Mechanisms of Change in Dialectical Behaviour Therapy

Journal of Mental Health ISSN: 0963-8237 (Print) 1360-0567 (Online) Journal homepage: http://www.tandfonline.com/loi/ijmh20 Mechanisms of change in dialectical behaviour therapy and cognitive behaviour therapy for borderline personality disorder: a critical review of the literature Susie Rudge, Janet Denise Feigenbaum & Peter Fonagy To cite this article: Susie Rudge, Janet Denise Feigenbaum & Peter Fonagy (2017): Mechanisms of change in dialectical behaviour therapy and cognitive behaviour therapy for borderline personality disorder: a critical review of the literature, Journal of Mental Health, DOI: 10.1080/09638237.2017.1322185 To link to this article: http://dx.doi.org/10.1080/09638237.2017.1322185 Published online: 08 May 2017. Submit your article to this journal Article views: 16 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=ijmh20 Download by: [The UC San Diego Library] Date: 10 May 2017, At: 20:33 http://tandfonline.com/ijmh ISSN: 0963-8237 (print), 1360-0567 (electronic) J Ment Health, Early Online: 1–11 ß 2017 Informa UK Limited, trading as Taylor & Francis Group. DOI: 10.1080/09638237.2017.1322185 REVIEW ARTICLE Mechanisms of change in dialectical behaviour therapy and cognitive behaviour therapy for borderline personality disorder: a critical review of the literature Susie Rudge1, Janet Denise Feigenbaum1,2, and Peter Fonagy1 1 Research Department of Clinical, Educational and Health Psychology, University College London, London, UK and 2North East London NHS Foundation Trust, London, UK Abstract Keywords Background: Little is known about the ‘‘active ingredients’’ of psychological therapy for Borderline Personality Disorder (BPD) despite a growing evidence base documenting its clinical effectiveness. This information can be used by clinicians to inform service planning and care pathways. Aims: The aim of this study was to review published empirical research investigating the potential mechanisms underlying therapeutic change in Dialectical Behaviour Therapy (DBT) and Cognitive Behaviour Therapy (CBT) for BPD. Method: A thorough search of the PsychInfo, CINAHL Plus, PubMed, MEDLINE and EMBASE databases revealed research into potential mechanisms of change. Results: A total of 52 abstracts were reviewed. After a full text screen of the most relevant studies, 14 met inclusion criteria. Twelve examined DBT and two CBT. Mechanisms of change identified broadly fell into three categories: emotion regulation/self-control, skills use and therapeutic alliance/investment in treatment. Outcomes measured included general mental health diagnoses (e.g. anxiety, depression) and BPD-specific symptoms (e.g. self-harm/ suicidality, impulsivity, substance misuse, anger). Conclusion: Further empirically robust research is required to test hypotheses about the influence of the proposed mechanisms on therapeutic change in psychological therapies for BPD. Borderline personality disorder, CBT, DBT, outcomes, therapeutic change Background Borderline Personality Disorder (BPD) is arguably the most common subtype of Personality Disorder seen by services (Coid et al., 2006; de Ruiter & Greeven, 2000) and has been extensively studied due to its association with suicide, selfharm, violence and substance misuse (American Psychiatric Association, 2013). Symptoms of BPD result in high levels of service usage (Bender et al., 2001; Comtois et al., 2003) and high mortality rates (American Psychiatric Association, 2001). Several characteristics of the disorder (e.g. impulsivity, recurrent suicidal behaviour) unfortunately lend themselves to early disengagement from treatment and difficulty committing to and engaging with the therapeutic process. Additionally, BPD is characterised by difficulties in establishing trusting and collaborative interpersonal relationships Correspondence: Dr Susie Rudge, c/o Dr Janet Feigenbaum, Research Department of Clinical, Educational, and Health Psychology, University College London, Gower Street, London WC1E 6BT, UK. E-mail: susie.webb.10@ucl.ac.uk, susie.rudge@hotmail.co.uk History Received 5 August 2016 Revised 19 November 2016 Accepted 27 February 2017 Published online 5 May 2017 and, ‘‘frantic efforts to avoid real or imagined abandonment’’ (American Psychiatric Association, 2013), which naturally extend to difficulties in the therapeutic relationship. Indeed, a recent qualitative study confirmed patients’ reluctance to be open and honest with their therapist because of fears of rejection and abandonment (Morris et al., 2014). Owing to the stigma associated with BPD, clinicians may find it difficult to communicate the diagnosis in a patient-centred manner (Sulzer et al., 2015), further exacerbating problematic therapist–patient relationships. Although research has sought to identify effective therapeutic treatments for the condition, the majority of BPD research to date focuses on outcome data with relatively few studies identifying reasons why therapies are successful, and what the specific processes through which improvements occur might be (Lynch et al., 2006). Linehan (2000) notes the need to identify ‘‘active’’ components of psychological therapy so that those aspects can be emphasised when striving for the most effective treatment. Clarkin & Levy (2006) highlight the difference between the vast number of outcome studies and the relatively few studies of mechanisms 2 S. Rudge et al. of change clarifying that, ‘‘the question of the mechanisms of change in psychotherapy seeks to learn how a particular therapy works, not what is the outcome of the treatment per se’’. Elliott (2010) refers to this research as ‘‘change process research’’ describing it as, ‘‘a necessary complement to randomised clinical trials and other forms of efficacy research’’. Even in the most rigorously researched psychotherapeutic interventions, researchers lack insight into the mechanisms through which these treatments result in successful outcomes and future investigations should strive towards evidencing this as the next step in psychotherapy research (Kazdin, 2007). Identifying specific mechanisms through which symptoms improve with treatment has vast implications for the future of psychological therapy for BPD. Pre-assessment, these data could allow clinicians to predict which patients are more likely to do well with which treatment. As well as informing care pathways, this information could assist in the planning and formation of new services. Current therapies for BPD include Dialectical Behaviour Therapy (DBT; Linehan et al., 2006a), Cognitive Behaviour Therapy (CBT; Beck et al., 2004), Mentalization Based Therapy (MBT; Bateman & Fonagy, 2008; Bateman & Fonagy, 2004), Transference-Focused Psychotherapy (TFP; Clarkin et al., 2006) and Schema-Focused Therapy (SFT; Kellogg & Young, 2006). Evidence suggests that CBT is an effective treatment for BPD (Davidson et al., 2006; Leichsenring & Leibing, 2003) with therapy focusing particularly on the development of functional new core beliefs. The therapeutic relationship is seen as a vital means for exploring the patient’s style of relating to others and for fostering more adaptive future interactions. Arntz (1994) describes CBT for BPD as consisting of five stages: (i) construction of working relationship, (ii) symptom-management, (iii) correction of thinking errors, (iv) emotional processing and cognitive re-evaluation of childhood trauma and schema changes and (v) termination. Developed by Linehan (1993), DBT has a large and robust evidence base (Bloom et al., 2012; Feigenbaum et al., 2011; Feigenbaum, 2007; Kliem et al., 2010; Linehan et al., 2006a; Panos et al., 2014). It uses strategies developed to aid enhanced regulation of emotions as well as teaching distress tolerance and using third wave approaches such as mindfulness to promote awareness and acceptance. Because of the large number of studies of DBT and CBT for BPD, this review focuses on potential mechanisms of change in these two treatments only. Method Searching, identifying and selecting studies for inclusion Searches of paper titles, abstracts and full text content were initially performed in July and August 2012 then updated in February 2014, in PsychInfo, CINAHL Plus, PubMed, MEDLINE and EMBASE databases. Search terms used were (a) ‘‘mechanism* change borderline personality disorder’’, (b) ‘‘mechanism* change’’ and ‘‘borderline personality disorder’’, (c) ‘‘mechanism* change’’ and ‘‘BPD’’, (d) ‘‘mechanism* change’’ and ‘‘borderline personality J Ment Health, Early Online: 1–11 disorder’’ and ‘‘treatment’’ and (e) ‘‘borderline personality disorder’’ and ‘‘therapeutic change.’’ Studies included in the review involved participants who  met standardised diagnostic criteria for BPD  had received either CBT or DBT treatment for their BPD  were treated as outpatients (due to the limited number of manualised DBT/CBT studies of inpatients or partially hospitalised patients with BPD)  were treated within full text, peer-reviewed, empirical studies published in English since 1990 (as this was the earliest that the literature began to report CBT and DBT treatment of BPD)  were adults (aged 18+ years) at the time of treatment (as there is a limited research presence investigating emerging BPD in adolescents) The review excluded  single case studies  studies which did not ultimately produce evidence on mechanisms of change and were, therefore, better defined as outcome studies. To ascertain this, two questions were asked: (1) Is the studied variable theorised to be a mechanism of change in a (separately defined) outcome variable? (2) Does the data presented investigate an association/ correlation between the proposed mechanistic variable and an outcome variable? Using the five search terms within each of the aforementioned databases yielded a total of 104 references which, following a title screen, reduced to 34 abstracts for review. It was clear that 15 papers did not meet inclusion criteria and, therefore, 19 full texts were appraised. Nine studies were ultimately excluded for presenting outcome data only, failing to separate out potential mechanisms by treatment type, or for combining inpatient/outpatient data. One study (Stepp et al., 2008) did not strictly meet inclusion criteria as only 63% of participants met diagnostic criteria for BPD and because at least one participant was under 18 years. It was decided, however, that because of the study’s relevance to the review it would still be beneficial to include, albeit with caveats. This left ten studies for inclusion. In October 2016, identical searches to those completed in 2012 and 2014 were re-run in all five databases. This search returned 261 references, yielding a total of 18 new abstracts for review after removal of studies appraised previously and those not relevant to BPD. Eleven further full texts were reviewed and a further four studies met inclusion criteria, taking the total to 14. See Figure 1 for flow chart of the review process. Studies included The 14 studies reviewed were published between 2000 and 2016 (Table 1). Twelve studies examined mechanisms of change in DBT and two in CBT. The two CBT studies (Gibbons et al., 2009; Wenzel et al., 2006) utilised the same sample of participants (n ¼ 32) from a previous larger trial (Brown et al., 2004), but investigated different hypothesised mechanisms of change. Two of the DBT studies Mechanisms of Change in DBT and CBT for BPD DOI: 10.1080/09638237.2017.1322185 Figure 1. Flow chart of review database. 1990-2014 3 2014-2016 Electronic database search results Electronic database search results n = 104 n = 261 Excluded following electronic title screen Excluded following electronic title screen n = 70 n = 243 Abstracts screened Abstracts screened n = 34 n = 18 Excluded following abstract screen Excluded following abstract screen n=7 n = 15 Full texts screened Full texts screened n = 19 n = 11 Excluded following full text screen Excluded following full text screen n=7 n=9 Papers identified for inclusion in final review Papers identified for inclusion in final review n = 10 n=4 Papers included in final review n = 14 (Bedics et al., 2012, 2015) also used the same sample (n ¼ 101). Ten studies took place in the USA, two in Switzerland, one in Canada and one in the UK. Sample sizes ranged from 24 to 165 and combined, the studies included 961 participants of which 912 (95%) were female. Ages ranged from 16–61 years with a mean of 31.2 years. Analysis Studies were evaluated based on their design and findings and measured against a critical appraisal checklist (DBC; Downs & Black, 1998) which assesses the methodological quality of both randomised and non-randomised studies of healthcare interventions. Due to difficulty in ascertaining reliable scores for the final item (power analyses, item 27) which awards up to five points, this item was instead scored either ‘‘0’’ (no power calculation completed or power not met) or ‘‘1’’ (power calculation completed, and met). Therefore, a maximum score of 28 was possible (items 1–27 are scored either ‘‘0’’ or ‘‘1’’ with the exception of item five which is worth up to two points). A summary of each study’s performance against the DBC can be found in Table 2 (DBT) or Table 3 (CBT). Across the 14 studies evaluated, three main themes emerged under which all identified mechanisms of change could be categorised, (i) emotion regulation and self-control, (ii) skills use and (iii) therapeutic alliance and investment in treatment. Each study is described in detail below under one of these three categories. Results Emotion regulation and self-control Axelrod et al. (2011) posited that greater control of emotions in BPD would lead to less impulsive behaviour which would, in turn, reduce the need to self-medicate using substances to regulate emotions. Females with substance dependence and BPD received a 20-week course of outpatient DBT and emotion regulation was assessed using the Difficulties in Emotion Regulation Scale (Gratz & Roemer, 2004). Substance use was recorded for 30 d preceding treatment and for the final 30 d of treatment, corroborated by weekly self-report, clinician assessment, urine toxicology and alcohol breathalysers. The study concluded that improvements in emotion regulation explained the variance in decreased substance use frequency. Changes in substance use lost their significance when improvement in emotion regulation was controlled for. This study has some limitations. First, the authors use the term ‘‘behavioural control’’ as their primary outcome but they measured this only by substance use. There are other aspects of behavioural control relevant to BPD (e.g. 4 Table 1. Papers included in the review. Sample size 2011 USA DBT No n ¼ 27 Barnicot et al. 2015 UK DBT No n ¼ 70 Bedics et al. 2015 USA DBT Yes n ¼ 101 Bedics et al. 2012 USA DBT Yes n ¼ 101 Gibbons et al. 2009 USA CBT Yes n ¼ 32 Kramer et al. 2016 Switzerland DBT Yes n ¼ 41 Linehan et al. 2015 USA DBT Yes n ¼ 99 99 women aged 18–60 years (no mean age reported). McMain et al. 2013 Canada DBT Yes n ¼ 80 67 women and 13 men with BPD. Mean age ¼ 32.6 years (S.D.¼10.1 years) Affect balance, problem solving and ability to identify and describe emotions Neacsiu et al. 2010 USA DBT Yes n ¼ 108 Increasing use of DBT skills O’Toole et al. 2012 USA DBT No n ¼ 165 63 recurrently suicidal women with BPD and 45 women with BPD. Mean age ¼ 31.4 years (S.D.¼7.4 years) 165 women with BPD. Mean age ¼ 37.1 years (S.D.¼12.04 years) Perroud et al. 2012 Switzerland DBT No n ¼ 54 Mindfulness and acceptance Stepp et al. 2008 USA DBT No n ¼ 27 Turner 2000 USA DBT Yes n ¼ 24 Wenzel et al. 2006 USA CBT No n ¼ 32 47 women and 7 men with BPD and suicidal/self-harm behaviour. Mean age ¼ 30.5 years (S.D.¼7.7 years) 23 women and 4 men. 63% met DSM-IV criteria for BPD. Mean age ¼ 30.4 (range ¼ 16–61 years) 19 women and 5 men with BPD. Mean age ¼ 22.0 years (range ¼ 18-27 years) 28 women and 4 men with BPD. Mean age ¼ 29.0 years (range ¼ 20–55 years) (from Brown et al., 2004) Axelrod et al. Participants 27 women with BPD and substance dependence. Mean age¼ 38.0 years (range ¼ 27–51 years) 63 women and 7 men with BPD. Mean age ¼ 32.0 years (S.D.¼10.2 years) 101 women with BPD. Mean age ¼ 29.3 years (S.D.¼7.5 years) 101 women with BPD. Mean age ¼ 29.3 years (S.D.¼7.5 years) 28 women and 4 men with BPD. Mean age ¼ 29.0 years (range ¼ 20-55 years) (from Brown et al., 2004) 36 women and 5 men with BPD. Mean age ¼ 34.4 years (S.D.¼9.1 years) Mechanism(s) of change Main findings Improvements in emotion regulation Improved emotion regulation can account for increased behavioural control Improvements in DBT skills use Facets of therapeutic alliance Personality factors and intrapsychic change, perception of therapeutic alliance Self-understanding and compensatory skills More frequent use of DBT skills was independently associated with less frequent self-harm Reduction in self-harm associated with increase in patient-rated total alliance in DBT group DBT patients reported self-affirmation, protection, love and less self-attack than controls Productive use of ager as an emotion and DBT skills use DBT skills use The use of DBT skills resulted in greater symptom reduction in the DBT skills group Mindfulness skills Increased DBT skills use Quality of therapeutic alliance Belief change, reduction in hopelessness, improvement in attitude towards treatment Change in compensatory skills observed in CBT group Standard DBT was not superior to other forms of DBT on suicide outcomes but was superior in reducing self-harm and improving other mental health problems Participants with improvements in affect balance, problem solving, and the ability to identify and describe emotions showed greaterimprovements in symptom distress and interpersonal function DBT skills use mediated decreases in suicide attempts and depression The use of mindfulness skills predicted emotional well-being. Women reporting greater use of mindfulness skills also reported less use of healthcare services although there was no associated change to prescription levels Increase in skill of accepting without judgement correlated with improvements in BPD symptoms Increased overall DBT skills use had a significant effect on measure of BPD symptoms DBT group improved more than controls on most outcomes Positivity towards treatment correlated with improvement in BPD diagnostic criteria J Ment Health, Early Online: 1–11 Control group? Author(s) S. Rudge et al. Country Primary therapeutic orientation Publication year Mechanisms of Change in DBT and CBT for BPD DOI: 10.1080/09638237.2017.1322185 5 Table 2. Checklist appraisal of DBT studies according to the DBC (Downs & Black, 1998). Paper Strengths according to DBC Axelrod et al. (2011) Reporting, sampling Barnicot et al. (2015) Reporting, outcome measure, statistical techniques Bedics et al. (2015) Randomisation, control group, large sample Bedics et al. (2012) Randomisation, control group, large sample Kramer et al. (2016) Linehan et al. (2015) Neacsiu et al. (2010) O’Toole et al. (2012) Randomisation, blinding of observers, control group Randomisation, comparison groups, blind assessors, participant matching Randomisation, statistical control of data, control of confounds, control group Randomisation, control group, blind assessors Reporting, sampling, large sample size, analyses Perroud et al. (2012) Turner (2000) Stepp et al. (2008) Outcome measures, sampling Randomisation, blind, independent assessors Reporting, statistical analyses McMain et al. (2013) Total DBC score (/28) Limitations according to DBC Attrition (44.4% did not complete treatment), small sample, lack of control group Attrition (46% did not complete treatment), lack of control group Reliance on self-report and lack of control of confounding variables Reliance on self-report, unable to determine treatment compliance Non-blinding of participants Non-blinding of participants 16 25 24 Non-blinding of participants and researchers 25 Non-blinding of participants Lack of control group, presence of confounding variables Lack of control group Lack of information about non-completers Non-randomisation, lack of control group, small sample size 26 18 21 21 21 19 21 15 Table 3. Checklist appraisal of CBT studies according to DBC (Downs & Black, 1998). Paper Gibbons et al. (2009) Wenzel et al. (2006) Strengths according to DBC Limitations according to DBC Randomisation, large sample, control/comparison groups Management of data from participants lost to follow-up impulsivity, self-harm) that could have been measured to more convincingly argue the role of emotion regulation as a mechanism of change. Further, the lack of controls of course impedes the possibility of attributing emotion regulation improvements exclusively to DBT. Additionally, most DBT studies utilise lengthier treatment (12 months/40 + sessions) so it is unclear whether this study replicated a full, comparable ‘‘dose’’. The study’s all-female sample does not facilitate conclusions about emotion regulation in males receiving DBT, although it is perhaps justified (and other BPD samples are also female-dominated) due to the ratio of BPD treatment-seeking females to males (currently estimated at 3:1; American Psychiatric Association, 2001), as well as the fact that current NICE (2009) guidance for BPD recommends the use of DBT treatment for females only. Using participants from a larger randomised controlled trial (RCT), over nine months McMain et al. (2013) used self-report measures to investigate the role of affect, problemsolving and emotional control. BPD symptoms and interpersonal function were assessed as outcomes every four months using well-known, standardised measures. The RCT compared the clinical effectiveness of DBT versus general psychiatric management (GPM) in 80 patients (67 female) diagnosed with BPD. Defining improved affect balance as an increased positive to negative ratio of emotions, results supported associations between improved affect balance with both reductions in symptom distress and improved interpersonal function. The researchers managed their data conscientiously; however, the final sample size in each parameter was reasonably small, limiting generalisability. Additionally, there Non-blinding of participants Lack of control group Total DBC score (/28) 20 19 was inadequate statistical power to test for between group differences and effects were, therefore, potentially confounded by the differences between DBT and GPM. Using a primarily female sample of 41 participants with BPD, Kramer et al. (2016) compared a 20 session version of DBT-informed skills training against a treatment as usual (TAU) control condition in an RCT (DBT: n ¼ 21, TAU: n ¼ 20) which investigated two forms of anger: primary, adaptive, ‘‘assertive’’ anger and secondary, ‘‘rejecting’’ anger. The latter could be classified in BPD patients as hostility or aggression. The former is a more accepted and adaptive emotional experience in which a person’s rights are rationally defended. The researchers hypothesised that the occurrence of rejecting anger would remain stable over time for both DBT and TAU but that observed expressions of primary assertive anger would increase more in DBT participants. Anger was measured both early and late in treatment using an intensive behavioural assessment of a standardised psychological interview (Perry et al., 2005). As well as finding that symptom reduction was greater in the DBT group than TAU, as expected the researchers discovered that the DBT group displayed increased use of assertive anger compared to TAU, whereas no effect was found for less productive rejecting anger. The authors link their findings to the theory that reactive angry responses may drive the staterelated problematic behaviour seen in BPD (e.g. self-harm, suicidal ideation, interpersonal aggression; Brown et al., 2002). A particular strength of this study lies in its robust methodology; however, because participants were not blind to treatment condition, expectation biases may have occurred. 6 S. Rudge et al. Not uncommonly, attrition was a problem; however, the researchers adequately addressed this using intent to treat analyses. Throughout the paper, the DBT administered is described as ‘‘DBT-informed skills training’’ (a group format) meaning that it lacked the other components of standardised DBT (individual therapy sessions, therapist team consultations and between-session telephone coaching) which comprise the complete treatment. It is possible that the observed effects could be lost if the skills focus of the treatment program was diluted with comprehensive DBT. Additionally, effects were not demonstrated at followup which could have been due to post-treatment dropouts lowering statistical power or because the shortened treatment format may have represented an insufficient dose of DBT. Skills use A key aspect of DBT is the teaching of specific behavioural skills which aim to replace maladaptive behaviours with more adaptive responses (Linehan, 1993). Neacsiu et al. (2010) noted that no study to date had directly tested this mechanism of change. They investigated DBT skills use in a sample of 108 women with BPD who were participating in a 12-month RCT with a 4-month follow-up. Participants included 63 recurrently suicidal women and 45 women with drug dependence (there were no significant demographic differences between the two groups). Participants received either DBT or one of three control treatments: Community Treatment by Experts (CTBE), Comprehensive Validation Therapy (CVT) or TAU in conjunction with a 12-step program. Measures of DBT skills use, anger, suicidal/selfharm behaviour and depression were gathered using a combination of self-report and semi-structured interviews. Although anger suppression and expression was not found to mediate outcome, significant mediation effects did indicate that use of DBT skills fully mediated decreases in suicide attempts and depression symptom severity and an increase in the control of anger over time. The use of DBT skills also partially mediated a decrease in self-harm over time. Participants who received DBT reported using three times more skills by the end of their treatment (mean skills use increased by 15.3%), compared with control participants (mean skills use increased by 4.6%). At follow up, DBT participants maintained increased skill use but control participants had decreased by 5%. Although this study demonstrated support for the DBT skills deficit model of BPD, it is limited by its primary reliance on subjective selfreport as well as using retrospective methods which are subject to both memory biases and over/underreporting responder biases. Assessing skills use on a daily basis using a more objective measure would increase reliability. When using standard mediation analysis, an assumption is made that there are no confounds manipulating the mediator and outcome (Robins & Rotnitzky, 2005) and it is possible that uncontrolled extraneous variables influenced the meditational analysis in this study such that an increase in DBT skills use was not the only variable influencing positive outcomes. Nevertheless, methodologically, this remains a robust study, reflected by the highest score awarded by the DBC. J Ment Health, Early Online: 1–11 Gibbons et al. (2009) were specifically interested in the acquisition of compensatory skills, self-understanding and perception, which they considered to be theoretically important mechanisms of change affecting outcomes in cognitivebased therapy for BPD. As part of a larger clinical trial (Brown et al., 2004), 34 participants with a primary diagnosis of BPD received 12 months of cognitive therapy tailored to BPD with self-report questionnaires used to measure selfunderstanding and acquisition of compensatory skills. Outcomes of depression, anxiety and quality of life were measured using well-known, validated self-report measures. The researchers found that change in compensatory skills was apparent in the BPD group and that in particular, a decrease in negative compensatory responses/negative thinking cooccurred with symptom improvement. This study used data from a larger trial which also included participants with a primary diagnosis of depression or anxiety so outcome measures were perhaps too broad to capture some of the additional symptoms experienced by those with BPD. Not uncommonly, this study relied heavily on self-report, creating the possibility of biased responding. The researchers concede that the relatively small within-study sample sizes and their associated limitations on statistical power meant that the use of a pooled database was not the best way to investigate mechanisms of change in specific treatments for specific diagnostic categories. Over a 12-month period, Perroud et al. (2012) investigated improved skills in mindfulness, a key component of DBT. Fifty-two (predominantly female) participants with a BPD diagnosis were regularly administered the Kentucky Inventory of Mindfulness Skills (KIMS; Baer et al., 2004), a self-report questionnaire which categorises mindfulness into four discrete dimensions: observing, describing, acting with awareness and accepting without judgment. Self-report measures of depression and hopelessness were also administered at regular intervals, as were standardised diagnostic clinician-administered assessments of BPD psychopathology. Accepting without judgement was the only dimension found to significantly increase following statistical adjustment for potential confounds. Increases in this dimension specifically correlated with improved BPD symptoms. Mindfulness is perhaps a construct inherently difficult to measure objectively so selfreport may be the best way to capture it, despite potential response biases. However, this study lacked a control group, limiting the possibility of drawing conclusions about whether observed improvements are exclusive to the acquisition of the accepting without judgement skill or whether they are partially or otherwise explained by a natural change in mindfulness skills and/or correlate with an uncontrolled confound. O’Toole et al. (2012) also studied improvements in mindfulness skills in 165 women with BPD recruited from five DBT programs. Self-report measures of perceived social support and physical and emotional well-being were used to assess outcomes. Mindfulness skills were measured using the Five Facet Mindfulness Questionnaire (FFMQ; Baer et al., 2008), a 39-item self-report measure. Mindfulness emerged as the strongest predictor of emotional well-being and women who reported greater use of mindfulness skills reported more infrequent use of healthcare services. Mindfulness was, DOI: 10.1080/09638237.2017.1322185 however, unrelated to the use of prescription medication. The large sample size and robust statistical techniques add to the reliability of these findings although the voluntary and diverse recruitment process creates potential biases and as with some other studies, data relied solely on self-report. This study, nevertheless, makes an important contribution to the evidence regarding the role of mindfulness skills as a DBT change mechanism. Using a multi-level repeated measures, non-randomised, uncontrolled design, Stepp et al. (2008) set out to identify possible ‘‘active ingredients’’ of DBT that may account for improved BPD symptoms. Their sample of 27 participants (85% female) ranged in age from 16 to 61 years but only 63% met diagnostic criteria for BPD. To assess BPD symptoms, the Personality Assessment Inventory-Borderline Features Scale (PAI-BOR; Morey, 1991), a 24-item self-report measure was administered at the start of each new skills module (teaching sequence: (i) mindfulness, (ii) interpersonal effectiveness, (iii) emotion regulation and (iv) distress tolerance). Skills use was assessed by weekly self-report diary cards. Analyses revealed that overall skills use produced a significant effect on PAI-BOR total scores over the course of 12 months. This finding held when analyses controlled for baseline levels of distress and diary card compliance. The methodological limitations of this study meant that it achieved the lowest score on the DBC: the researchers concede that their findings could reflect motivation rather than skills utilisation; more motivated participants would arguably also be more committed to completing diary cards. However, this confound was controlled as much as possible. Both skills use and outcome could be more reliably assessed by using blinded performance-based observer ratings. Despite the study’s power-maximising design, small effect sizes could have been missed due to the small sample and the fact that one-third of participants did not meet DSM-IV criteria for BPD introduces an uncontrolled confound limiting the conclusions that can be drawn about BPD from this research. The lack of a control group means that results may not be reflective of skills specific to DBT and could be a finding common to any skills utilisation program. Within a single-blind, randomised trial of three different types of DBT, Linehan et al. (2015) aimed to ascertain the effect of DBT skills use on outcomes of suicide attempts, selfharm and mental health problems in 99 women diagnosed with BPD. Outcome measures included the Suicide Attempt Self Injury Interview (SASII; Linehan et al., 2006b) and the Suicidal Behaviours Questionnaire (Addis & Linehan, 1989). Assessments were conducted by blinded independent assessors prior to treatment, every two months during the 12month treatment and then 12 months post-treatment. The researchers’ methodology dismantled DBT into the following formats: skills training plus case management (DBT-S), DBT individual therapy plus activities group (DBT-I) and standard DBT (this includes both skills training and individual therapy). Standard DBT was not shown to be superior over other forms of DBT with regards to suicide-related outcomes; all three formats were equally effective at reducing suicidality among high-risk participants. However, the two interventions which incorporated DBT skills training were more effective in reducing self-harm incidents and improving other mental Mechanisms of Change in DBT and CBT for BPD 7 health problems. This rigorously controlled, single-blind study used computerised randomisation and matching to allocate participants to their DBT program and a particular strength lies in the management of the data and control of, and statistical investigation into, potential confounds. Because participants were informed of their treatment allocation during their first therapy session, the possible effect of expectation biases cannot be ruled out, however. Barnicot et al. (2015) used a predominantly female sample of 70 participants with BPD, aiming to examine whether DBT skills use was associated with positive treatment outcomes independent of treatment processes common to most psychological therapies: therapeutic alliance, treatment credibility and self-efficacy. Skills use, the proposed change mechanism, was measured by self-report every two months and self-harm was clinician-assessed every two months using items selected from the SASII (Linehan et al., 2006b). The researchers concluded that more frequent use of DBT skills was independently associated with less frequent concurrent selfharm. This is a solid study; however, due to the use of DBTspecific terminology, it was not possible to compare skills use in a control group not receiving DBT. Therapeutic alliance and investment in treatment Bedics et al. (2012) aimed to explore the therapeutic alliance as a mechanism of change on self-harm outcomes in DBT. One hundred and one females were randomised to receive either DBT or a control condition, Community Treatment by Experts (CTBE). As well as meeting criteria for BPD, all participants had a history of self-harm and at least one incident in the eight weeks prior to commencing the study. The quality of the therapeutic alliance was rated by patients using the Structural Analysis of Social Behaviour (SASB; Benjamin, 1974). Results showed that in comparison to CTBE participants, DBT participants reported their therapists as increasingly more affirming, protecting and controlling during treatment. Additionally, DBT participants reported a stronger association between increased therapist affirmation and protection with decreased self-harm. Despite the strength of the RCT data, the reasonable sample size and the use of multiple time points to assess symptomatic change and the therapeutic relationship, this study has some limitations. Assessment of BPD symptoms was limited to self-harm only and the researchers note the value that further research could add in extending these results to other relevant domains. Additionally, reliability of the data is limited because of the lack of clinician-recorded or blinded observations. Continuing their research, Bedics et al. (2015) used DBT and CTBE comparison data from their previous sample (Bedics et al., 2012), this time employing the California Psychotherapy Alliance Scale (CALPAS; Gaston, 1991) to investigate the association between different components of the alliance and BPD outcomes (suicide attempts, self-harm, introject and depression) which were measured using standardised, validated tools. No differences were found between patient ratings of the alliance between the two treatments; however, a reduction in self-harm was associated with an increase in patient-rated total alliance in DBT but not in CTBE. Although it did not quite reach significance, 8 S. Rudge et al. researchers also discovered that DBT participants who perceived greater understanding and involvement from their therapist reported reductions in self-harm. As per their rationale, investigating specific facets of the alliance highlighted mechanisms that may not be revealed when focusing on total alliance ratings; however, the researchers did not control for the possibility that early symptom improvement may have influenced alliance ratings, causing subsequent changes in symptoms. Turner (2000) tested the effects of DBT versus a ClientCentred Treatment control condition (CCT) in a naturalistic evaluation of 24 (primarily female) participants with a diagnosis of BPD. In order to understand its role in differences in outcomes (depression, anxiety, anger, selfharm/suicidality, hospitalisation) between the two therapies, the quality of the therapeutic alliance was measured using the Helping Relationship Questionnaire (HRQ; Luborsky, 1984). Participants were randomly assigned to receive either DBT or CCT and outcomes were evaluated using a combination of self-report and a blind rating assessor. Differences in ratings of the quality of the therapeutic alliance were found to account for significant variance in outcomes across both DBT and CCT but no significant difference in therapeutic alliance was observed between the two treatments. This suggests that the alliance accounted for as much variance in symptom improvement as did differences in the treatment conditions themselves. Researchers rated the quality of the alliance at one single time point rather than measuring a change (improvement) in alliance over time, making it harder to infer its role as a mechanism of change linked explicitly to improved BPD symptoms. Like others, this study relied heavily on self-report within a relatively small sample but the use of randomisation and controls contribute to the reliability of these important findings. Wenzel et al. (2006) proposed that change in dysfunctional beliefs, reduction in hopelessness and improvement in attitude toward treatment all function as mechanisms of change in CBT for BPD. Using data from 32 participants diagnosed with BPD as part of a wider clinical trial (Brown et al., 2004), the researchers conducted clinical evaluations at baseline, 6 months and 12 months then again at 6 months follow-up. Baseline assessments involved clinician-administered interviews, self-report questionnaires and review of treatment histories. Attitude towards treatment was measured using the Attitudes and Expectations Questionnaire (ERQ; adapted from Elkin et al., 1989). Results showed that 66.7% of participants who had positive attitudes toward treatment no longer met criteria for BPD after 12 months of treatment, compared with 14.3% of participants with a negative attitude toward treatment. This may be a spurious link, however; it is not clarified how changes in attitude towards treatment specifically influence outcome and without the benefit of data obtained at more than one time point it is perhaps not reliably described as a mechanism of change. The researchers investigated other hypothesised mechanisms of change concluding, in support of their hypotheses, that reductions in hopelessness were associated with significant reductions in borderline beliefs. However, this conclusion does not shed much light on the specific processes by which change occur, as both belief change and reduction in hopelessness might be J Ment Health, Early Online: 1–11 more reliably classed as outcomes rather than change mechanisms. Further, the small sample size precludes the possibility of making generalisable inferences to larger samples and the standard critique of self-report measures also applies, although the use of clinician-administered assessments and treatment records did introduce more objective ratings. Discussion Results from this review show that there are at least three distinct categories of mechanisms of change in DBT and CBT for BPD. Empirical support for improvements in emotion regulation and behavioural control as change mechanisms in DBT is perhaps unsurprising given that Linehan’s (1993) DBT biosocial theory views BPD as a disorder of persistent emotional dysfunction occurring largely due to deficits in the ability to regulate difficult emotions and because of emotional instability and vulnerability. By contrast, no studies reviewed investigated emotion regulation and self-control in CBT, which, again, is perhaps unsurprising given that the core aim of CBT for BPD focuses more on dysfunctional schema identification and cognitive restructuring (Arntz, 1994). Axelrod et al. (2011) discovered an association between emotion regulation and behavioural control in DBT, although both variables might be considered both mechanisms of change and outcomes in their own right. Similarly, as well as discovering a link between improved affect balance and reductions in symptom distress, McMain et al. (2013) found an association between balanced affect and improved interpersonal function, a key component in DBT skills training. Also supporting a link between emotion regulation and skills use, Kramer et al. (2016) showed that receipt of a DBT skills training module was associated with improved ability to use productive, assertive anger, an association also apparent in Neacsiu et al.’s (2010) mediation analysis which concluded that the frequency of DBT skills use by patients with BPD was associated with an increase in control of anger over time. In addition to Neacsiu et al. (2010), several DBT studies (Barnicot et al., 2015; Linehan et al., 2015; O’Toole et al., 2012; Perroud et al., 2012; Stepp et al., 2008) provided some support for Linehan’s (1993) skills deficit theory of BPD suggesting that acquisition of new skills is associated with better DBT outcomes. Indeed, a study of 49 women found that both mindfulness and crisis survival skills were practiced frequently by DBT participants and that the majority practiced their DBT-learned skills on most treatment days (Lindenboim et al., 2007). Arntz (1994) lists ‘‘correction of thinking errors’’ as one of the five main components to be addressed in CBT for BPD. It is, therefore, again encouraging, yet unsurprising, that Gibbons et al. (2009) concluded that compensatory skills use was association with reductions in negative cognitions which correlated with BPD symptom improvement. Across a range of psychotherapies, the therapeutic alliance is considered helpful in retaining patients in therapy as well as contributing to positive outcomes (Horvath & Luborsky, 1993). Patients view a trusting alliance as something to be prioritised (Morris et al., 2014). However, it remains a difficult concept to quantify and could easily be conflated DOI: 10.1080/09638237.2017.1322185 with other mechanisms of change such that its role as an independent factor becomes less clear. The most sensible definition for considering therapeutic alliance as a mechanism of change might be to measure change in the alliance over time to show that as it develops (and hopefully improves) so BPD symptoms reduce – a positive, measureable outcome. Because of their similarity, therapeutic alliance and investment in treatment were clustered to form one category in the current review but examination of both proved problematic in terms of identifying isolated processes which could reliably be classed as mechanisms of change. Barnicot et al.’s (2015) study into the association between DBT skills use and treatment outcome (self-harm, drop-out) investigated the mediating effect of three common treatment processes but of the three, both therapeutic alliance and treatment credibility were not found to be linked with decreased self-harm. More promisingly, Wenzel et al. (2006) found that a positive attitude towards treatment was associated with a reduction in BPD symptoms in CBT, although it was unclear how much this factor alone was responsible for patients no longer meeting diagnostic criteria for BPD post-treatment. Exploring components of the therapeutic alliance in further detail, Bedics et al. (2015) discovered that participants who perceived greater understanding and involvement from their therapist reported a reduction in self-harm. This fits with Linehan’s (1993) model as instead of the invalidating environment that BPD patients are accustomed to, therapists provide warm, emotionally validating settings, fostering increased emotional regulation and decreased instability and impulsivity (which manifest in behaviours such as deliberate self-harm). Therapeutic alliance and investment in treatment are examples of non-specific processes theorised to be common across all psychological therapies, as opposed to specific effects that are produced by different therapists/models (see Wampold (2001) for review of the value of specific versus non-specific processes in psychotherapy mechanism research and Lynch et al. (2006) for a discussion of theorised common and unique mechanisms in DBT treatment). Interestingly, although Bedics et al. (2012) and Turner (2000) both concluded that a more positively-perceived alliance was associated with improved DBT outcomes, they produced contradicting evidence on the importance of the alliance as a positive change process. Bedics et al. (2012) found that the alliance was reported more favourably in DBT than controls whereas Turner (2000) found no significant difference in patient-reported alliance between DBT and controls. Nevertheless, the importance of the therapeutic alliance in the studies reviewed is a finding in agreement with a review of factors predicting outcome in BPD treatment (Barnicot et al., 2012) and is particularly promising given the difficulty BPD patients have with interpersonal relationships. Martin et al.’s (2000) meta-analysis of studies measuring alliance concluded that the overall relationship between the alliance and outcome is moderate but consistent regardless of any hypothesised confounds. They found great diversity in measures of alliance, suggesting that this research may not be easily replicable, adding to the difficult task of producing robust evidence of the alliance as a mechanism of change. Mechanisms of Change in DBT and CBT for BPD 9 As well as the inherent risk of this search omitting relevant studies, this review was limited by ultimately being primarily DBT-focused with little CBT evidence. This is likely due to the prominence of DBT in the most current BPD clinical guideline (NICE, 2009) as well as the fact that DBT was created specifically for BPD. It explains why the DBT model more aptly describes the mechanisms of change identified herewith than the CBT model. This does, however, suggest that further research into mechanisms of change in cognitive therapy for BPD is warranted, especially as NICE (2009) advises that should these data be produced, future revisions may recommend CBT for BPD. The empirical data reviewed highlight the difficulty in demonstrating causality, much of the evidence relying on associative relationships, and the majority of studies revealed difficulties in obtaining large enough sample sizes and in establishing satisfactory scientific rigour from which to base conclusions. Whether some variables were classed as mechanisms of change or could more reliably be considered as outcomes was also somewhat muddied. Indeed, in their study of DBT partial hospitalisation, Yen et al. (2009) concluded that, ‘‘BPD is a complex, heterogeneous disorder for which there is no single pathognomonic criterion, so that each criterion should be considered individually in determining its potential effect on treatment outcomes.’’ Further robust research and hypothesis testing will help to corroborate the identified mechanisms and attempts to establish causality would be highly beneficial in concluding which components of therapy to focus on. These data could assist practitioners in testing the efficacy of briefer interventions that incorporate the specific mechanisms that are most likely to lead to positive outcomes, particularly benefitting those for whom only brief treatments are available. Encouragingly, there was a large increase in the number of relevant studies published between the 2012 and 2016 searches, with more than a quarter of the studies that met inclusion criteria being published in the last two years. Demonstrating that this is a research-worthy area of growing interest, searches also revealed several studies of mechanisms of change in other treatments for BPD such as TFP and SFT. Conclusion There are several potential mechanisms of change associated with the theoretical underpinnings of BPD treatment (Lynch et al., 2006) and this review is a start in a long journey towards being able to confirm which specific mechanisms are active in treatments for such a complex, challenging disorder. Three broad categories of mechanism of change were identified which are well explained by Linehan’s (1993) DBT biosocial model of BPD: initial deficits in emotion regulation and self-control are improved via the therapeutic alliance and investment in treatment which result in increased skills use leading to favourable outcomes on mood and anxiety symptoms, and on measures of BPD symptoms including self-harm, impulsivity, substance misuse and borderline beliefs. Declaration of interest Susie Rudge and Peter Fonagy have no conflicts of interest to declare. Janet Feigenbaum is an international senior trainer in 10 S. Rudge et al. DBT with British Isles DBT, and Chairman of the Board of Accreditation for the Society for DBT UK. ORCID Peter Fonagy http://orcid.org/0000-0003-0229-0091 References Addis M, Linehan MM. (1989). Predicting suicidal behaviour: psychometric properties of the Suicidal Behaviours Questionnaire. Poster presented at: Annual Meeting of the Association for the Advancement Behaviour Therapy; Nov 2–5, 1989; Washington, DC. American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders, 5th ed. Arlington, VA: American Psychiatric Publishing. American Psychiatric Association. (2001). Practice guideline for the treatment of patients with borderline personality disorder— Introduction. Am J Psychiatry, 158, 1–52. Axelrod SR, Perepletchikova F, Holtzman K, Sinha R. (2011). Emotion regulation and substance use frequency in women with substance dependence and borderline personality disorder receiving dialectical behaviour therapy. Am J Drug Alcohol Abuse, 37, 37–42. Arntz A. (1994). Treatment of borderline personality disorder: A challenge for cognitive behavioural therapy. Behav Res Ther, 32, 419–30. Baer RA, Smith GT, Allen KB. (2004). Assessment of mindfulness by self-report: The Kentucky Inventory of Mindfulness Skills. Assessment, 11, 191–206. Baer RA, Smith GT, Lykins E, et al. (2008). Construct validity of the Five Facet Mindfulness Questionnaire in meditating and nonmeditating samples. Assessment, 15, 329–42. Barnicot K, Gonzalez R, McCabe R, Priebe S. (2015). Skills use and common treatment processes in dialectical behaviour therapy for borderline personality disorder. J Behav Ther Exp Psychiatry, 52, 147–56. Barnicot K, Katsakou C, Bhatti N, et al. (2012). Factors predicting the outcome of psychotherapy for borderline personality disorder: A systematic review. Clin Psychol Rev, 32, 400–12. Bateman AW, Fonagy P. (2004). Psychotherapy for borderline personality disorder: Mentalization-based treatment. Oxford: Oxford University Press. Bateman AW, Fonagy P. (2008). 8-Year follow-up of patients treated for borderline personality disorder: Mentalization-based treatment versus treatment as usual. Am J Psychiatry, 165, 631–8. Beck AT, Freeman A, Davis DD. (2004). Cognitive therapy of personality disorders, 2nd ed. New York: Guilford. Bedics JD, Atkins DC, Comtois KA, Linehan MM. (2012). Treatment differences in the therapeutic relationship and introject during a 2-year randomised controlled trial of dialectical behaviour therapy versus non-behavioural psychotherapy by experts for borderline personality disorder. J Consult Clin Psychol, 80, 66–77. Bedics JD, Atkins DC, Harned MS, Linehan MM. (2015). The therapeutic alliance as a predictor of outcome in dialectical behaviour therapy versus non-behavioural psychotherapy by experts for borderline personality disorder. Psychotherapy (Chic), 52, 67–77. Bender DS, Dolan RT, Skodol AE, et al. (2001). Treatment utilization by patients with personality disorders. Am J Psychiatry, 158, 295–302. Benjamin LS. (1974). Structural analysis of social behaviour. Psychol Rev, 81, 392–425. Bloom JM, Woodward EN, Sasmaras T, Pantalone DW. (2012). Use of dialectical behaviour therapy in inpatient treatment of borderline personality disorder: A systematic review. Psychiatr Serv, 63, 881–8. Brown GK, Newman CF, Charlesworth SE, et al. (2004). An open clinical trial of cognitive therapy for borderline personality disorder. J Personal Disord, 18, 257–71. Brown M, Comtois KA, Linehan MM. (2002). Reasons for suicide attempts and non-suicidal self-injury in women with borderline personality disorder. J Abnorm Psychol, 111, 198–202. Clarkin JF, Levy KN. (2006). Psychotherapy for patients with borderline personality disorder: Focusing on the mechanisms of change. J Clin Psychol, 62, 405–10. J Ment Health, Early Online: 1–11 Clarkin JF, Yeomans FE, Kernberg O. (2006). Psychotherapy for borderline personality, focusing on object relations. New York: American Psychiatric Publishing. Coid J, Yang M, Tyrer P, et al. (2006). Prevalence and correlates of personality disorder in Great Britain. Br J Psychiatry, 188, 423–31. Comtois KA, Russo J, Snowden M, et al. (2003). Factors associated with high use of public mental health services by persons with borderline personality disorder. Psychiatr Serv, 54, 1149–54. Davidson K, Norrie J, Tyrer P, et al. (2006). The effectiveness of cognitive behaviour therapy for borderline personality disorder: Results from the BOSCOT Trial. J Personal Disord, 20, 450–65. de Ruiter C, Greeven PJG. (2000). Personality disorders in a Dutch forensic psychiatric sample: Convergence of interview and self-report measures. J Personal Disord, 14, 162–70. Downs SH, Black N. (1998). The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. J Epidemiol Community Health, 52, 377–84. Elkin I, Shea MT, Watkins JT, et al. (1989). National Institute of Mental Health Treatment of Depression Collaborative Research Program: General effectiveness of treatments. Arch Gen Psychiatry, 46, 971–82. Elliott R. (2010). Psychotherapy change process research: Realizing the promise. Psychother Res, 20, 123–235. Feigenbaum J. (2007). Dialectical behaviour therapy: An increasing evidence base. J Ment Health, 16, 51–68. Feigenbaum JD, Fonagy P, Pilling S, et al. (2011). A real-world study of the effectiveness of DBT in the UK National Health Service. Br J Clin Psychol, 51, 121–41. Gaston L. (1991). Reliability and criterion-related validity of the California Psychotherapy Alliance Scales – Patient Version. J Consult Clin Psychol, 3, 68–74. Gibbons MBC, Crits-Christoph P, Barber JP, et al. (2009). Unique and common mechanisms of change across cognitive and dynamic psychotherapies. J Consult Clin Psychol, 77, 801–13. Gratz KL, Roemer L. (2004). Multidimensional assessment of emotion regulation and dysregulation: Development, factor structure, and initial validation of the difficulties in emotion regulation scale. J Psychopathol Behav Assessment, 26, 41–54. Horvath AO, Luborsky L. (1993). The role of the therapeutic alliance in psychotherapy. J Consult Clin Psychol, 61, 561–73. Kazdin AE. (2007). Mediators and mechanisms of change in psychotherapy research. Annu Rev Clin Psychol, 3, 1–27. Kellogg SH, Young JE. (2006). Schema therapy for borderline personality disorder. J Clin Psychol, 62, 445–58. Kramer U, Pascual-Leone A, Berthoud LD, et al. (2016). Assertive anger mediates effects of dialectical behaviour-informed skills training for borderline personality disorder: A randomized controlled trial. Clin Psychol Psychother, 23, 189–202. Kliem S, Kröger C, Kosfelder J. (2010). Dialectical behaviour therapy for borderline personality disorder: A meta-analysis using mixedeffects modelling. J Consult Clin Psychol, 78, 936–51. Leichsenring F, Leibing E. (2003). The effectiveness of psychodynamic therapy and cognitive behavior therapy in the treatment of personality disorders: A meta-analysis. Am J Psychiatry, 160, 1223–32. Lindenboim N, Comtois KA, Linehan MM. (2007). Skills practice in dialectical behaviour therapy for suicidal women meeting criteria for borderline personality disorder. Cogn Behav Pract, 14, 147–56. Linehan MM. (1993). Cognitive-behavioural treatment of borderline personality disorder. New York: The Guilford Press. Linehan MM. (2000). The empirical basis of dialectical behaviour therapy: Development of new treatments versus evaluation of existing treatments. Clin Psychol Sci Pract, 7, 113–19. Linehan MM, Comtois KA, Murray AM, et al. (2006a). Two-year randomized controlled trial and follow-up of dialectical behaviour therapy verses therapy by experts for suicidal behaviours and borderline personality disorder. Arch Gen Psychiatry, 63, 757–66. Linehan MM, Comtois KA, Brown MZ, et al. (2006b). Suicide Attempt Self-Injury Interview (SASII): Development, reliability, and validity of a scale to assess suicide attempts and intentional self-injury. Psychol Assessment, 18, 303–12. Linehan MM, Korslund KE, Harned MS, et al. (2015). Dialectical behaviour therapy for high suicide risk in individuals with borderline personality disorder: A randomised clinical trial and component analysis. JAMA Psychiatry, 72, 475–82. DOI: 10.1080/09638237.2017.1322185 Luborsky L. (1984). Principles of psychoanalytic psychotherapy. New York: Basic Books. Lynch TR, Chapman AL, Rosenthal MZ, et al. (2006). Mechanisms of change in dialectical behaviour therapy: Theoretical and empirical observations. J Clin Psychol, 62, 459–80. Martin DJ, Garske JP, Davis MK. (2000). Relation of the therapeutic alliance with outcome and other variables: A meta-analytic review. J Consult Clin Psychol, 68, 438–50. McMain S, Links P, Guimond T, et al. (2013). An exploratory study of the relationship between changes in emotion and cognitive processes and treatment outcome in borderline personality disorder. Psychother Res, 23, 658–73. Morey L. (1991). Personality assessment inventory: Professional manual. Odessa, FL: Psychological Assessment Resources. Morris C, Smith I, Alwin N. (2014). Is contact with adult mental health services helpful for individuals with diagnosable BPD? A study of service users’ views in the UK. J Ment Health, 23, 251–5. National Institute for Health and Clinical Excellence. (2009). Borderline personality disorder: Treatment and management. CG78. London: National Institute for Health and Clinical Excellence. Neacsiu AD, Rizvi SL, Linehan MM. (2010). Dialectical behaviour therapy skills use as a mediator and outcome of treatment for borderline personality disorder. Behav Res Ther, 48, 832–9. O’Toole SK, Diddy E, Kent M. (2012). Mindfulness and emotional wellbeing in women with borderline personality disorder. Mindfulness, 3, 117–23. Panos PT, Jackson JW, Hasan O, Panos A. (2014). Meta-analysis and systematic review assessing the efficacy of dialectical behaviour therapy (DBT). Res Soc Work Pract, 24, 213–23. Mechanisms of Change in DBT and CBT for BPD 11 Perroud N, Nicastro R, Jermann F, Huguelet P. (2012). Mindfulness skills in borderline personality disorder patients during dialectical behaviour therapy: Preliminary results. Int J Psychiatry Clin Pract, 16, 189–96. Perry JC, Fowler JC, Semeniuk TT. (2005). An investigation of tasks and techniques associated with dynamic interview adequacy. J Nerv Ment Dis, 193, 136–9. Robins J, Rotnitzky A. (2005). Estimation of treatment effects in randomised trials with non-compliance and dichotomous outcome using structural mean models. Biometrika, 91, 763–83. Stepp S, Epler A, Jahng S, Trull T. (2008). The effect of dialectical behaviour therapy skills use on borderline personality disorder features. J Personal Disord, 22, 549–63. Sulzer SH, Muenchow E, Potvin A, et al. (2015). Improving patientcentred communication of the borderline personality disorder diagnosis. J Ment Health, 25, 5–9. Turner RM. (2000). Naturalistic evaluation of dialectical behaviour therapy-oriented treatment for borderline personality disorder. Cogn Behav Pract, 7, 413–9. Wampold BE. (2001). The great psychotherapy debate: Model, methods, and findings. Mahwah, NJ: Lawrence Erlbaum Associates. Wenzel A, Chapman JE, Newman CF, et al. (2006). Hypothesised mechanisms of change in cognitive therapy for borderline personality disorder. J Clin Psychol, 62, 503–16. Yen S, Johnson J, Costello E, Simpson EB. (2009). A 5-day dialectical behaviour therapy partial hospital program for women with borderline personality disorder: Predictors of outcome from a 3-month follow-up study. J Psychiatr Pract, 15, 173–82.