The clinical sample comprised 202 Spanish participants diagnosed with BPD, 80.7%, n = 163, of whom were female. The mean age was 28.93 years (SD = 9.40). Regarding their marital status, 46.5%, n = 94, were married or had a partner, and 53.46%, n = 108, were single, divorced, or widowed. Regarding the educational level, 2.5%, n = 5, had no studies; 27.7%, n = 56, had primary school level studies; 48%, n = 97, had a high school education; and 21.8%, n = 44, had university level studies. Regarding the participants’ psychiatric comorbidity, 78.2%, n = 158, matched the criteria for another mental disorder. Of them, 62.5%, n = 127, met eating disorder criteria; 6.5%, n = 13, met abuse substance disorder criteria; 3.6%, n = 7, met obsessive compulsive disorder criteria; 2.9%, n = 6, met anxiety disorder criteria; and 2.7 %, n = 5, met mood disorder criteria. The mean score on the Global assessment of functioning from the DSM-IV was 56.18 (SD = 33.69). The number of inpatient hospitalizations in the past year was 0.84 (SD = 1.87), the frequency of suicide attempts in the past year was 0.41 (SD = 0.98), and the frequency of non-suicidal self-injuries (NSSI) in the past year was 2.89 (SD = 7.63). Moreover, 34.2% presented a physical illness.
The inclusion criterion was: patients who met the full DSM-IV (39) criteria for BPD. The exclusion criteria included moderate or severe intellectual disability and meeting the criteria for schizophrenia or another psychotic disorder. Participants were European Whites. They were recruited consecutively. Participants were volunteers who did not receive any compensation for their participation, and they signed an informed consent form. Ethical approval to carry out this study was granted by the University Ethics Committee of Clinical Studies.
The non-clinical sample comprised 201 Spanish university students without mental disorder diagnoses; 72.6%, n = 146, were female. Regarding their marital status, 38.3%, n = 71, were married or had a partner, and 61.7%, n = 124, were single. Participants ranged in age from 18 to 60 years, with a mean age of 22.37 years (SD = 5.42). Participation was voluntary, and they did not receive any compensation. Informed consent was obtained from all participants.
Structured clinical interview for DSM-IV axis I disorders (SCID I) (40). This is an interview for making the major DSM-IV-TR (39) Axis I diagnoses. It offers good psychometric properties: Kappa .66, demonstrating reliability (41). The Spanish version shows psychometric properties similar to those of the original scale (42).
Structured clinical interview for DSM-IV axis II personality disorders (SCID II) (43). This is an interview for making DSM-IV-TR (39) Axis II Personality Disorder diagnoses. It includes 119 questions and has a Kappa of .74, demonstrating reliability for admitted patients (43). The Spanish version shows psychometric properties similar to those of the original scale (44).
Relevant clinical information inventory. Created ad hoc for this research, it collects the frequency of NSSI (from 0 to the maximum number of NSSI). NSSI were conceptualized as self-injurious behaviours that were not intended to be an attempt to end one’s life. The number of NSSI in the year prior to the initial assessment was assessed through an open question: Have you ever caused yourself any self-directed and deliberate injuries, such as cutting, hitting, scratching etc., with no suicidal intent? (yes/no). How many times/days have you caused yourself such injuries in the past year? Suicidal Attempts were conceptualized as self-inflicted, potentially injurious behaviours with a non-fatal outcome, but with evidence of the intention to die (45). The number of suicide behaviours in the past year was assessed with the following question, created ad hoc for this research: Have you ever tried to end your life? (yes/no). How many times have you attempted suicide in the past year? The responses related to methods were categorized by clinical psychologists. Moreover, we collected the frequency of inpatient hospitalizations in the past year.
Quality of life index (QLI) (46). QoL was assessed by the Spanish version (47) of the QLI, which consists of 10 items that can be rated from 0 (poor) to 10 (excellent). Each item represents one relevant dimension of QoL: Physical Well-being, Psychological/Emotional Well-being, Self-care and Independent Functioning, Occupational Functioning, Interpersonal Functioning, Social Emotional Support, Community and Services Support, Personal Fulfilment, Spiritual Fulfilment, and Overall Quality of Life. Each dimension contains a brief explanation in parentheses, designed to allow flexibility in the interpretation depending on the individual’s cultural and experiential background (48). The final score ranges from 1 to 10 and is obtained by calculating the average of the scores on all the items. The translation of this scale has shown satisfactory test-retest reliability (Cronbach's α = 0.89) and validity properties, and in our sample, the reliability was adequate (α = 0.88).
Resilience Scale (RS-15) (49) Resilience was measured by the short form of the Spanish version of the Resilience Scale (50), which was originally developed by Wagnild and Young (51). The Spanish version replicated the bi-factorial structure of the original scale (“personal competence” and “acceptance of self and life”). Cronbach’s α for the total scale was .93 for the general population and .94 for patients with eating disorders. The RS-15 is a 15-item self-report measure of resilience, defined by the authors as a positive personality trait that promotes adaptability amid adversity. Each item is rated by the participant using a 7-point Likert scale ranging from disagree (1) to agree (7). Possible scores range from 15 to 105, with higher scores indicating higher perceived resilience. Psychometric properties and internal consistency are adequate and similar to the original scale (49). In our sample, the reliability was adequate (α = 0.93).
Beck Depression Inventory-II (BDI-II) (52). This inventory consists of 21 items with four response options (0 – 4) that rate depressive symptomatology. Its Spanish version offers good psychometric properties (53). It has presented adequate reliability (Cronbach's α = 0.90) in Spanish participants. In our sample, the reliability was adequate (α = 0.91).
The clinical sample was collected from treatment-seeking patients in three Specialized services in personality disorders in Spain between 2011 and 2018. All participants were informed about the study and gave their written informed consent. Several expert clinical psychologists with more than 10 years of experience with BPD treatment conducted the assessment to ensure that patients met the inclusion criteria. The participants received one of these possible treatments, depending on their clinical situation and the study underway in the clinical centres at the time of recruitment: DBT (54), STEPPS (55), or TAU-CBT. DBT is a treatment with broad empirical support for BPD (56). It draws on dialectical tensions of the behaviours, which can be functional and dysfunctional at the same time, and it targets a balance between acceptance and change (57). It consists of a combination of individual psychotherapy, group skills training, telephone coaching, and a therapist consultation team (58). STEPPS is a cognitive behavioural model in group format that incorporates skills training and creates a common language for the patients and their system (55). In this treatment, maladaptive schemas are confronted, and BPD is reframed as an emotional intensity disorder (59). TAU-CBT is the standard treatment protocol in clinical centres; namely, it is a cognitive behavioural program focused on treating the symptomatology by using CBT strategies. In our study, it consisted of one hour of individual therapy – in which personality psychopathology symptoms were also addressed – and one two-hour group session per week. The TAU-CBT group format was adapted by the clinical team and included psychoeducation, cognitive restructuring, and consolidation of achievements. All these programs lasted about 6 months, and patients completed self-reports on resilience and QoL both before and after the treatment. All the psychotherapeutic groups received a similar number of psychotherapy sessions.
For the nonclinical sample, a convenience sample of university students was recruited. The samples were matched on age. They completed the questionnaires during their normal day at the university.
Means and standard deviations and zero-order correlations were calculated for all the variables at baseline (T1) and at post-treatment (T2). First, to calculate whether there were differences in QoL between the non-clinical population and the participants with a diagnosis of BPD, a t-test was performed. Second, to examine whether there were statistically significant differences between DBT, STEPPS, and TAU in the improvement in QoL after treatment, a MANOVA was performed, and the effect sizes were calculated (Cohen´s d). Moreover, we analysed whether the treatments produced a clinically significant change in QoL Three conditions are necessary to consider a change clinically significant (28): a) the Reliable change index (RCI) is calculated; RCI is a methodology that indicates whether the change detected after a treatment represents a real modification in the patient’s clinical condition, or if it simply reflects a measurement error surrounding test-retest difference scores. If there is a reliable change, the assumption can be made that changes are due to treatment and not to a measurement error; b) after treatment, the QoL scores should be situated in the mean range of the normal population (+/- SD) to interpret the functional direction (60); and c) the effect size is calculated using Hedge´s g with the range of a normal population to confirm that there are no statistically significant differences in QoL after the treatment (61).
Finally, we performed two linear regression analyses. In the first model, we took Resilience before treatment (RS T1) as the predictor variable and QoL pre-treatment (QoL T1) as the dependent variable. In the second linear regression analysis, we took Resilience after treatment (RS T2) as the predictor variable and QoL post-treatment (QoL T2) as the dependent variable. In the two prediction models, Type of psychotherapy, Gender, Age, and depression (BDI-II) were controlled. Potential multicollinearity between prediction variables was rejected due to tolerance values and a variance inflation factor between 0.9 and 1.3, respectively, which meet good statistical criteria (62). Data were analysed using SPSS 24 (63).