [3]);

• Addictive Behavior Questionnaire (ABQ [13]);

• Childhood Trauma Questionnaire – Short Form (CTQ-SF [8]);

• Toronto Alexithymia Scale − 20 (TAS-20 [4]);

• Barratt Impulsiveness Scale − 11 (BIS-11 [7]);

• Coping Inventory for Stressful Situations (CISS [25]);

• Social and Occupational Functioning Assessment Scale (SOFAS [47]);

• Satisfaction Profile (SAT-P [39]);

The BPDSI is a semi-structured clinical interview assessing the frequency and severity of specific BPD symptoms. The interview consists of eight items scored on a 10-point frequency scale (0 = never; 10 = daily), including “abandonment,” “interpersonal relationships,” “impulsivity,” “parasuicidal behavior,” “affective instability,” “feelings of emptiness,” “outbursts of anger,” “dissociation and paranoid ideation,” and of one item scored on a four-point severity scale, concerning “identity.” The total score is the sum of the nine averaged criteria scores (range 0–90). The index, as well as the separate criteria, possesses adequate reliability, as well as discriminant, concurrent, and construct validity, in both the original version and the Italian translation [23].

The ABQ is the Addictive Behavior Questionnaire (ABQ), a self-assessment tool that allows the formulation of a diagnosis of substance-related disorder or addiction disorder according to DSM-5 criteria. It is a multidimensional instrument consisting of a Severity Index (SI) and a Seven Domains Addiction Scale (7DAS). These two sections explore different constructs: SI provides a measure of the intensity and frequency of addictive behaviors, while 7DAS explores the vulnerability/predisposition to addiction. The SI investigates 4 areas of Addiction (Substance, Alcohol, Gambling, and Internet), and for each of the 4 areas, there are two separate scores referring to the last 12 months. Section A of the SI assesses addictive behaviors (yes/no), and Section B measures levels of Craving (never/always). For each section, severity levels have been established, ranging from mild to severe. Section B of the SI involves responses on a Likert scale (never/rarely/sometimes/often/always) and assesses the incidence of three distinct factors in the experience of craving: obsessiveness, impulsivity, and compulsivity. As for the 7DAS scale, it explores 7 domains of vulnerability (impulsivity, compulsivity, emotional regulation, reward sensitivity, attachment, metacognitive functioning, stress sensitivity) believed to be risk factors for the development and maintenance of pathological addiction. Each scale consists of 7 items with a 5-point frequency scale (never to always). ABQ demonstrates good psychometric properties, including strong internal consistency (Cronbach’s α = 0.80–0.90), a stable factorial structure, and satisfactory convergent and discriminant validity [13].

The CTQ-SF is a retrospective questionnaire to evaluate abuse and neglect experiences in childhood. It is a standardized measure to detect a positive history of trauma exposure. CTQ-SF consists of 28 items and five subscales that investigate five different types of childhood trauma: emotional abuse, physical abuse, sexual abuse, emotional neglect, and physical neglect. There is one additional scale to explore the tendency to minimize or deny. Each item is scored on a 5-point frequency scale (1 = never true; 5 = very often true). Scoring for each subscale ranges between 5 and 25. Higher scores indicate more severe exposure to traumatic events. Scoring for the scale minimization/denial ranges between 0 and 3. The scale is valid in the original version and in the Italian translation [33].

The TAS 20, an instrument used to assess alexithymia, is a questionnaire consisting of 20 items and 3 factorial scales: difficulty in identifying feelings and distinguishing between feelings and physical sensation (5 items); difficulty in describing one’s feelings to others (7 items); and externally oriented cognitive style (8 items). Patients can receive a score from 1 to 5 for each item (1 = strongly disagree; 5 = strongly agree). The total alexithymia score is the sum of responses to the 20 items. The TAS-20 score ranges from 20 to 100, with a cut-off for alexithymia corresponding to 60. Values between 52 and 60 are indicative of possible alexithymia, while values below 51 rule out the presence of alexithymia [4].

The BIS-11 is a 30-item self-report questionnaire that measures impulsivity on a 4-point Likert scale. Higher scores for each item indicate higher levels of impulsivity. Twelve items are reverse-scored in order to avoid response sets. Is it possible to identify three factors: attentive impulsiveness, motor impulsiveness, and nonplanning impulsiveness. The global score is obtained by summing these factors. The BIS-11 demonstrates adequate reliability and construct validity in both US [7] and Italian [29] samples.

The CISS is an easy-to-administer multidimensional instrument to measure coping styles (maneuver-oriented, emotion-oriented, or avoidance-oriented). The CISS is made of three scales, each consisting of 16 items. The Task-oriented (T) scale describes efforts aimed at solving a problem through cognitive restructuring. The focus of attention is on the task and design. The Emotion (E) scale describes self-oriented emotional reactions aimed at reducing stress. The scale Avoidance (A) describes cognitive actions and changes enacted to avoid a stressful situation through distraction (task-oriented), or through social diversion (person-oriented) [25].

The SOFAS is a clinician-rated scale to measure impairment in social and occupational areas. It is independent of the psychiatric diagnosis and the severity of the patient’s symptoms. The score ranges from 0 to 100. Higher scores indicate better functioning.

The SAT-P is a self-administered questionnaire consisting of 32 scales that provide a satisfaction profile in daily life and can be considered an indicator of subjective quality of life. The SAT-P considers five different factors: “psychological functioning”; “physical functioning”; “work”; “sleep, food, and free time”; “and social functioning”. The SAT-P asks the patient to evaluate his satisfaction over the last month for each of the 32 life aspects on a 10-cm analog scale ranging from “extremely dissatisfied” to “extremely satisfied”.

The assessment was performed by an investigator (G.L.) who had received training in psychometric instruments before starting the investigation.

Statistical analysis

Descriptive analyses were first conducted to characterize the socio-demographic and clinical characteristics of the two groups: (1) patients with BPD + SUD, and (2) patients with single BPD.

A chi-square test was performed to compare categorical variables between the two groups: group 1 (individuals with BPD and comorbid SUD) and group 2 (individuals with BPD without SUD), while analysis of covariance (ANCOVA) was performed to compare continuous variables between the two groups, adjusting for gender.

In order to achieve the second aim of the study, two hierarchical regressions were conducted by entering as independent variables those that showed a significant difference in the between-group comparisons. In both hierarchical regressions, independent variables were grouped into the following nine blocks: 1): socio-demographical variables; 2) psychiatric history; 3) BPD psychopathology; 4) psychopathological vulnerability to develop SUD; 5) traumatic experiences; 6) alexithymia; 7) impulsivity; 8) coping strategies; 9) global functioning, functional milestones, and subjective quality of life. Given the large number of variables relative to the sample size, the total score was included for each psychometric scale rather than the subscale scores. The only exception is the Coping Inventory for Stressful Situations (CISS), in which the subscales describe different coping strategies that must be considered separately. We conducted a hierarchical logistic regression with the presence of SUD as the dependent variable, followed by a hierarchical linear regression with the severity of addictive behaviors as the dependent variable.

When the total score of a rating scale was significantly associated with the outcome (dependent variable) in the last step of a hierarchical regression that significantly increased the model fit, we conducted a further regression analysis entering the subscale scores of that/those rating scale(s) as the independent variables.

Statistical analyses were conducted using the IBM Statistical Package for the Social Sciences (SPSS), Version 29.0 for Windows (IBM Corp., Armonk, NY).

The significance level was p ≤ 0.05.

Results

Seventy-five patients (43 with BPD + SUD and 32 with BPD without SUD) were recruited for the study. Descriptive statistics of the sociodemographic and clinical variables of the sample are shown in Table 1.

Table 1.

Socio-demographic and clinical characteristics of the sample

Variable	BPD + SUD (n = 43)		BPD (n = 32)
	Mean / N	SD / %	Mean / N	SD / %
Age, y	31	10	34	14
Level of education, y	13.9	3.5	14.3	3.0
Gender (F)	24	55.8	29	90.6
Employed	27	62.8	19	59.4
Stable romantic relationship	19	44.2	14	43.8
Independent living	11	25.6	10	31.3
Age of BPD onset, y	17.5	4.3	19.3	4.6
Comorbidity with MDD	16	37.2	19	59.3
Mood stabilizers	29	67.4	21	65.6
Atypical antipsychotics	19	44.2	14	43.8
Antidepressants	4	9.3	3	9.4
Disulfiram	2	4.7	-	-
Acamprosate	2	4.7	-	-
Methadone	1	2.3	-	-
Buprenorphine	2	4.7	-	-
Alcohol use	28	66.1	-	-
Cocaine use	14	32.5	-	-
MDMA	4	9.3	-	-
LSD	5	11.6	-	-
Cannabis	24	55.8	-	-
Opioids	7	16.3	-	-
Polydrug use	35	81.4	-	-
ABQ SI total score	27.0	17.6	4.7	10.3
BPDSI total score	35.0	16.2	28.1	10.4
ABQ-7DAS total score	187.6	43.2	147.8	37.4
BIS-11 total score	91.5	11.9	64.2	10.1
CISS Task-oriented coping subscale	49.8	13.8	52.7	9.9
CISS Emotion-oriented coping subscale	52.8	13.1	57.2	7.4
CISS Avoidance-oriented coping subscale	28.8	7.2	29.8	7.3
CTQ-SF total score	66.8	16.0	64.0	13.1
TAS-20 total score	75.5	15.0	59.4	9.0
SAT-P total score	173.7	35.9	186.8	52.6
SOFAS total score	67.8	15.6	74.4	13.5

Abbreviations: y = years; BPD = Borderline Personality Disorder; SUD = Substance Use Disorder; F = Female; MDD = major depressive disorder; MDMA = 3,4 – Methylenedioxymethamphetamine; LSD = Lysergic Acid Diethylamide; ABQ SI = Addictive Behavior Questionnaire Severity Index for Substances total score; ABQ-7DAS = Addictive Behavior Questionnaire Seven Domain Addictive Scale; BPDSI = Borderline Personality Disorder Severity Index; BIS-11 = Barratt Impulsiveness Scale; TAS-20 = Toronto Alexithymia Scale; CISS = Coping Inventory for Stressfull Situations; SAT-P = Satisfaction Profile; SOFAS = Social and Occupational Functioning Assessment Scale; SD = standard deviation

As in our sample, the gender distribution was significantly unbalanced between groups (in the group with BPD + SUD: 24 females and 19 males, in the group with single BPD: 29 females and 3 males), we performed ANCOVA controlling for gender.

Chi-square test and ANCOVA revealed significant differences between groups in: gender (p = 0.001), age of BPD onset (p = 0.006), ABQ SI-substances total score (p < 0.001), ABQ-7DAS (<0.001), BPDSI total score (p = 0.006), BIS-11 total score (p < 0.001), CISS “Task-Oriented Coping” subscale (p = 0.04), TAS-20 total score (p < 0.001), SAT-P total score (p = 0.04).

ANCOVA’s effect sizes (η²) of variables that showed a significant between-group difference ranged between 0.05 and 0.58 (mean value: 0.21).

Variables in which gender had a significant effect in the ANCOVA were: age of BPD onset (p = 0.006), BPDSI total score (p = 0.028), and CISS “Task-Oriented coping” subscale (p = 0.006).

No statistically significant differences were found between the two groups in the following variables: age (p = 0.33), level of education (p = 0.85), employment (p = 0.47), stable romantic relationship (p = 0.58), independent living (p = 0.38), comorbidity with major depression (p = 0.57), treatment with antidepressants (p = 0.99), treatment with mood stabilizers (p = 0.87), treatment with atypical antipsychotics (p = 0.97), CISS “Emotion-oriented coping subscale” (p = 0.27), CISS “Avoidance-oriented coping subscale” (p = 0.34), and SOFAS (p = 0.22).

These results of the between-group comparisons are summarized in Table 2.

Table 2.

Comparison of demographic and clinical variables between patients with BPD + SUD (n = 43) and patients with single BPD (n = 32), using χ² test and ANCOVA (controlling for gender)

Variable	BPD + SUD (n = 43) Mean ± SE / N (%)	BPD (n = 32) Mean ± SE / N (%)	F / χ2	p
Age, y	31.21 ± 1.91	34.21 ± 2.24	0.97	0.332
Female, n	24 (55.8%)	29 (90.6%)	10.73	0.001
*Age of BPD onset, y	16.98 ± 0.66	19.96 ± 0.78	7.86	0.006
Level of education, y	14.00 ± 0.51	14.15 ± 0.6	0.03	0.850
Employment	27 (62.8%)	19 (59.4%)	0.09	0.763
Stable romantic relationship	19 (44.2%)	14 (43.7%)	0.01	0.969
Independent living	11 (25.6%)	10 (31.2%)	0.29	0.583
Comorbidity with major depression	16 (37.2%)	19 (59.3%)	2.98	0.084
Treatment with antidepressants	4 (9.3%)	3 (9.4%)	<0.01	0.990
Treatment with mood stabilizers	29 (67.4%)	21 (65.6%)	0.03	0.868
Treatment with atypical antipsychotics	19 (44.2%)	14 (43.7%)	<0.01	0.971
ABQ SI-substances total score	26.81 ± 2.36	5.41 ± 2.77	31.15	<0.001
ABQ 7DAS total score	190.98 ± 6.31	143.26 ± 7.4	22.4	<0.001
*BPDSI total score	36.25 ± 2.16	26.39 ± 2.52	8.16	0.006
BIS-11 total score	91.81 ± 1.77	63.73 ± 2.08	98.53	<0.001
*CISS Task-Oriented Coping subscale	48.48 ± 1.86	54.52 ± 2.18	4.14	0.040
CISS Emotion-Oriented coping subscale	53.33 ± 1.74	56.43 ± 2.04	1.24	0.272
CISS Avoidance-Oriented coping subscale	28.49 ± 1.14	30.25 ± 1.33	0.94	0.344
TAS-20 total score	75.21 ± 2.03	59.81 ± 2.39	22.50	<0.001
SAT-P total score	170.38 ± 6.78	191.33 ± 7.95	4.52	0.041
SOFAS	68.68 ± 2.3	73.24 ± 2.7	1.53	0.223

*Variables in which gender has a significant effect in the ANCOVA. Bold p-values are statistically significant (<0.05)

Abbreviations: y = years; BPD = Borderline Personality Disorder; SUD = Substance Use Disorder; ABQ SI = Addictive Behavior Questionnaire Severity Index for Substances total score; ABQ-7DAS = Addictive Behavior Questionnaire Seven Domain Addictive Scale; BPDSI = Borderline Personality Disorder Severity Index; BIS-11 = Barratt

Hierarchical logistic regression analysis revealed a significant association between gender and the presence of SUD in comorbidity with BDP, accounting for 14.6% of the Cox and Snell pseudo-R² (C&S-R²). Female gender was associated with a lower probability of comorbid SUD (Exponential -Exp- B = 0.131; p = 0.003). In Step 2, adding the psychiatric history (age of BPD onset) to the model led to an increase in the C&S-R² to 23.1%. This increase was significant (χ² of the modification –χ²-M– = 7.91; p = 0.005). An earlier BDP onset was associated with a higher likelihood of a comorbid SUD (Exp B of the age of BDP onset = 0.831; p = 0.011). In Step 3, when entering the model BPD psychopathology (BPDSI), the C&S-R² increased to 28.3%. This change was significant (χ²-M = 5.27; p = 0.022). More severe BPD symptoms were associated with the presence of SUD in comorbidity with BPD (Exp B = 1.044; p = 0.032). In Step 4, the inclusion of psychopathological vulnerability for SUD (ABQ-7DAS) in the model led the C&S-R² to 37%. This increase was significant (χ²-M = 9.66; p = 0.022). Having a higher psychopathological vulnerability for addictive behaviors was associated, as expected, with a higher probability of co-occurring SUD (Exp B = 1.025; p = 0.005). Moreover, the association with BPD psychopathology was no longer significant. In Step 5, the model showed an increase in the C&S-R² to 40% (χ²-M = 3.88; p = 0.046). Including alexithymia (TAS-20) in the model invalidated the significance of the impact of gender on having comorbidity of SUD and BPD. Vulnerability to SUD remained significant in the model (Exp B = 1.073; p = 0.047). The association between alexithymia and the outcome (SUD comorbidity) was not significant. In Step 6, when adding impulsivity (BIS-11), the model led to an uptrend in the C&S-R² value (67.9%). This change was statistically significant (χ²-M = 46.85; p = <0.001). Higher levels of impulsiveness were associated with the likelihood of dual diagnosis (Exp B = 1.642; p = 0.004). In Step 7, incorporating the coping strategy (CISS “Task-oriented coping subscale”) into the model resulted in a further significant increase (χ²-M = 8.22; p = 0.008) in C&S-R² to 74.5%. Task-oriented coping strategy was significantly associated with a higher chance of a comorbid SUD (Exp B = 1.474; p = 0.016). When this variable was included in the model, the association between vulnerability to SUD and SUD comorbidity was no longer significant. In Step 8, the entering of the subjective quality of life (SAT-P) into the model was not associated with a significant increase in the C&S-R² (χ²-M = 0.54; p = 0.461).

The results of the hierarchical logistic regression are shown in Table 3.

Table 3.

Hierarchical logistic regression (n = 75) including demographical and clinical variables associated with the presence of SUD in patients with BPD

Step	Variables in the model	Exp B	CI 95% Exp B	p	R2C&S	χ2-M	p
1 Socio-demographics	Female gender	0.131	0.034; 0.495	0.003	14.6%	11.82	<0.001
2 Psychiatric history	Female gender	0.069	0.015; 0.328	<0.001	23.1%	7.91	0.005
	Age of BPD onset	0.831	0.726; 0.958	0.011
3 BPD psychopathology	Female gender	0.056	0.011; 0.282	<0.001	28.3%	5.27	0.022
	Age of BPD onset	0.852	0.741; 0.982	0.027
	BPDSI	1.044	1.004; 1.087	0.032
4 Vulnerability for SUD	Female gender	0.050	0.009; 0.275	<0.001	37.0%	9.66	0.002
	ABQ-7DAS	1.025	1.008; 1.044	0.005
5 Alexithymia	ABQ-7DAS	1.043	1.002 1.067	0.047	40.0%	3.88	0.046
6 Impulsivity	ABQ-7DAS	1.057	1.003; 1.115	0.041	67.9%	46.85	<0.001
	BIS-11	1.642	1.176; 2.288	0.004
7 Coping strategies	ABQ-7DAS	1.081	1.003; 1.172	0.042	74.5%	8.22	0.008
	BIS-11	2.533	1.290; 4.878	0.007
	CISS Task-Oriented Coping subscale	1.474	1.074; 2.041	0.016
8 Global functioning, functional milestones, subjective quality of life	No variables statistically significant	-	-	-	-	0.54	0.461

Dependent variable: “case (BPD+ SUD+) / control (BPD+ SUD-)”

Independent variables significantly associated with the dependent variable within each step were reported

Abbreviations: C&S = Cox and Snell; Exp B = exponentiated B; CI = confidence interval; BPD = Borderline Personality Disorder; ABQ 7DAS = Addictive Behavior Questionnaire Seven Domain Addictive Scale total score; BPDSI = Borderline Personality Disorder Severity Index; BIS-11 = Barratt Impulsiveness Scale; TAS-20 = Toronto Alexithymia Scale; CISS = Coping Inventory for Stressful Situation

The logistic regression analysis including the subscales (BIS-11 subscales) of the clinical variables significantly associated with the presence of SUD (impulsivity and task-oriented coping strategy) in the last step of the hierarchical logistic regression that significantly increased the model fit (Step 7), showed that “motor impulsiveness” (BIS-11, motor impulsiveness subscale) was significantly and independently associated with the presence of SUD in comorbidity with BPD (Exp B = 1.216; p = 0.017). On the contrary, “attentive impulsiveness” (BIS-11, attentive impulsiveness subscale) (Exp B = 0.925; p = 0.271); “nonplanning impulsiveness” (BIS-11, nonplanning impulsiveness subscale) (Exp B = 1.102; p = 0.126); and task-oriented coping strategy (CISS, Task-oriented coping subscale) (Exp B = 0.996; p = 0.875), were not significantly associated with dual diagnosis.

These results are shown in Table 4.

Table 4.

Logistic regression analysis (n = 75), adding the subscales of the clinical variables that resulted statistically significant in the last step of the hierarchical logistic regression, evaluating the association with the presence of SUD in patients with BPD

Variables	Exp B	CI 95% Exp B	p
BIS-11, Attentive impulsiveness subscale	0.925	0.806; 1.061	0.271
BIS-11, Motor impulsiveness subscale	1.216	1.035; 1.428	0.017
BIS-11, Nonplanning impulsiveness subscale	1.102	0.973; 1.248	0.126
CISS, Task-oriented coping subscale	0.996	0.949; 1.046	0.875

Dependent variable: “case (BPD+ SUD+) / control (BPD+ SUD-)”

Bold p-values are statistically significant (<0.05)

Abbreviations: Exp = exponentiated; CI = confidence interval; BPD = Borderline Personality Disorder; SUD = substance use disorder; BIS-11 = Barratt Impulsiveness Scale; CISS = Coping Inventory for Stressful Situations.; SE = standard error

Results from the hierarchical linear regression analysis revealed a significant effect of gender on SUD severity (ABQ-SI) in Step 1, accounting for 9% of the variance (R2). Female gender was associated with a lower severity of addictive behaviors (Standardized -St.-β = −0.301; p = 0.009). In Step 2, adding into the model the psychiatric history (age of BPD onset), 5.1% of the variance of severity of SUD (ABQ-SI) was further explained (R² = 14.1%). This change was significant (F of the modification –Fm– = 4.250; p = 0.043). An earlier BDP onset was associated with more severe SUD symptoms (St. β = −0.230; p = 0.043). Adding BPD psychopathology (BPDSI) in Step 3 explained an additional 5.8% of the variance in severity of SUD (ABQ-SI) (R² = 19.9%). This change was significant (Fm = 5.176; p = 0.026). More severe BPD symptoms were associated with SUD severity (St. β = 0.247; p = 0.026). In Step 4, the psychopathological vulnerability for SUD (ABQ-7DAS) was added, resulting in a significant increase in R² of 10.3% (Fm = 10.32; p = 0.002). With the addition of this independent variable, the association between the severity of SUD and the type and severity of BPD-specific symptoms was no longer significant. Conversely, a vulnerability to SUD was positively associated with the severity of the disorder (St. β = 0.397; p = 0.002).

In Step 5, alexithymia (TAS-20) was added, resulting in a significant increase in R² of 18.6% (Fm = 25.13; p < 0.001). A more marked alexithymia was associated with more severe SUD symptoms (St. β = 0.620; p = <0.001). Furthermore, when alexithymia is included in the regression model, the association with vulnerability to substance use disorder is no longer significant, nor is the association with gender. With the addition of impulsivity (BIS-11) in Step 6, R² further significantly increased (R² = 63.9%), explaining an additional 15.1% of the variance in severity of SUD (Fm = 28.28; p < 0.001). Together, alexithymia and impulsiveness explained about 60% of the variance in the severity of SUD (Adj. R² = 60.7%). These two symptom dimensions made the largest contributions to SUD severity (alexithymia: St. β = 0.298, p = 0.016; impulsivity: St. β = 0.571, p < 0.001). In Step 7, adding the coping strategies (CISS “Task- oriented coping subscale”) to the model did not significantly increase model fit, resulting in a non-significant difference of explained variance (Fm = 0.821; p = 0.368). Similarly, in Step 8, entering the subjective quality of life (SAT-P) into the model did not significantly increase the explained variance (Fm = 0.193; p = 0.662).

The results of the hierarchical linear regression are shown in Table 5.

Table 5.

Hierarchical linear regression analysis (n = 75) including demographical and clinical variables associated with the severity of SUD in patients with BPD

Step	Variables in the model	St. β	β	IC 95% β	p	R2	Adj. R2	ΔR2	Fm	p
1 Socio-demographics	Feale gender	− 0.301	− 12.148	− 21.151; − 3.144	0.009	9.0%	7.8%	7.8%	7.231	0.009
2 Psychiatric history	Female gender	− 0.348	− 14.079	− 23.086 − 5.071	0.003	14.1%	11.7%	5.1%	4.250	0.043
	Age of BPD onset	− 0.230	− 0.955	− 1.879; − 0.032	0.043
3 BPD psychopathology	Female gender	− 0.374	− 15.120	− 23.926; − 6.314	0.001	19.9%	16.5%	5.8%	5.176	0.026
	BPDSI	0.247	0.319	0.039; 0.598	0.026
4 Vulnerability for SUD	Female gender	− 0.344	−13.901	− 22.217; − 5.584	0.001	30.2%	26.2%	10.3%	10.32	0.002
	ABQ-7DAS	0.397	0.163	0.062; 0.264	0.002
5 Alexithymia	TAS-20	0.620	0.765	0.460; 1.069	<0.001	48.8%	45.1%	18.6%	25.13	<0.001
6 Impulsivity	TAS-20	0.298	0.368	0.078; 0.665	0.016	63.9%	60.7%	15.1%	28.28	<0.001
	BIS-11	0.571	0.634	0.396; 0.872	<0.001
7 Coping strategies	No variables statistically significant	-	-	-	-	-	-	-	0.821	0.368
8 Global functioning, functional milestones, subjective quality of life	No variables statistically significant	-	-	-	-	-	-	-	0.193	0.662

Dependent variable: Addictive Behavior Questionnaire - Severity Index (ABQ-SI)

Independent variables significantly associated with the dependent variable within each step were reported

Abbreviations: St. = standardized; Adj. = adjusted; CI = confidence interval; BPD = Borderline Personality Disorder; ABQ-7DAS = Addictive Behavior Questionnaire Seven Domain Addictive Scale total score; BPDSI = Borderline Personality Disorder Severity Index; BIS-11 = Barratt Impulsiveness Scale; TAS-20 = Toronto Alexithymia Scale; CISS = Coping Inventory for Stressful Situations

The linear regression analysis including the subscales (BIS-11 and TAS-20 subscales) of the clinical variables significantly associated with SUD severity (alexithymia and impulsivity) in the last step of the hierarchical linear regression that significantly increased the model fit (Step 6) revealed that difficulty describing feelings and externally-oriented thinking (TAS-20, “Difficulty describing feelings subscale” and “externally-oriented thinking subscale”) were significantly associated with the severity of SUD (difficulty describing feelings: St. β = 0.306, p = 0.002; externally-oriented thinking: St. β = 0.616, p < 0.001). On the contrary, the three subscales of the BIS-11: “attentive impulsiveness subscale” (St. β = 0.209, p = 0.567), “motor impulsiveness subscale” (St. β = 0.061, p = 0.549), nonplanning impulsiveness subscale” (St. β = 0.037, p = 0.697) and the subscale of the TAS-20 “difficulty identifying feelings” (St. β = 0.087, p = 0.409) were not significantly associated with the severity of SUD.

These results are shown in Table 6.

Table 6.

Linear regression analysis (n = 75), adding the subscales of the clinical variables that resulted statistically significant in the last step of the hierarchical linear regression, evaluating the association with the severity of SUD in patients with BPD

Variables	St. β	β	CI 95% β	p
BIS-11, Attentive impulsiveness subscale	0.209	0.817	− 0.025; 1.659	0.567
BIS-11, Motor impulsiveness subscale	− 0.061	− 0.266	− 1.149; 0.617	0.549
BIS-11, Nonplanning impulsiveness subscale	− 0.037	− 0.135	−0.821; 0.552	0.697
TAS-20, Difficulty describing feelings subscale	0.306	1.022	0.382; 1.663	0.002
TAS-20, Difficulty identifying feelings subscale	− 0.087	−0.265	−0.902; 0.372	0.409
TAS-20, Externally-oriented thinking subscale	0.616	1.265	0.901; 1.629	<0.001

Dependent variable: Addictive Behavior Questionnaire - Severity Index (ABQ-SI)

Bold p-values are statistically significant (<0.05)

Abbreviations: St. = standardized; CI = confidence interval; BPD = Borderline Personality Disorder; SUD = substance use disorder; ABQ SI-substances = Addictive Behavior Questionnaire Severity Index for Substances; BIS-11 = Barratt Impulsiveness Scale; TAS-20 = Toronto Alexithymia Scale; St. standardized

Discussion

The first aim of the present study was to evaluate socio-demographic, clinical, and psychopathological features in patients with BPD comorbid with SUD compared with patients with BPD alone. The second objective of the study was to identify which factors were associated with the presence and/or the severity of SUD within a sample of individuals with BPD.

Some recent studies have compared specific characteristics of patients with BPD and SUD with patients with other personality disorders [65], patients with single SUD [9, 30], patients with PTSD [34], and the general population [48]. The number of studies comparing patients with BPD with or without comorbid SUD is very limited [10, 37].

Overall, our findings confirmed and extended available evidence suggesting the key role of emotional dysfunction and impulsive dyscontrol as shared vulnerability dimensions underlying SUD in comorbidity with BPD [36, 59, 60].

Consistent with prior literature, patients with comorbid BPD and SUD were more frequently male and exhibited a more severe clinical profile. In particular, our results showed that subjects with comorbidity of BPD and SUD, compared with subjects with BPD alone, presented an earlier BPD onset, more severe BPD symptoms, more severe addictive behaviors, higher vulnerability to SUD, higher impulsiveness, more compromised task-oriented coping strategies, more marked alexithymia, and a lower level of subjective well-being.

In our study, gender emerged as a significant factor in the initial steps of the regression models, with female gender associated with both a lower likelihood and lower severity of SUD. This result is in line with existing literature, which indicates that males with BPD are more prone to develop SUD as a dysfunctional coping strategy, whereas females with BPD are more likely to engage in self-injurious behaviors and exhibit depressive symptoms [9, 12, 19, 42, 55, 56]. From our results, female gender could be considered as a protective factor for both the presence and the severity of SUD.

Notably, the effect of gender was no longer significant after including alexithymia in both the logistic and linear regression models. In BPD, unlike in the general population, alexithymia is more pronounced among women [2], and, at the same time, a deficit in emotional awareness is prominent in patients with SUD. Thus, adding alexithymia to the model could offset the impact of gender in having comorbidity of SUD and BPD. This interpretation is supported by evidence indicating that alexithymia plays a transdiagnostic role across psychiatric disorders and may partially account for gender-related differences in symptom expression [2, 41].

An earlier BDP onset was associated with a higher likelihood of comorbid SUD as well as a higher severity of addictive behaviors. Early onset of BPD in patients with SUD is characterized by more severe, impulsive, and treatment-resistant symptoms. This comorbid young population frequently shows higher suicide risk and faster progression to addiction [31]. Early-emerging emotional dysregulation could increase the likelihood of resorting to maladaptive coping strategies, including substance use [18, 60]. Furthermore, early-onset borderline personality disorder has been associated with a more severe and persistent clinical course, particularly in the presence of comorbid conditions [63].

Severity of BPD symptoms was associated with both SUD comorbidity and severity. When adding vulnerability to SUD to the regression models, the association between specific BPD symptom severity and SUD (comorbidity and severity) was no longer significant. Moreover, a higher psychopathological vulnerability for SUD and severity of addictive behaviors was associated, as expected, with a higher probability of co-occurring SUD. A possible explanation is that addiction-specific vulnerability represents a more proximal index of SUD outcomes than BPD symptoms severity per se [26, 53, 60] and that the instrument of evaluation used to identify substance use vulnerability (ABQ- 7DAS) appears to be more suitable for detecting comorbid conditions.

One of the most relevant findings of the present study concerns the role of alexithymia and impulse dyscontrol. In particular, when we consider alexithymia and impulsivity together, they account for a large proportion (60.7%) of the variance in SUD severity. Impulse dyscontrol also surfaced as a key factor in dual diagnosis. This finding suggested that these two psychopathological dimensions (alexithymia and impulsivity) can be considered as core transdiagnostic dimensions. In fact, both dimensions can be underpinned by fronto-limbic dysfunctions, circuits involved in emotion and behavioral regulation [44, 61]. Greater fronto-limbic dysregulation (increased amygdala activity and decreased prefrontal control) is generally associated with higher severity of both BPD and SUD symptoms [42, 43, 62].

Difficulty in identifying, processing, describing, and working with one’s own feelings could lead to maladaptive strategies, including substance abuse, to regulate emotions and relieve stress. Thus, addictive behaviors could represent a mode of coping with unprocessed emotions. This finding is in agreement with the relevant literature reporting high rates of alexithymia in individuals with substance use [32, 45, 46, 58], especially in the young population [24, 49]. In our study, among the different alexithymia constructs, difficulty describing feelings and externally oriented thinking were independently associated with SUD severity. Externally-oriented thinking, defined as a cognitive style characterized by a preference for focusing on external events and objects, may lead individuals to rely on substances as a way of regulating internal states [20]. Our findings indicate that alexithymia is not a significant factor in distinguishing between the presence and absence of SUD, but it is a key factor in explaining the intensity and frequency of addictive behaviors. With regard to impulsivity, our results are consistent with those from previous studies showing higher levels of impulse dyscontrol in BPD patients, who prefer immediate, smaller rewards to higher future benefits [6]. The authors suggested that in patients with comorbid BPD and SUD, impulsivity is the origin of the tendency to pursue short-term compensations through substance use. This powerful mechanism is particularly active at the early stages of substance use, when “positive” reward effects predominate [5, 9, 54], and seems to be a cross-cutting mechanism that contributes to both the onset and maintenance of the two conditions [60]. Moreover, impulsivity significantly contributed to the severity of SUD, confirming its role not only in the initiation but also in the maintenance of addictive behaviors [11, 22].

Among the components of impulsivity, motor impulsivity was found to be independently associated with dual diagnosis. This finding is consistent with previous studies suggesting that a lack of impulse control and a tendency to act without reflection are closely linked to substance use and the risk of relapse [17, 21, 51].

Results related to coping strategies are not easy to describe or comment on. No differences were found between the two groups (BPD + SUD patients and single BPD patients) in the use of emotion-oriented coping strategies and in avoidance-oriented coping strategies. In contrast, the use of a task-oriented coping strategy (a proactive, problem-focused strategy aimed at managing stress by directly addressing its cause) was significantly lower in patients with dual diagnosis. Moreover, difficulties in task-oriented coping strategies were associated with a higher likelihood of SUD in the hierarchical logistic regression, but not with SUD severity in the linear regression. Previous studies have reported that individuals with BPD and SUD show reduced coping flexibility and rely on maladaptive strategies regardless of the type of coping strategy [35, 40]. The fact that, among coping strategies, the impairment of task-oriented coping strategies is what determines a tendency to develop SUD in patients with BPD may be linked to both impulsivity and alexithymia. Indeed, we hypothesize that individuals with a dual diagnosis tend to focus more on external factors and engage in action-oriented behaviors without adequate emotional processing and action planning. In other words, they do not tackle the problem or the cause of stress by formulating a plan to act on it directly. However, they diverge without reflection on substance abuse as an alternative coping strategy.

In the present study, a subjectively negative perception of quality of life was significantly higher in patients with dual diagnosis than in patients with BPD alone. Dissatisfaction with several areas of one’s life may facilitate the tendency to seek detachment through substance misuse, as a maladaptive strategy aimed at temporarily alleviating psychological distress. Conversely, having a dual diagnosis can contribute to dissatisfaction with one’s quality of life. Available evidence reported a significant association between low life satisfaction and the use of substances, suggesting that dissatisfied individuals may turn to substance use as a dysfunctional coping strategy [38, 57, 66]. Subjective perception of quality of life did not significantly contribute to the regression models, indicating that the level of quality of life does not provide information beyond impulsivity, alexithymia, and task-oriented coping strategies in distinguishing between the presence and the severity of SUD for detecting patients with dual diagnosis and more severe addictive behaviors.

The present study suffers from some limitations. The main limitations of the study relate to its cross-sectional design, which does not allow for the establishment of a causal relationship between the observations and the outcomes studied—namely, the presence and severity of SUD—and to the selection of a clinically stable sample with an underrepresentation of males in the group of BPD patients without SUD. The observed results cannot be generalized to populations of patients with acute symptoms, such as, for example, BPD patients experiencing a major depressive episode or SUD patients with a recent withdrawal syndrome. Furthermore, despite the statistical tools used to account for gender imbalance, this phenomenon may still have partially influenced the results, limiting their generalizability to populations of men with DBP and without SUD. Moreover, participants were recruited from a general psychiatry outpatient setting rather than a specialized substance use unit. They were primarily drawn from BPD programs rather than substance abuse services, which may have led to the underrepresentation of individuals with a primary SUD. Finally, the sample should be enlarged to strengthen the validity and reliability of the findings and to enable stratification of BDP patients by the substance(s) used.

Several strengths should also be noted, particularly the focus on comparing two subgroups of patients with BPD with or without SUD and the use of scales and statistical methods that enabled a detailed, granular examination of the contributions of alexithymia, impulsivity, and coping strategies to the presence and severity of SUD in BPD patients.

Conclusions

Results of the present study, in substantial accordance with current literature, suggested several differences between patients with a dual diagnosis (BPD and SUD) and those with BPD alone. Patients with dual diagnosis were more frequently male and exhibited a more severe clinical profile: an earlier BPD onset, more severe BPD symptoms, more severe addictive behaviors, higher vulnerability to SUD, higher levels of alexithymia and impulsiveness, more impaired task-oriented coping strategies, and a lower level of subjective well-being.

Finally, findings highlighted the importance of considering the contribution of multiple factors, particularly impulsivity and alexithymia, to comorbidity with SUD and severity of addictive behaviors in BPD patients. These results have important clinical implications, suggesting that interventions targeting emotional awareness and impulse control might be particularly beneficial in reducing the risk and severity of substance use in BPD subjects.

Abbreviations

ABQ

Addictive Behavior Questionnaire

ABQ SI-substances

Addictive Behavior Questionnaire - Severity Index for Substances

ABQ 7DAS

Addictive Behavior Questionnaire - Seven Domains Addiction Scale

ANCOVA

Analysis of covariance

ANOVA

Analysis of variance

BIS-11

Barratt Impulsiveness Scale - 11

BPD

Borderline Personality Disorder

BPDSI-IV

Borderline Personality Disorder Severity Index - IV

CTQ-SF

Childhood Trauma Questionnaire - Short Form

CISS

Coping Inventory for Stressful Situations

DSM-5

Diagnostic and Statistical Manual of Mental Disorders, 5th Edition

MDD

Major Depressive Disorder

PTSD

Post-Traumatic Stress Disorder

MDMA

3,4-Methylenedioxymethamphetamine, SAT-P Satisfaction Profile

SOFAS

Social and Occupational Functioning Assessment Scale

SUD

Substance Use Disorder

SCID-5-CV

Structured Clinical Interview for DSM-5 - Clinical Version

SCID-5-PD

Structured Clinical Interview for DSM-5 - Personality Disorders

SPSS

Statistical Package for the Social Sciences

TAS-20

Toronto Alexithymia Scale – 20

Author contributions

P.B. & S.B. initiated, conceptualized, and designed the study; drafted the initial manuscript. G.M. & G.L. collected data. C.B. performed the statistical analyses. P.B. & C.B. wrote the final version of the manuscript. P.R. & S.B. supervised and edited the manuscript. All of the authors revised, edited, and approved the final manuscript.

Funding

None.

Data availability

Our institution does not allow the sharing of raw data concerning sensitive patient issues (not entirely anonymous). The data is stored by the authors and protected by passwords.

Declarations

Ethics approval and consent to participate

The trial was carried out in accordance with the recommendations of the Ethics Committee: Comitato Etico Interaziendale A.O.U. Città della Salute e della Scienza di Torino - A.O. Ordine Mauriziano - A.S.L. Città di Torino (approval code: 00483/2022). All participants signed a written consent for participation, but not consent for publication sensitive data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.