Received: 22 September 2021 | Accepted: 23 January 2022 DOI: 10.1111/acps.13403 ORIGINAL ARTICLE “Skills for pills”: The dialectical-­behavioural therapy skills training reduces polypharmacy in borderline personality disorder Joaquim Soler1,2,3 | Elisabet Casellas-­Pujol1,2,3 | Isabel Fernández-­Felipe4 | Ana Martín-­Blanco1,2,3 | David Almenta1,2,3 | Juan C. Pascual1,2,3 1 Department of Psychiatry, Hospital de la Santa Creu i Sant Pau, Barcelona, Abstract Spain Objective: Polypharmacy and overprescription of off-­label medications are 2 Universitat Autònoma de Barcelona common in patients with borderline personality disorder (BPD). The aim of the (UAB), Barcelona, Spain 3 present naturalistic study was to explore whether the skills training module of Institut d’Investigació Biomèdica-­ Sant Pau (IIB-­SANT PAU), Centro de dialectical-­behavioural therapy (DBT) can reduce polypharmacy in these patients Investigación Biomédica en Red de in routine clinical practice. Salud Mental, CIBERSAM, Barcelona, Methods: Retrospective, observational study of 377 patients with a primary Spain 4 diagnosis of BPD consecutively admitted to the BPD outpatient unit from 2010 Labpsitec, Laboratorio de Psicología y Tecnología. Dpto. Psicología Básica, through 2020. All patients were invited to participate in the DBT skills training Clínica y Psicobiología, Universitat module (DBT-­ST). DBT-­ST participants (n = 182) were compared with a con- Jaume I, Castelló, Spain trol group who did not participate in DBT-­ST (n = 195). Pre-­post intervention Correspondence changes in medication load and use of antidepressants, benzodiazepines, mood Juan C. Pascual., Department of stabilizers, and antipsychotics were evaluated. Psychiatry, Hospital de la Santa Creu i Sant Pau, Av. Sant Antoni Mª Claret Results: At baseline, most patients (84.4%) were taking at least one medication 167, 08025 Barcelona. Spain. and 46.9% were on polypharmacy. Compared to controls, patients in the DBT-­ST Email: jpascual@santpau.cat group presented a significant reduction in the number of medications (2.67–­1.95 Funding information vs. 2.16–­2.19; p < 0.001), medication load (4.25–­3.05 vs. 3.45–­3.48; p < 0.001), Centro de Investigación Biomédica en use of benzodiazepines (54.4%–­27.5% vs. 40%–­40.5%; p < 0.001), mood stabiliz- Red de Salud Mental ers (43.4%–­33% vs. 36.4%–­39.5%; p < 0.001), and antipsychotics (36.3%–­29.1% vs. 34.4%–­36.9%; p < 0.001). Conclusions: These findings suggest that patients with BPD can benefit from the DBT-­ST module, which may reduce the medication load, particularly of sedatives. The results suggest that DBT-­ST may be useful to treat overmedication in patients with BPD and could help to promote “deprescription” in clinical practice. KEYWORDS borderline personality disorder, deprescription, dialectical-­behavioural therapy, polypharmacy, psychopharmacology treatment Joaquim Soler and Elisabet Casellas-­Pujol contributed equally to this work. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2022 The Authors. Acta Psychiatrica Scandinavica published by John Wiley & Sons Ltd. 332 | wileyonlinelibrary.com/journal/acps  Acta Psychiatr Scand. 2022;145:332–342. SOLER et al.    | 333 1 | I N T RO DU CT ION Significant outcomes Borderline personality disorder (BPD) is a mental disorder • Most patients were on pharmacotherapy and characterized by a persistent pattern of emotional dysreg- polypharmacy was common. ulation, self-­image and interpersonal relationship insta- • Patients who participated in the skills training bility, and marked impulsivity. It is a severe but common module of dialectical-­behavioural therapy pre- disorder, with an estimated prevalence ranging from 1.6% sented a significant reduction in the number in the general population to 20% in psychiatric inpatient of medications, medication load, and use of populations.1,2 BPD is associated with severe psychoso- sedatives. cial and occupational dysfunction, intensive use of men- • Dialectical-­behavioural therapy skills training tal health resources, and carries a substantial economic may be an effective strategy in helping to re- burden.3,4 duce polypharmacy and could promote depre- All of the main clinical guidelines support psychother- scription in patients with BPD. apy as the treatment of choice for BPD (for an overview see5–­7). Although various psychological treatments have Limitations proven to be efficacious treatments for BPD, dialectical behavioural therapy (DBT) is the most frequently studied • Because of the non-­randomized, retrospective treatment, supported by robust empirical evidence from study design, there is a risk of selection bias. Cochrane reviews and meta-­analyses.6,7 However, access • The observed decrease in the medication load to psychological therapies such as DBT is frequently lim- after skills training does not necessarily imply ited, which is why most individuals with BPD only receive better clinical and/or functional outcomes. pharmacological treatment, even though no drugs have been specifically approved for the treatment of BPD and some clinical guidelines specifically recommend avoid- Several strategies have been developed to reduce poly- ing psychotropic drugs to treat BPD.8–­10 While psychotro- pharmacy and promote quality pharmacological treat- pic drugs are commonly used to treat Axis I comorbidity ment, including efforts to familiarize physicians with disorders, they are also used to reduce the characteristic the concept of "deprescribing" (defined as the process of symptoms in patients with BPD such as intense mood reducing medications toward the minimum helpful treat- instability, anger, and/or impulsive behaviour.7,11–­13 ment), to emphasize the need to prescribe drugs only on However, the optimal therapeutic drug for BPD-­specific an "as needed" basis, and to establish a support group for symptoms is controversial, in part because of the scant re- physicians to avoid case overload.16 search on pharmacological management in patients with One approach that could be useful in deprescribing BPD but also to the lack of consensus pharmacological drugs in patients with BPD is psychotherapy. One of the recommendations in clinical guidelines for BPD.5,7,12 most effective psychotherapeutic treatments for BPD is Several studies have shown that more than 80% of DBT to develop cognitive-­behavioural principles to replace patients with BPD in western countries receive psycho- maladaptive behaviours with healthier coping skills.18 tropic drugs and more than 50% of patients receive poly- DBT is an outpatient treatment supported by empirical ev- pharmacy.11–­15 In a study conducted by our group, nearly idence in BPD populations with other comorbid disorders three quarters of patients with BPD were taking antide- such as post-­traumatic stress disorder, eating disorders, and pressants, 55% benzodiazepines, 40% mood stabilizers, substance abuse disorders (DBT-­S).19 Given that DBT is a and 35% antipsychotics.15 Some authors have described multicomponent therapy, that is both expensive and time the risks of medication use in BPD, which include in- consuming, several studies have sought to identify the most teractions with substances of abuse, addiction potential, relevant components of this therapy to reduce costs and increased risk of suicidal tendencies and disinhibition, se- treatment time. For example, the DBT skills training mod- rious side effects, a high risk for self-­injury with pills, and ule (DBT-­ST) alone has proven to be effective in BPD.20,21 psychological risks such as diverting attention and energy DBT skills training comprises four modules focused on from psychotherapeutic aims, thus altering the effective- mindfulness, distress tolerance, interpersonal effective- ness of psychotherapy.16 The reasons for overprescription ness, and emotion regulation.18,20 Importantly, offering of psychotropic drugs in this population include: lack of DBT to patients with BPD does not rule out the concur- perceived clinical alternatives; the prescription of drugs to rent use of medications, which should be considered po- address a temporary crisis that are never withdrawn (thus tential adjuncts to psychological therapy.18,19 Nonetheless, becoming chronic); attempts to treat all symptoms with DBT has been shown to promote the "skills for pills" ef- medications; and treatment of comorbid disorders.12,16,17 fect, whose aim is to reduce the use of problematic drugs 334 |    SOLER et al. such as benzodiazepines, especially DBT-­S in patients with specific psychotherapeutic intervention for all individu- substance abuse disorders.19 The main idea is to taper the als, groupal DBT skills training. medication while simultaneously increasing the use and proficiency of the trained skills. In this way, physicians can encourage patients to learn and practice these skills, thus 2.2 | Study design allowing them to reduce their use of prescription medi- cations, when appropriate. Consequently, to ensure that This was a naturalistic, retrospective cohort study. All pharmacological and psychological treatment are tightly participants underwent a clinical interview, and two coordinated, the clinician prescriber should ideally work semistructured diagnostic interviews were conducted by closely with the psychology treatment team, which would an experienced psychiatrist and a clinical psychologist. also improve continuity of care. Admission to the BPD unit required a confirmed BPD diagnosis with the validated Spanish language versions of the Structured Clinical Interview for DSM-­IV Axis 1.1 | Aims of the study II Disorder (SCID-­II)22,23 and the Diagnostic Interview Revised for Borderlines (DIB-­R).24,25 Inclusion criteria The aim of the present naturalistic study was to explore, in were as follows: (a) age between 18 and 55 years; (b) pri- a naturalistic clinical setting, whether DBT-­ST reduces the mary diagnosis of BPD (DSM-­IV criteria) confirmed by use of medications and/or polypharmacy in patients with structured interviews (SCID-­II and DIB-­R); (c) absence BPD. We hypothesized that providing patients with train- of comorbid psychotic or bipolar disorders; and (d) no ing in specific skills such as mindfulness, emotion regula- neurological disease, intellectual disability, or any severe tion, and distress tolerance would reduce or eliminate the physical condition that could affect the psychotherapeutic need for problematic drugs such as benzodiazepines. intervention. All study data were obtained from patient medical re- cords, which included data relative to admission to the 2 | M AT E R IAL AN D M ET H OD S BPD unit (time 1) and after completion of the psychother- apeutic intervention and follow-­up in the BPD program 2.1 | Participants (time 2). All patients included in this historical cohort in the BPD program were invited to participate in the Data were retrospectively collected from 377 patients di- DBT-­ST intervention. The cohort was divided into two agnosed with BPD and admitted to the outpatient BPD groups: those who agreed to participate in the DBT-­ST in- unit at the Department of Psychiatry at the Hospital tervention (DBT-­ST group; n = 182) and a control group de la Santa Creu i Sant Pau, between January 2010 and consisting of those who did not participate in the inter- December 2020. This outpatient program is part of Spain's vention (control group; n = 195). Therefore, this was a Public National Mental Health Service and provides spe- non-­randomized, observational study. cialized care for people with BPD referred from other psy- In parallel with the psychotherapeutic intervention, chiatric clinical units (eg, psychiatric emergency units, all patients received periodic psychiatric evaluation and acute hospitalization units, general mental health outpa- follow-­up visits, including supervision of pharmacolog- tient services, private mental health centers, among oth- ical treatment to prevent excessive use of medications. ers). A high percentage of these patients had previously Patients did not receive any other psychotherapeutic inter- received pharmacological and psychological treatment ventions. The psychiatrists who performed these periodic in other clinical units. Because of the lack of human and evaluations were not affiliated with DBT psychotherapy. economic resources, general mental health public services The follow-­up period in the BPD program was variable in Spain have some limitations of offering a complete as- and depended on the following factors: duration of ther- sistance to individuals with BPD. Compared with general apy, whether or not the patient repeated the psychological mental health center, the BPD Unit offers: reliable con- intervention (DBT-­ST), and whether the patient dropped firmation of BPD diagnosis with validated instruments, out during follow-­up. greater accessibility to the unit, emergency attention in The following sociodemographic and clinical data were crisis, higher frequency and duration of visits, therapeu- collected: age; sex; follow-­up period (months); presence of tic team with specific experience and sensitivity for BPD, lifetime comorbid axis I disorders (yes/no) classified into family care, psychoeducation of disorder, general man- four groups: affective, anxiety, substance use, and eating agement and non-­harmful strategies, and, finally, supervi- disorders; DIB-­R total and subscale scores (affect [DIB-­ sion of pharmacological treatment avoiding the excessive aff]; cognition [DIB-­cog]; impulsive action patterns [DIB-­ use of medication. Moreover, our BPD unit offers, as imp]; and interpersonal relationships [DIB-­per]). SOLER et al.    | 335 Data on pharmacological treatment were obtained and high = 2. For antidepressants and mood stabilizers, the classified as follows: antidepressants, including selective se- doses and types were coded according to the method pro- rotonin reuptake inhibitors (SSRIs), tricyclic antidepressants posed by Sackeim.27 Antipsychotic doses were converted and dual-­acting agents; benzodiazepines; mood stabilizers; into chlorpromazine dose equivalents and coded using the and first and second generation antipsychotics (FGA and mean effective daily dose as reference.28 The lorazepam SGA, respectively). For these drugs, we determined whether dose was also coded to the reference of the midpoint of the the participant was taking a given medication (yes/no) and recommended daily dose range from the Physician's Desk the total number of medications (all categories) per patient. Reference.29 Scores for each of the individual medication The study adhered to the principles outlined in the were summed to create a total medication load index. The Declaration of Helsinki and was approved by the Clinical composite total medication load index seeks to reflect the Research Ethics Committee at the Hospital de la Santa dose and variety of the different medications taken. The Creu i Sant Pau. Written consent to participate in the study index score is the sum of all individual medication codes was not considered necessary as all data were collected in each medication category for each participant. We also retrospectively from routine admission data and subse- calculated the sedative load index, which includes all quently anonymized. We checked the medical records of medications that contribute to sedative load: benzodiaz- all patients to verify that no written objection to the use of epines, antipsychotics, and drugs with a sedative com- this information was included in the record. ponent or side effect, including pregabalin, gabapentin, lamotrigine, oxcarbazepine, carbamazepine, topiramate, and mirtazapine. Only those medications that were regu- 2.3 | Instruments larly used by the patients were considered when compar- ing the indices at each time point. The following instruments were administered to all participants. 2.4 | Psychotherapeutic intervention 2.3.1 | Structured Clinical Interview for 2.4.1 | Dialectical-­behavioural therapy DSM-­IV Axis II Disorder (SCID-­II) skills training The SCID-­II is a semistructured interview designed to as- The DBT skills training intervention is an adaptation of sess DSM-­IV personality disorders. The Spanish valida- the DBT format drawn from one of the four intervention tion study showed that this instrument discriminates well modes of the standard version.18,20,21 DBT-­ST consisted between personality disorders and has good inter-­rater of weekly skills training sessions (120 min each) held reliability.22,23 over a 6-­month period. Upon completion of this 6-­month program, patients were invited to repeat the program to further reinforce the skills learned. All training sessions 2.3.2 | Diagnostic interview revised for were conducted by two experienced psychotherapists, borderlines (DIB-­R) each with more than 10 years of clinical experience and specific training in DBT (Behavioral Tech Inc.). The treat- The DIB-­R is an instrument designed to diagnose BPD ment groups consisted of 9–­12 participants. None of the and to assess the severity of the disorder within the last participants received any other type of individual or group 2 years. The Spanish version has demonstrated good inter- psychotherapy during the study period. nal consistency (Cronbach's alpha, 0.89; sensitivity, 0.81; Dialectical-­behavioural therapy skills training aims to and specificity, 0.94).24,25 promote behavioural change, to help participants learn how to be interpersonally effective, and to learn how to better regulate emotions, and foster acceptance, mindful- 2.3.3 | Medication load index ness, and distress tolerance. The group sessions were all structured in a similar manner: teaching the content, in-­ This index measures the total medication load. Following session practice of the strategies, and weekly homework the method proposed by Hassel et al.,26 the total medi- assignments. The skills taught in a given week were re- cation load index was calculated by coding the dose of viewed at the following session. each antidepressant, mood stabilizer, antipsychotic, and The DBT skills training consists of four modules, as anxiolytic medication as follows: absent = 0, low = 1, or follows: 336 |    SOLER et al. Mindfulness 2.5 | Statistical analysis This training module aims to help participants to develop non-­judgmental awareness, attentional control, and sense Data were analyzed with the IBM-­SPSS Statistics for of self. Other aspects taught in this module include ob- Windows, v. 25.0. All data were screened for skewness serving and describing thoughts, emotions, bodily sen- and kurtosis to test assumptions of normality. All hypoth- sations, and events; and fully engaging in experiences eses were tested to a two-­sided significance level of 0.05. without judgment, focusing on one thing at a time, and First, patient demographic and clinical characteristics being effective. at baseline (time 1) were described using typical measures of frequency, central tendency, and dispersion. To com- Distress tolerance pare these characteristics between groups (DBT-­ST vs. This module involves training participants to perceive control group), chi-­square tests (or Fisher's exact test if the environment without demanding that it be different. expected frequencies were <5) were used for categorical Participants learn to experience and accept painful emo- variables, and t-­tests for independent samples were used tions without trying to change them. This module focuses for continuous variables. on teaching strategies to help manage crises and skills to Second, to determine the impact of the group allocation accept reality. on the use of medication, the percentage of withdrawal of each class of drug (antidepressants, benzodiazepines, Emotion regulation mood stabilizers, and antipsychotics) at time 2 was com- In this module, participants learn skills to help identify, pared between both groups (DBT-­ST vs. control group) label, and describe emotions, and to experience emo- using chi-­square tests (or Fisher's exact tests if expected tions from a mindful attitude to decrease vulnerability frequencies were <5). to unpleasant emotions, increase pleasant emotions. Finally, to evaluate the impact of treatment on pre-­ Participants learn to act in the opposite way when emo- post differences in the mean number of drugs taken per tions are not justified. patient, and in the medication and sedative load indices, we conducted multivariate repeated-­measures ANOVAs. Interpersonal effectiveness Treatment effects were assessed by entering each variable This module contains strategies for solving problems at as dependent variables; time (pre-­ and post-­treatment) was the interpersonal level. The focus is on teaching partici- entered as a within-­subjects factor, and group condition pants how to reach personal objectives while maintaining (DBT-­ST and control group) was entered as a between-­ relationships and self-­respect. Participants also learn how subjects factor. to ask for what they need and on knowing how to say no Post hoc analyses were carried out when significant in- when appropriate. teractions were found. Effect sizes are reported by partial eta squared, with values up to 0.01, 0.06, and 0.14 consid- Control group ered small, moderate, and large, respectively.31 This group included all patients that did not participate in the DBT-­ST intervention for any reason, which included any of the following: scheduling conflicts (n = 48 [24.6%], 3 | RESULTS mainly for academic reasons); work-­related issues (n = 53, 27.2%); preference for other types of therapy (n = 13, 3.1 | Sample characteristics 6.7%); preference for individual therapy (n = 66, 33.8%); preference for private psychotherapy (n = 10, 5.1%); and The final sample consisted of 377 patients with a primary others (n = 5, 2.6%). diagnosis of BPD. Table 1 summarizes the sociodemo- Although these individuals did not receive any specific graphic and clinical characteristics of the participants. psychotherapeutic intervention for BPD, compared with The typical patient profile was a 30-­year-­old female with general mental health services, they valued the higher fre- moderate clinical severity (based on DIB-­R total score). quency of psychiatric visits, attention in crisis, family care, Most patients presented at least one lifetime comorbid and greater experience and sensitivity in the management Axis I disorder, most commonly substance use and eating of BPD. This follow-­up visits also include supervision of disorders. pharmacological treatment avoiding, if possible, the ex- At admission to the BPD program, 84.4% of the patients cessive use of medications, as recommended by all clini- were taking at least one medication. The mean number cal guidelines. They also received non-­harmful strategies of medications was 2.4 (range, 0–­8), and nearly half of based on the Handbook of Good Psychiatric Management the patients were receiving polypharmacy (≥3 medica- for Borderline Personality Disorder.30 tions). Nearly three quarters of the sample (71.9%) were SOLER et al.    | 337 TABLE 1 Baseline demographic and clinical characteristics of the sample with differences between groups Total sample DBT-­ST Control group Variables (377) (182) (195) χ2 t p Age, mean (SD) 30.51 (8.5) 30.92 (8.0) 30.14 (9.0) n. s Females, n (%) 336 (89.1%) 168 (92.3%) 168 (86.2%) n. s Married/stable couple, n (%) 140 (37.1%) 68 (37.4%) 72 (36.9%) n. s Employed, n (%) 138 (36.6%) 72 (39.6%) 66 (33.8%) n. s Comorbidities Axis I comorbidity 266 (70.6%) 133 (73.1%) 133 (68.2%) n. s Affective disorders 87 (23.1%) 47 (25.8%) 40 (20.5%) Anxiety disorders 54 (14.3%) 26 (14.3%) 28 (14.4%) Eating disorders 116 (30.8%) 61 (33.5%) 55 (28.2%) Substance use disorders 133 (35.3%) 62 (34.1%) 71 (36.4%) DIB-­R total score, mean (SD) 7.24 (1.2) 7.46 (1.2) 7.05 (1.2) −3.27 0.001 Pharmacological treatment Medications, mean (SD) 2.41 (1.7) 2.67 (1.7) 2.16 (1.7) −2.00 0.003 0 59 (15.6%) 18 (9.9%) 41 (21.0%) 1 67 (17.8%) 35 (19.2%) 32 (16.4%) 2 74 (19.6%) 32 (17.6%) 42 (21.5%) 3 85 (22.5%) 45 (24.7%) 40 (20.5%) 4 48 (12.7%) 25 (13.7%) 23 (11.8%) ≥5 44 (11.7%) 27 (14.8%) 17 (8.6%) Polypharmacy 177 (46.9%) 97 (53.3%) 80 (41%) 5.69 0.017 Antidepressants 271 (71.9%) 142 (78.0%) 129 (66.2%) 6.56 0.014 Benzodiazepines 177 (46.9%) 99 (54.4%) 78 (40.0%) 7.83 0.005 Mood stabilizers 150 (39.8%) 79 (43.4%) 71 (36.4%) n. s Antipsychotics 133 (35.3%) 66 (36.3%) 67 (34.4%) n. s Medication load, mean (SD) 3.83 (2.9) 4.25 (2.8) 3.45 (2.8) −2.75 0.006 Sedation load, mean (SD) 2.24 (2.1) 2.49 (2.1) 2.01 (2.0) −2.28 0.023 Abbreviations: DIB-­R, Revised Diagnostic Interview for Borderlines; n. s., not significant; SD, standard deviation. on antidepressants (mostly SSRIs), 47% on benzodiaze- using this variable as a covariate, without finding any in- pines, 40% on mood stabilizers, and 35% on antipsychotics fluence on the results. (mostly SGAs). Participants in the DBT-­ST group were taking, on av- erage, more medications than controls, with a higher proportion receiving polypharmacy, as well as higher 3.2 | Clinical and pharmacological medication and sedation load indices (Table 1). In addi- differences between the DBT-­ST group and tion, in the DBT-­ST group, a higher percentage of patients controls non-­DBT-­ST group were taking antidepressants and benzodiazepines. There were no significant between-­group differences in terms of sociodemographic characteristics or the preva- 3.3 | Differences between groups in lence of Axis I comorbidities. However, participants in the prescription changes pre-­post intervention DBT-­ST group had higher mean DIB-­R total scores (7.46 vs. 7.05), indicating greater clinical severity (Table 1). Changes in prescription patterns are shown in Table 2 There were also small but statistically significant dif- and Figures 1 and 2. Patients who participated in DBT-­ST ferences between the groups at the timeframe (time in intervention experienced a significant decrease in the months between time 1 and time 2): DBT-­ST 11.1 (SD: 5) number of medications prescribed [F(1,375) = 69.74, vs. Control group 9.77 (SD: 4.7; p = 0.007). We re-­analyzed p < 0.001] and in the medication and sedation load indices 338 |    SOLER et al. [F(1,375) =86.77, p < 0.001; F(1,375) = 127.56, p < 0.001, countries13,14 and higher than recommended in some respectively] over the course of the psychotherapeutic clinical guidelines (eg, United Kingdom and Australian intervention compared with control group. The effect of guidelines), which do not recommend prescribing phar- intervention for all variables was large. Participants in macological agents to treat BPD symptoms, but only to the DBT-­ST group also significantly reduced their use of treat comorbid disorders or for short term use during pe- benzodiazepines (from 54.4% to 27.5%), mood stabilizers riods of crisis.7–­10 Although clinical guidelines in other (from 43.4% to 33%), and antipsychotics (from 36.3% to countries (Netherlands, Germany, and the United States) 29.1%). differ and have a more symptom-­oriented approach, all In the control group, no decrease in medication use guidelines seem to agree that pharmacotherapy should was observed in any class of drugs. Moreover, for some not be used as a first-­line strategy because of the risks as- medications, small increases were observed. The most no- sociated with these drugs.32 table finding was a significant decrease in the prescription Interestingly, the patients who agreed to participate in of benzodiazepines, mood stabilizers, and antipsychotics the DBT-­ST intervention had higher DIB-­R total scores in the DBT-­ST group versus controls. (indicating greater severity) and also took more prescrip- tion drugs than controls. Given the study design (obser- vational, non-­randomized study), we hypothesize that 4 | DI S C USSION patients with more severe BPD symptoms may be more motivated (ie, negatively reinforced) to change, and thus The findings of the present study confirm previous re- more likely to agree to participate in a skills training pro- ports that a high proportion of patients with BPD treated gram to learn how to better manage their condition. This in clinical settings receive pharmacotherapy and poly- increased motivation to change also points to a greater pharmacy. However, patients who received DBT skills interest in reducing pharmacological treatment. In this training achieved a significant pre-­post reduction in the regard, findings from other studies have shown that indi- mean number of medications and in the medication and viduals with greater levels of pretreatment distress were sedation load indices. These results confirm our main hy- less likely to drop out of treatment, benefit more from pothesis that DBT skills training would help to reduce the DBT, and have better outcomes.33 In turn, this suggests medication burden. Compared to controls, patients who that patients with less severe BPD may less motivated than received DBT-­ST also significantly reduced their use of those with greater severity to enroll in a weekly DBT pro- sedatives (benzodiazepines and antipsychotics). gram, which requires a strong commitment and high level In line with previous studies, we found that most pa- of motivation.34 tients with BPD were taking at least one medication, with The findings of the current study suggest that DBT-­ST nearly half of those patients receiving polypharmacy. This leads to deprescription in these patients, thus reducing the prescription practice in the public mental health system in medication load, as we had hypothesized. The deprescrip- Spain is consistent with clinical practice in other western tion effect observed in patients who attended the DBT-­ST T A B L E 2 Between-­group differences DBT-­ST (182) Control group (195) p value in changes in prescriptions pre-­post Pre Post Pre Post intervention Number of medications, 2.66 (1.7) 1.95 (1.3) 2.16 (1.7) 2.19 (1.6) <0.001 mean (SD) Medication Load, mean 4.23 (2.8) 3.05 (2.2) 3.46 (2.8) 3.48 (2.7) <0.001 (SD) Sedation Load, mean 2.48 (2.1) 1.38 (1.6) 2.0 (2.0) 2.04 (2.0) <0.001 (SD) Percentage of patients With medication 90.1 89.6 79 79.5 n. s Antidepressants 78 80.8 66.2 70.8 n. s Benzodiazepines 54.4 27.5 40.0 40.5 <0.001 Mood stabilizers 43.4 33.0 36.4 39.5 <0.001 Antipsychotics 36.3 29.1 34.4 36.9 <0.001 Abbreviations: DBT-­ST, dialectical-­behavioural therapy skills training; n.s., not significant; SD, standard deviation. SOLER et al.    | 339 F I G U R E 1 Differences between groups in prescription changes pre-­post intervention F I G U R E 2 Pre-­post intervention differences in the prescription of antidepressants, benzodiazepines, mood stabilizers, and antipsychotics sessions is consistent with previous studies showing that of DBT-­ST is to help patients to build a new behavioural DBT-­ST is an important component of the more compre- repertoire, which should enhance emotion regulation hensive DBT program.20,21 Those studies demonstrated the and functioning. The strategies taught in the four DBT-­ST relevant role of this DBT modality in mediating reductions modules allow patients to better understand and regulate in suicide, depression, anger, and—­at least partially—­self-­ their emotions, better tolerate suffering, to more effec- harm, as well as in improving variables associated with tively respond on an interpersonal basis, and to participate psychopathology, mood, and affect. Those same studies mindfully and non-­judgmentally in life. The strategies have also shown that group DBT-­ST may even be more ef- taught in DBT-­ST allow patients to better understand their fective than individual DBT therapy alone.20,21 behaviours and thus to more effectively manage crises, by Dialectical-­behavioural therapy is based on a bioso- reducing emotional vulnerability and fostering the capac- cial model in which a skill deficit leads to the behavioural ity to identify different emotional states to deploy respond problems associated with BPD, such as parasuicide or ag- more effectively to those emotional states. Skills training gressive behaviours, which are often a response to emotion may empower the patients and increase self-­efficacy, thus dysregulation.18 Accordingly, one of the main objectives consequently decreasing the number of medications, 340 |    SOLER et al. although this was not a goal per se of the intervention, but The prescription of mood stabilizers and antipsychot- rather a consequence. ics, especially SGAs, for patients with BPD, has increased Although we observed an overall trend toward depre- significantly in the last two decades,15 likely because of scription for some medications, this was not true for all the promising results reported in clinical trials (especially prescription medications. For example, in both groups, for olanzapine, quetiapine, topiramate, and lamotrigine) antidepressant prescriptions were either unchanged or in the treatment of affect dysregulation and impulse con- increased slightly, perhaps because of the prescribing trol.7 Antipsychotics have a sedative profile and could be clinicians' confidence in the effectiveness of antidepres- an alternative to benzodiazepines. In any case, the appli- sants, leading them to exclude these from the deprescrip- cation of DBT-­ST in patients with BPD would likely re- tion goal. The preference to reduce sedative drugs may be duce the need for these drugs. explained because DBT skills training increases tolerance Dialectical-­behavioural therapy skills training could be to distress, promotes coping and exposition, leading to ha- a highly effective strategy to improve emotion regulation bituation, extinction, or a change in the emotional experi- in patients with BPD. Unfortunately, DBT is not always ence, thus reducing dependence on medication as a main available in the public health system, in part because it strategy for emotion regulation. is a resource-­intensive treatment, requiring specific clini- In some contexts, the use of certain medications (no- cal training, staffing, and time. Consequently, it is essen- tably benzodiazepines and antipsychotics) allows the pa- tial to consider other efficient options that could reduce tient to avoid or escape from negative emotional states. pharmacological strategies. In this regard, the application Thus, when effective alternative strategies are available of the principles recommended in the Good Psychiatric to achieve the same objective, this consequently leads Management model30 could help to reduce the medication to a reduction in medication intake. Not surprisingly, load and provide alternative tools to help BPD patients no changes in prescription patterns were observed in cope with distress. the control group, perhaps because these patients—­who This study has several limitations. First, because of obviously did not receive DBT-­ST—­may believe that the non-­randomized, retrospective study design, there is symptom-­targeted pharmacological treatment is the only a risk of selection bias. Although clinical guidelines rec- viable strategy to manage BPD and comorbid distress. ommend avoiding polypharmacy and unnecessary drugs Maybe, these patients were not looking for new strate- in individuals with BPD, there may have been a more ac- gies or skills and they preferred a complete assistance tive deprescription activity in the DBT-­ST group compared in the BPD unit with greater accessibility, attention in to controls. It is also possible that there exist differences crisis, higher frequency and duration of visits, and fam- between groups regarding the motivation to reduce med- ily care. Indeed, it is also possible that non-­participation ication on both individuals and psychiatrists. Therefore, in DBT skills training influenced psychiatrists' practice these findings will need to be confirmed in a prospective, because they were less motivated to reduce medication double-­blind randomized clinical trial. Another potential since patients did not have alternative tools to manage limitation is that the observed decrease in the medication difficulties. burden does not necessarily imply better functional out- The most significant change observed in the study comes. Consequently, other variables need to be included group was the marked reduction in the prescription of and evaluated in future studies to confirm that the reduced benzodiazepines, which decreased from 54.4% of patients medication burden is because of clinical improvement at baseline to 27.5% post-­intervention. All major clinical and not because of other factors. Finally, uncontrolled guidelines advise against the use of benzodiazepines in clinical factors could also have influenced the results. By BPD (except for short-­term use to address a specific, tem- contrast, the main strength of this study is the naturalistic porary crisis).8–­10 Notwithstanding the clear recommenda- study design, which allowed us to analyze data obtained tions against benzodiazepines, prescription rates for these in the course of routine clinical practice, in contrast to pharmaceuticals remain high. Because of the addictive other studies performed outside of real-­life clinical prac- potential of these medications and their risk of increasing tice (eg, clinical trials). This design provides clinically rel- suicidal tendencies and disinhibition, an important goal evant information since it was carried out in the context of in the psychiatric management of BPD is to reduce their usual psychiatric care in the Spanish public health system. use. These drugs are often prescribed in psychiatric emer- Finally, the results of this study support the value of a co- gency rooms to manage exacerbations or crises, as they ordinated and synergistic approach between psychophar- can rapidly alleviate anger and/or negative emotions. In macologic and psychotherapeutic interventions in BPD. this regard, if patients received DBT-­based skills training In conclusion, the findings of this study suggest that to improve emotion regulation and distress tolerance, the DBT skills training appears to be an effective strategy to use of these drugs would likely be reduced. reduce polypharmacy in patients with BPD. Providing SOLER et al.    | 341 4. Alvarez-­Tomás I, Soler J, Bados A, et al. Long-­term course of patients with tools to effectively manage distress could borderline personality disorder: a prospective 10-­year follow-­up reduce the need for these patients to resort to the use of study. J Pers Disord. 2017;31(5):590-­605. pharmaceuticals as their main approach to reducing dis- 5. Riffer R, Farkas M, Streibl L, Kaiser E, Sprung M. tress. Further research into the role of deprescription in Psychopharmacological treatment of patients with borderline the context of skills training would be invaluable to better personality disorder: comparing data from routine clinical care understand whether the decrease in medication load is with recommended guidelines. Int J Psychiatry Clin Pract. followed by clinical and functional improvement. 2019;23(3):178-­188. 6. Stoffers-­Winterling JM, Völlm BA, Rücker G, Timmer A, Huband N, Lieb K. Psychological therapies for people with ACKNOWLEDGMENTS borderline personality disorder. Cochrane Database Syst Rev. This study was supported by Centro de Investigación 2012;8:CD005652. Accessed 21 September 2021. Biomédica en Red de Salud Mental (CIBERSAM). The au- 7. Simonsen S, Bateman A, Bohus M, et al. European guidelines thors would like to thank Bradley Londres for professional for personality disorders: past, present and future. Borderline English language editing. Pers Disord Emot Dysregulation. 2019;6(9):1-­10. 8. National Institute for Health and Clinical Excellence. Borderline Personality Disorder: Treatment and Management. CONFLICT OF INTEREST NICE Clinical Guideline 78 Summary. National Institute for The authors declare that they have no conflicts of interest. Health and Clinical Excellence; 2009. 9. National Health and Medical Research Council. Clinical AUTHOR CONTRIBUTIONS Practice Guideline for the Management of Borderline JS and JCP conceived the study. AMB and JCP performed Personality Disorder. Melbourne: National Health and Medical the statistical analyses. ECP and IFF drafted the first ver- Research Council; 2012. sion of the manuscript. JS, ECP, and DA performed the 10. GPC. Grupo de trabajo de la guía de práctica clínica sobre psychotherapeutic intervention. All authors contributed trastorno límite de la personalidad Fórum de Salud Mental y AIAQS, coordinadores. Guía de práctica clínica sobre trastorno to the writing and reviewing of the manuscript. límite de la personalidad. Barcelona, Spain: Agència d’Infor- mació, Avaluació i Qualitat en Salut. Servei Català de la Salut. PEER REVIEW Pla Director de Salut Mental i Addiccions. Departament de The peer review history for this article is available at Salut. Generalitat de Catalunya; 2011. https://publo​ns.com/publo​n/10.1111/acps.13403. 11. Pascual JC, Martín-­Blanco A, Soler J, et al. A naturalistic study of changes in pharmacological prescription for borderline per- DATA AVAILABILITY STATEMENT sonality disorder in clinical practice: from APA to NICE guide- The data that support the present study are available from lines. Int Clin Psychopharmacol. 2010;25(6):349-­355. 12. Martín-­Blanco A, Ancoechea A, Soler J, et al. Changes over the corresponding author upon reasonable request. the last 15 years on psychopharmacological management of subjects with Borderline Personality Disorder. Acta Psychiatr ORCID Scand. 2017;136(3):323-­331. Joaquim Soler https://orcid.org/0000-0001-8077-3641 13. Bridler R, Häberle A, Müller ST, et al. Psychopharmacological Elisabet Casellas-­Pujol https://orcid. treatment of 2195 in-­patients with borderline personality dis- org/0000-0001-8496-0719 order: a comparison with other psychiatric disorders. Eur Ana Martín-­Blanco https://orcid. Neuropsychopharmacol. 2015;25(6):763-­772. 14. Paton C, Crawford MJ, Bhatti SF, Patel M, Barnes TR. The use org/0000-0002-3917-0854 of psychotropic medication in patients with emotionally unsta- Juan C. Pascual https://orcid.org/0000-0001-8039-9346 ble personality disorder under the care of UK Mental Health Services. J Clin Psychiatry. 2015;76(4):512-­518. REFERENCES 15. Pascual JC, Martín-­Blanco A, Soler J. Twenty-­year trends in 1. American Psychiatric Association. Diagnostic and Statistical psychopharmacology treatment of outpatients with borderline Manual of Mental Disorders (DSM-­5). 5th ed. Washington, DC: personality disorder. CNS Drugs. 2021;35(9):1023-­1032. American Psychiatric Association; 2013. 16. Fineberg SK, Gupta S, Leavitt J. Collaborative deprescribing in 2. Grant BF, Chou SP, Goldstein RB, et al. Prevalence, correlates, borderline personality disorder: a narrative review. Harv Rev disability, and comorbidity of DSM-­IV borderline personal- Psychiatry. 2019;27(2):75-­86. ity disorder: results from the wave 2 National Epidemiologic 17. Pascual JC, Córcoles D, Castaño J, et al. Hospitalization and phar- Survey on alcohol and related conditions. J Clin Psychiatry. macotherapy for borderline personality disorder in a psychiat- 2008;69:533-­445. ric emergency service. Psychiatr Serv. 2007;58(9):1199-­1204. 3. Trull TJ, Jahng S, Tomko RL, Wood PK, Sher KJ. Revised 18. Linehan M. Cognitive-­Behavioral Treatment of Borderline NESARC personality disorder diagnoses: gender, prevalence, Personality Disorder. Guilford Press; 1993. and comorbidity with substance dependence disorders. J Pers 19. Burckell L, McMain S. Dialectical Behavior Therapy adapted Disord. 2010;24:412-­426. to the treatment of concurrent borderline personality disorder 342 |    SOLER et al. and substance use disorders. Clinical Addiction Psychiatry; 28. Davis JM, Chen N. Dose response and dose equivalence of anti- 2010;207-­217. psychotics. J Clin Psychopharmacol. 2004;24(2):192-­208. 20. Linehan M, Korslund KE, Harned MS, et al. Dialectical behav- 29. Physicians’ Desk Reference. Ativan Tablets (lorazepam)-­ Drug ior therapy for high suicide risk in individuals with borderline Summay; 2016 [20/02/2021]. http://www.pdr.net/drug-­summa​ personality disorder: a randomized clinical trial and compo- ry/Ativa​n-­Table​ts-­loraz​epam-­2135.1869 nent analysis. JAMA Psychiatry. 2015;72(5):475-­482. 30. Gunderson J, Masland S, Choi-­Kain L. Good psychiatric man- 21. Soler J, Pascual JC, Tiana T, et al. Dialectical behaviour ther- agement: a review. Curr Opin Psychol. 2018;21:127-­131. apy skills training compared to standard group therapy in bor- 31. Cohen J. Statistical Power Analysis for the Behavioral Sciences. derline personality disorder: a 3-­month randomised controlled 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates; 1988. clinical trial. Behav Res Ther. 2009;47(5):353-­358. 32. Ingenhoven T. Pharmacotherapy for borderline patients: busi- 22. First MB, Gibbon M, Spitzer RL, et al. User’s Guide for the ness as usual or by default? J Clin Psychiatry. 2015;76(4):522-­523. Structured Clinical Interview for DSM-­IV Axis II Personality 33. Landes SJ, Chalker SA, Comtois KA. Predicting dropout in Disorders: SCID-­II. Washington, DC: American Psychiatric outpatient dialectical behavior therapy with patients with bor- Press; 1997. derline personality disorder receiving psychiatric disability. 23. Gómez-­Beneyto M, Villar M, Renovell M, et al. The diagnosis of Borderline Pers Disord Emot Dysregul. 2016;3(1):1-­9. personality disorder with a modified version of the SCID-­II in a 34. Soler J, Trujols J, Campins MJ, et al. Stages of change in dialec- Spanish clinical sample. J Pers Disord. 1994;8(2):104-­110. tical behaviour therapy for patients with borderline personality 24. Zanarini M, Gunderson JG, Frankenburg FR, Chauncey DL. disorder. Br J Clin Psychol. 2008;47(Pt4):417-­426. The revised diagnostic interview for borderlines: discriminating BPD from other axis II disorders. J Pers Disord. 1989;3(1):10-­18. 25. Barrachina J, Soler J, Campins MJ, et al. Validation of a Spanish How to cite this article: Soler J, Casellas-­Pujol E, version of the diagnostic interview for borderlines -­ revised Fernández-­Felipe I, Martín-­Blanco A, Almenta D, (DIB-­R). Actas Esp Psiquiatr. 2004;32(5):293-­298. Pascual JC. “Skills for pills”: The dialectical-­ 26. Hassel S, Almeida JR, Kerr N, et al. Elevated striatal and de- behavioural therapy skills training reduces creased dorsolateral prefrontal cortical activity in response to emotional stimuli in euthymic bipolar disorder: no asso- polypharmacy in borderline personality disorder. ciations with psychotropic medication load. Bipolar Disord. Acta Psychiatr Scand. 2022;145:332–­342. doi:10.1111/ 2008;10(8):916-­927. acps.13403 27. Sackeim HA. The definition and meaning of treatment-­resistant depression. J Clin Psychiatry. 2001;62(16):10-­17.